MDEV-27058 fixup: Bogus assertion !block->page.is_io_fixed()
buf_page_get_gen(): After recv_sys_t::recover_low() returned,
the page must not be read-fixed, but it may be write-fixed,
because the io-fix state is protected by block->page.lock,
which we are not holding yet.
Also, let us copy the block descriptor state to a local variable
for examination, so that in case an assertion would fail again,
we will have the sampled state in the core dump. In a core dump of
the assertion failure, we had block->page.fix() == buf_page_t::UNFIXED,
that is, the assertion expression was holding again.
4 years ago  MDEV-13542: Crashing on corrupted page is unhelpful
The approach to handling corruption that was chosen by Oracle in
commit 177d8b0c125b841c0650d27d735e3b87509dc286
is not really useful. Not only did it actually fail to prevent InnoDB
from crashing, but it is making things worse by blocking attempts to
rescue data from or rebuild a partially readable table.
We will try to prevent crashes in a different way: by propagating
errors up the call stack. We will never mark the clustered index
persistently corrupted, so that data recovery may be attempted by
reading from the table, or by rebuilding the table.
This should also fix MDEV-13680 (crash on btr_page_alloc() failure);
it was extensively tested with innodb_file_per_table=0 and a
non-autoextend system tablespace.
We should now avoid crashes in many cases, such as when a page
cannot be read or allocated, or an inconsistency is detected when
attempting to update multiple pages. We will not crash on double-free,
such as on the recovery of DDL in system tablespace in case something
was corrupted.
Crashes on corrupted data are still possible. The fault injection mechanism
that is introduced in the subsequent commit may help catch more of them.
buf_page_import_corrupt_failure: Remove the fault injection, and instead
corrupt some pages using Perl code in the tests.
btr_cur_pessimistic_insert(): Always reserve extents (except for the
change buffer), in order to prevent a subsequent allocation failure.
btr_pcur_open_at_rnd_pos(): Merged to the only caller ibuf_merge_pages().
btr_assert_not_corrupted(), btr_corruption_report(): Remove.
Similar checks are already part of btr_block_get().
FSEG_MAGIC_N_BYTES: Replaces FSEG_MAGIC_N_VALUE.
dict_hdr_get(), trx_rsegf_get_new(), trx_undo_page_get(),
trx_undo_page_get_s_latched(): Replaced with error-checking calls.
trx_rseg_t::get(mtr_t*): Replaces trx_rsegf_get().
trx_rseg_header_create(): Let the caller update the TRX_SYS page if needed.
trx_sys_create_sys_pages(): Merged with trx_sysf_create().
dict_check_tablespaces_and_store_max_id(): Do not access
DICT_HDR_MAX_SPACE_ID, because it was already recovered in dict_boot().
Merge dict_check_sys_tables() with this function.
dir_pathname(): Replaces os_file_make_new_pathname().
row_undo_ins_remove_sec(): Do not modify the undo page by adding
a terminating NUL byte to the record.
btr_decryption_failed(): Report decryption failures
dict_set_corrupted_by_space(), dict_set_encrypted_by_space(),
dict_set_corrupted_index_cache_only(): Remove.
dict_set_corrupted(): Remove the constant parameter dict_locked=false.
Never flag the clustered index corrupted in SYS_INDEXES, because
that would deny further access to the table. It might be possible to
repair the table by executing ALTER TABLE or OPTIMIZE TABLE, in case
no B-tree leaf page is corrupted.
dict_table_skip_corrupt_index(), dict_table_next_uncorrupted_index(),
row_purge_skip_uncommitted_virtual_index(): Remove, and refactor
the callers to read dict_index_t::type only once.
dict_table_is_corrupted(): Remove.
dict_index_t::is_btree(): Determine if the index is a valid B-tree.
BUF_GET_NO_LATCH, BUF_EVICT_IF_IN_POOL: Remove.
UNIV_BTR_DEBUG: Remove. Any inconsistency will no longer trigger
assertion failures, but error codes being returned.
buf_corrupt_page_release(): Replaced with a direct call to
buf_pool.corrupted_evict().
fil_invalid_page_access_msg(): Never crash on an invalid read;
let the caller of buf_page_get_gen() decide.
btr_pcur_t::restore_position(): Propagate failure status to the caller
by returning CORRUPTED.
opt_search_plan_for_table(): Simplify the code.
row_purge_del_mark(), row_purge_upd_exist_or_extern_func(),
row_undo_ins_remove_sec_rec(), row_undo_mod_upd_del_sec(),
row_undo_mod_del_mark_sec(): Avoid mem_heap_create()/mem_heap_free()
when no secondary indexes exist.
row_undo_mod_upd_exist_sec(): Simplify the code.
row_upd_clust_step(), dict_load_table_one(): Return DB_TABLE_CORRUPT
if the clustered index (and therefore the table) is corrupted, similar
to what we do in row_insert_for_mysql().
fut_get_ptr(): Replace with buf_page_get_gen() calls.
buf_page_get_gen(): Return nullptr and *err=DB_CORRUPTION
if the page is marked as freed. For other modes than
BUF_GET_POSSIBLY_FREED or BUF_PEEK_IF_IN_POOL this will
trigger a debug assertion failure. For BUF_GET_POSSIBLY_FREED,
we will return nullptr for freed pages, so that the callers
can be simplified. The purge of transaction history will be
a new user of BUF_GET_POSSIBLY_FREED, to avoid crashes on
corrupted data.
buf_page_get_low(): Never crash on a corrupted page, but simply
return nullptr.
fseg_page_is_allocated(): Replaces fseg_page_is_free().
fts_drop_common_tables(): Return an error if the transaction
was rolled back.
fil_space_t::set_corrupted(): Report a tablespace as corrupted if
it was not reported already.
fil_space_t::io(): Invoke fil_space_t::set_corrupted() to report
out-of-bounds page access or other errors.
Clean up mtr_t::page_lock()
buf_page_get_low(): Validate the page identifier (to check for
recently read corrupted pages) after acquiring the page latch.
buf_page_t::read_complete(): Flag uninitialized (all-zero) pages
with DB_FAIL. Return DB_PAGE_CORRUPTED on page number mismatch.
mtr_t::defer_drop_ahi(): Renamed from mtr_defer_drop_ahi().
recv_sys_t::free_corrupted_page(): Only set_corrupt_fs()
if any log records exist for the page. We do not mind if read-ahead
produces corrupted (or all-zero) pages that were not actually needed
during recovery.
recv_recover_page(): Return whether the operation succeeded.
recv_sys_t::recover_low(): Simplify the logic. Check for recovery error.
Thanks to Matthias Leich for testing this extensively and to the
authors of https://rr-project.org for making it easy to diagnose
and fix any failures that were found during the testing.
3 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago  TSAN: unprotected global counter
WARNING: ThreadSanitizer: data race (pid=1506937)
Write of size 8 at 0x0000067ab740 by thread T6:
#0 buf_page_get_low(page_id_t, unsigned long, unsigned long, buf_block_t*, unsigned long, mtr_t*, dberr_t*, bool) /storage/innobase/buf/buf0buf.cc:2946:8 (mariadbd+0x2014c7f)
#1 buf_page_get_gen(page_id_t, unsigned long, unsigned long, buf_block_t*, unsigned long, mtr_t*, dberr_t*, bool) /storage/innobase/buf/buf0buf.cc:3047:10 (mariadbd+0x2016216)
#2 btr_cur_search_to_nth_level_func(dict_index_t*, unsigned long, dtuple_t const*, page_cur_mode_t, unsigned long, btr_cur_t*, ssux_lock_impl<true>*, mtr_t*, unsigned long) /storage/innobase/btr/btr0cur.cc:1613:10 (mariadbd+0x1fb5bff)
#3 btr_pcur_open_low(dict_index_t*, unsigned long, dtuple_t const*, page_cur_mode_t, unsigned long, btr_pcur_t*, unsigned long, mtr_t*) /storage/innobase/include/btr0pcur.ic:439:8 (mariadbd+0x24ddead)
#4 row_search_on_row_ref(btr_pcur_t*, unsigned long, dict_table_t const*, dtuple_t const*, mtr_t*) /storage/innobase/row/row0row.cc:1215:7 (mariadbd+0x24dd537)
#5 row_purge_reposition_pcur(unsigned long, purge_node_t*, mtr_t*) /storage/innobase/row/row0purge.cc:81:23 (mariadbd+0x24c5369)
#6 row_purge_reset_trx_id(purge_node_t*, mtr_t*) /storage/innobase/row/row0purge.cc:748:6 (mariadbd+0x24c90c7)
#7 row_purge_record_func(purge_node_t*, unsigned char*, que_thr_t const*, bool) /storage/innobase/row/row0purge.cc:1174:4 (mariadbd+0x24c8262)
#8 row_purge(purge_node_t*, unsigned char*, que_thr_t*) /storage/innobase/row/row0purge.cc:1218:18 (mariadbd+0x24c5af3)
#9 row_purge_step(que_thr_t*) /storage/innobase/row/row0purge.cc:1267:3 (mariadbd+0x24c5996)
#10 que_thr_step(que_thr_t*) /storage/innobase/que/que0que.cc:653:9 (mariadbd+0x23d5298)
#11 que_run_threads_low(que_thr_t*) /storage/innobase/que/que0que.cc:709:25 (mariadbd+0x23d3f29)
#12 que_run_threads(que_thr_t*) /storage/innobase/que/que0que.cc:729:2 (mariadbd+0x23d3bdf)
#13 srv_task_execute() /storage/innobase/srv/srv0srv.cc:1692:3 (mariadbd+0x2562841)
#14 purge_worker_callback(void*) /storage/innobase/srv/srv0srv.cc:1864:10 (mariadbd+0x255f361)
#15 tpool::task_group::execute(tpool::task*) /tpool/task_group.cc:55:9 (mariadbd+0x260a5ca)
#16 tpool::task::execute() /tpool/task.cc:47:16 (mariadbd+0x260adf6)
#17 tpool::thread_pool_generic::worker_main(tpool::worker_data*) /tpool/tpool_generic.cc:550:11 (mariadbd+0x25fc590)
#18 void std::__invoke_impl<void, void (tpool::thread_pool_generic::*)(tpool::worker_data*), tpool::thread_pool_generic*, tpool::worker_data*>(std::__invoke_memfun_deref, void (tpool::thread_pool_generic::*&&)(tpool::worker_data*), tpool::thread_pool_generic*&&, tpool::worker_data*&&) /usr/lib/gcc/x86_64-linux-gnu/11/../../../../include/c++/11/bits/invoke.h:74:14 (mariadbd+0x26061b5)
#19 std::__invoke_result<void (tpool::thread_pool_generic::*)(tpool::worker_data*), tpool::thread_pool_generic*, tpool::worker_data*>::type std::__invoke<void (tpool::thread_pool_generic::*)(tpool::worker_data*), tpool::thread_pool_generic*, tpool::worker_data*>(void (tpool::thread_pool_generic::*&&)(tpool::worker_data*), tpool::thread_pool_generic*&&, tpool::worker_data*&&) /usr/lib/gcc/x86_64-linux-gnu/11/../../../../include/c++/11/bits/invoke.h:96:14 (mariadbd+0x2605f57)
#20 void std::thread::_Invoker<std::tuple<void (tpool::thread_pool_generic::*)(tpool::worker_data*), tpool::thread_pool_generic*, tpool::worker_data*> >::_M_invoke<0ul, 1ul, 2ul>(std::_Index_tuple<0ul, 1ul, 2ul>) /usr/lib/gcc/x86_64-linux-gnu/11/../../../../include/c++/11/bits/std_thread.h:253:13 (mariadbd+0x2605ecb)
#21 std::thread::_Invoker<std::tuple<void (tpool::thread_pool_generic::*)(tpool::worker_data*), tpool::thread_pool_generic*, tpool::worker_data*> >::operator()() /usr/lib/gcc/x86_64-linux-gnu/11/../../../../include/c++/11/bits/std_thread.h:260:11 (mariadbd+0x2605e35)
#22 std::thread::_State_impl<std::thread::_Invoker<std::tuple<void (tpool::thread_pool_generic::*)(tpool::worker_data*), tpool::thread_pool_generic*, tpool::worker_data*> > >::_M_run() /usr/lib/gcc/x86_64-linux-gnu/11/../../../../include/c++/11/bits/std_thread.h:211:13 (mariadbd+0x2605ac9)
#23 <null> <null> (libstdc++.so.6+0xd230f)
Previous write of size 8 at 0x0000067ab740 by thread T8:
#0 buf_page_get_low(page_id_t, unsigned long, unsigned long, buf_block_t*, unsigned long, mtr_t*, dberr_t*, bool) /storage/innobase/buf/buf0buf.cc:2946:8 (mariadbd+0x2014c7f)
#1 buf_page_get_gen(page_id_t, unsigned long, unsigned long, buf_block_t*, unsigned long, mtr_t*, dberr_t*, bool) /storage/innobase/buf/buf0buf.cc:3047:10 (mariadbd+0x2016216)
#2 btr_cur_search_to_nth_level_func(dict_index_t*, unsigned long, dtuple_t const*, page_cur_mode_t, unsigned long, btr_cur_t*, ssux_lock_impl<true>*, mtr_t*, unsigned long) /storage/innobase/btr/btr0cur.cc:1613:10 (mariadbd+0x1fb5bff)
#3 btr_pcur_open_low(dict_index_t*, unsigned long, dtuple_t const*, page_cur_mode_t, unsigned long, btr_pcur_t*, unsigned long, mtr_t*) /storage/innobase/include/btr0pcur.ic:439:8 (mariadbd+0x24ddead)
#4 row_search_on_row_ref(btr_pcur_t*, unsigned long, dict_table_t const*, dtuple_t const*, mtr_t*) /storage/innobase/row/row0row.cc:1215:7 (mariadbd+0x24dd537)
#5 row_purge_reposition_pcur(unsigned long, purge_node_t*, mtr_t*) /storage/innobase/row/row0purge.cc:81:23 (mariadbd+0x24c5369)
#6 row_purge_reset_trx_id(purge_node_t*, mtr_t*) /storage/innobase/row/row0purge.cc:748:6 (mariadbd+0x24c90c7)
#7 row_purge_record_func(purge_node_t*, unsigned char*, que_thr_t const*, bool) /storage/innobase/row/row0purge.cc:1174:4 (mariadbd+0x24c8262)
#8 row_purge(purge_node_t*, unsigned char*, que_thr_t*) /storage/innobase/row/row0purge.cc:1218:18 (mariadbd+0x24c5af3)
#9 row_purge_step(que_thr_t*) /storage/innobase/row/row0purge.cc:1267:3 (mariadbd+0x24c5996)
#10 que_thr_step(que_thr_t*) /storage/innobase/que/que0que.cc:653:9 (mariadbd+0x23d5298)
#11 que_run_threads_low(que_thr_t*) /storage/innobase/que/que0que.cc:709:25 (mariadbd+0x23d3f29)
#12 que_run_threads(que_thr_t*) /storage/innobase/que/que0que.cc:729:2 (mariadbd+0x23d3bdf)
#13 trx_purge(unsigned long, bool) /storage/innobase/trx/trx0purge.cc:1271:2 (mariadbd+0x25841b4)
#14 srv_do_purge(unsigned long*) /storage/innobase/srv/srv0srv.cc:1784:20 (mariadbd+0x2563224)
#15 purge_coordinator_callback_low() /storage/innobase/srv/srv0srv.cc:1881:35 (mariadbd+0x2562b3b)
#16 purge_coordinator_callback(void*) /storage/innobase/srv/srv0srv.cc:1910:3 (mariadbd+0x255f4ab)
#17 tpool::task_group::execute(tpool::task*) /tpool/task_group.cc:55:9 (mariadbd+0x260a5ca)
#18 tpool::task::execute() /tpool/task.cc:47:16 (mariadbd+0x260adf6)
#19 tpool::thread_pool_generic::worker_main(tpool::worker_data*) /tpool/tpool_generic.cc:550:11 (mariadbd+0x25fc590)
#20 void std::__invoke_impl<void, void (tpool::thread_pool_generic::*)(tpool::worker_data*), tpool::thread_pool_generic*, tpool::worker_data*>(std::__invoke_memfun_deref, void (tpool::thread_pool_generic::*&&)(tpool::worker_data*), tpool::thread_pool_generic*&&, tpool::worker_data*&&) /usr/lib/gcc/x86_64-linux-gnu/11/../../../../include/c++/11/bits/invoke.h:74:14 (mariadbd+0x26061b5)
#21 std::__invoke_result<void (tpool::thread_pool_generic::*)(tpool::worker_data*), tpool::thread_pool_generic*, tpool::worker_data*>::type std::__invoke<void (tpool::thread_pool_generic::*)(tpool::worker_data*), tpool::thread_pool_generic*, tpool::worker_data*>(void (tpool::thread_pool_generic::*&&)(tpool::worker_data*), tpool::thread_pool_generic*&&, tpool::worker_data*&&) /usr/lib/gcc/x86_64-linux-gnu/11/../../../../include/c++/11/bits/invoke.h:96:14 (mariadbd+0x2605f57)
#22 void std::thread::_Invoker<std::tuple<void (tpool::thread_pool_generic::*)(tpool::worker_data*), tpool::thread_pool_generic*, tpool::worker_data*> >::_M_invoke<0ul, 1ul, 2ul>(std::_Index_tuple<0ul, 1ul, 2ul>) /usr/lib/gcc/x86_64-linux-gnu/11/../../../../include/c++/11/bits/std_thread.h:253:13 (mariadbd+0x2605ecb)
#23 std::thread::_Invoker<std::tuple<void (tpool::thread_pool_generic::*)(tpool::worker_data*), tpool::thread_pool_generic*, tpool::worker_data*> >::operator()() /usr/lib/gcc/x86_64-linux-gnu/11/../../../../include/c++/11/bits/std_thread.h:260:11 (mariadbd+0x2605e35)
#24 std::thread::_State_impl<std::thread::_Invoker<std::tuple<void (tpool::thread_pool_generic::*)(tpool::worker_data*), tpool::thread_pool_generic*, tpool::worker_data*> > >::_M_run() /usr/lib/gcc/x86_64-linux-gnu/11/../../../../include/c++/11/bits/std_thread.h:211:13 (mariadbd+0x2605ac9)
#25 <null> <null> (libstdc++.so.6+0xd230f)
Location is global 'buf_dbg_counter' of size 8 at 0x0000067ab740 (mariadbd+0x67ab740)
The obvious fix is to make counter atomic.
4 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-14425 Improve the redo log for concurrency
The InnoDB redo log used to be formatted in blocks of 512 bytes.
The log blocks were encrypted and the checksum was calculated while
holding log_sys.mutex, creating a serious scalability bottleneck.
We remove the fixed-size redo log block structure altogether and
essentially turn every mini-transaction into a log block of its own.
This allows encryption and checksum calculations to be performed
on local mtr_t::m_log buffers, before acquiring log_sys.mutex.
The mutex only protects a memcpy() of the data to the shared
log_sys.buf, as well as the padding of the log, in case the
to-be-written part of the log would not end in a block boundary of
the underlying storage. For now, the "padding" consists of writing
a single NUL byte, to allow recovery and mariadb-backup to detect
the end of the circular log faster.
Like the previous implementation, we will overwrite the last log block
over and over again, until it has been completely filled. It would be
possible to write only up to the last completed block (if no more
recent write was requested), or to write dummy FILE_CHECKPOINT records
to fill the incomplete block, by invoking the currently disabled
function log_pad(). This would require adjustments to some logic around
log checkpoints, page flushing, and shutdown.
An upgrade after a crash of any previous version is not supported.
Logically empty log files from a previous version will be upgraded.
An attempt to start up InnoDB without a valid ib_logfile0 will be
refused. Previously, the redo log used to be created automatically
if it was missing. Only with with innodb_force_recovery=6, it is
possible to start InnoDB in read-only mode even if the log file
does not exist. This allows the contents of a possibly corrupted
database to be dumped.
Because a prepared backup from an earlier version of mariadb-backup
will create a 0-sized log file, we will allow an upgrade from such
log files, provided that the FIL_PAGE_FILE_FLUSH_LSN in the system
tablespace looks valid.
The 512-byte log checkpoint blocks at 0x200 and 0x600 will be replaced
with 64-byte log checkpoint blocks at 0x1000 and 0x2000.
The start of log records will move from 0x800 to 0x3000. This allows us
to use 4096-byte aligned blocks for all I/O in a future revision.
We extend the MDEV-12353 redo log record format as follows.
(1) Empty mini-transactions or extra NUL bytes will not be allowed.
(2) The end-of-minitransaction marker (a NUL byte) will be replaced
with a 1-bit sequence number, which will be toggled each time when the
circular log file wraps back to the beginning.
(3) After the sequence bit, a CRC-32C checksum of all data
(excluding the sequence bit) will written.
(4) If the log is encrypted, 8 bytes will be written before
the checksum and included in it. This is part of the
initialization vector (IV) of encrypted log data.
(5) File names, page numbers, and checkpoint information will not be
encrypted. Only the payload bytes of page-level log will be encrypted.
The tablespace ID and page number will form part of the IV.
(6) For padding, arbitrary-length FILE_CHECKPOINT records may be written,
with all-zero payload, and with the normal end marker and checksum.
The minimum size is 7 bytes, or 7+8 with innodb_encrypt_log=ON.
In mariadb-backup and in Galera snapshot transfer (SST) scripts, we will
no longer remove ib_logfile0 or create an empty ib_logfile0. Server startup
will require a valid log file. When resizing the log, we will create
a logically empty ib_logfile101 at the current LSN and use an atomic rename
to replace ib_logfile0 with it. See the test innodb.log_file_size.
Because there is no mandatory padding in the log file, we are able
to create a dummy log file as of an arbitrary log sequence number.
See the test mariabackup.huge_lsn.
The parameter innodb_log_write_ahead_size and the
INFORMATION_SCHEMA.INNODB_METRICS counter log_padded will be removed.
The minimum value of innodb_log_buffer_size will be increased to 2MiB
(because log_sys.buf will replace recv_sys.buf) and the increment
adjusted to 4096 bytes (the maximum log block size).
The following INFORMATION_SCHEMA.INNODB_METRICS counters will be removed:
os_log_fsyncs
os_log_pending_fsyncs
log_pending_log_flushes
log_pending_checkpoint_writes
The following status variables will be removed:
Innodb_os_log_fsyncs (this is included in Innodb_data_fsyncs)
Innodb_os_log_pending_fsyncs (this was limited to at most 1 by design)
log_sys.get_block_size(): Return the physical block size of the log file.
This is only implemented on Linux and Microsoft Windows for now, and for
the power-of-2 block sizes between 64 and 4096 bytes (the minimum and
maximum size of a checkpoint block). If the block size is anything else,
the traditional 512-byte size will be used via normal file system
buffering.
If the file system buffers can be bypassed, a message like the following
will be issued:
InnoDB: File system buffers for log disabled (block size=512 bytes)
InnoDB: File system buffers for log disabled (block size=4096 bytes)
This has been tested on Linux and Microsoft Windows with both sizes.
On Linux, only enable O_DIRECT on the log for innodb_flush_method=O_DSYNC.
Tests in 3 different environments where the log is stored in a device
with a physical block size of 512 bytes are yielding better throughput
without O_DIRECT. This could be due to the fact that in the event the
last log block is being overwritten (if multiple transactions would
become durable at the same time, and each of will write a small
number of bytes to the last log block), it should be faster to re-copy
data from log_sys.buf or log_sys.flush_buf to the kernel buffer,
to be finally written at fdatasync() time.
The parameter innodb_flush_method=O_DSYNC will imply O_DIRECT for
data files. This option will enable O_DIRECT on the log file on Linux.
It may be unsafe to use when the storage device does not support
FUA (Force Unit Access) mode.
When the server is compiled WITH_PMEM=ON, we will use memory-mapped
I/O for the log file if the log resides on a "mount -o dax" device.
We will identify PMEM in a start-up message:
InnoDB: log sequence number 0 (memory-mapped); transaction id 3
On Linux, we will also invoke mmap() on any ib_logfile0 that resides
in /dev/shm, effectively treating the log file as persistent memory.
This should speed up "./mtr --mem" and increase the test coverage of
PMEM on non-PMEM hardware. It also allows users to estimate how much
the performance would be improved by installing persistent memory.
On other tmpfs file systems such as /run, we will not use mmap().
mariadb-backup: Eliminated several variables. We will refer
directly to recv_sys and log_sys.
backup_wait_for_lsn(): Detect non-progress of
xtrabackup_copy_logfile(). In this new log format with
arbitrary-sized blocks, we can only detect log file overrun
indirectly, by observing that the scanned log sequence number
is not advancing.
xtrabackup_copy_logfile(): On PMEM, do not modify the sequence bit,
because we are not allowed to modify the server's log file, and our
memory mapping is read-only.
trx_flush_log_if_needed_low(): Do not use the callback on pmem.
Using neither flush_lock nor write_lock around PMEM writes seems
to yield the best performance. The pmem_persist() calls may
still be somewhat slower than the pwrite() and fdatasync() based
interface (PMEM mounted without -o dax).
recv_sys_t::buf: Remove. We will use log_sys.buf for parsing.
recv_sys_t::MTR_SIZE_MAX: Replaces RECV_SCAN_SIZE.
recv_sys_t::file_checkpoint: Renamed from mlog_checkpoint_lsn.
recv_sys_t, log_sys_t: Removed many data members.
recv_sys.lsn: Renamed from recv_sys.recovered_lsn.
recv_sys.offset: Renamed from recv_sys.recovered_offset.
log_sys.buf_size: Replaces srv_log_buffer_size.
recv_buf: A smart pointer that wraps log_sys.buf[recv_sys.offset]
when the buffer is being allocated from the memory heap.
recv_ring: A smart pointer that wraps a circular log_sys.buf[] that is
backed by ib_logfile0. The pointer will wrap from recv_sys.len
(log_sys.file_size) to log_sys.START_OFFSET. For the record that
wraps around, we may copy file name or record payload data to
the auxiliary buffer decrypt_buf in order to have a contiguous
block of memory. The maximum size of a record is less than
innodb_page_size bytes.
recv_sys_t::parse(): Take the smart pointer as a template parameter.
Do not temporarily add a trailing NUL byte to FILE_ records, because
we are not supposed to modify the memory-mapped log file. (It is
attached in read-write mode already during recovery.)
recv_sys_t::parse_mtr(): Wrapper for recv_sys_t::parse().
recv_sys_t::parse_pmem(): Like parse_mtr(), but if PREMATURE_EOF would be
returned on PMEM, use recv_ring to wrap around the buffer to the start.
mtr_t::finish_write(), log_close(): Do not enforce log_sys.max_buf_free
on PMEM, because it has no meaning on the mmap-based log.
log_sys.write_to_buf: Count writes to log_sys.buf. Replaces
srv_stats.log_write_requests and export_vars.innodb_log_write_requests.
Protected by log_sys.mutex. Updated consistently in log_close().
Previously, mtr_t::commit() conditionally updated the count,
which was inconsistent.
log_sys.write_to_log: Count swaps of log_sys.buf and log_sys.flush_buf,
for writing to log_sys.log (the ib_logfile0). Replaces
srv_stats.log_writes and export_vars.innodb_log_writes.
Protected by log_sys.mutex.
log_sys.waits: Count waits in append_prepare(). Replaces
srv_stats.log_waits and export_vars.innodb_log_waits.
recv_recover_page(): Do not unnecessarily acquire
log_sys.flush_order_mutex. We are inserting the blocks in arbitary
order anyway, to be adjusted in recv_sys.apply(true).
We will change the definition of flush_lock and write_lock to
avoid potential false sharing. Depending on sizeof(log_sys) and
CPU_LEVEL1_DCACHE_LINESIZE, the flush_lock and write_lock could
share a cache line with each other or with the last data members
of log_sys.
Thanks to Matthias Leich for providing https://rr-project.org traces
for various failures during the development, and to
Thirunarayanan Balathandayuthapani for his help in debugging
some of the recovery code. And thanks to the developers of the
rr debugger for a tool without which extensive changes to InnoDB
would be very challenging to get right.
Thanks to Vladislav Vaintroub for useful feedback and
to him, Axel Schwenke and Krunal Bauskar for testing the performance.
4 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-21132 Remove buf_page_t::newest_modification
At each mini-transaction commit, the log sequence number of the
mini-transaction must be written to each modified page, so that
it will be available in the FIL_PAGE_LSN field when the page is
being read in crash recovery.
InnoDB was unnecessarily allocating redundant storage for the
field, in buf_page_t::newest_modification. Let us access
FIL_PAGE_LSN directly.
Furthermore, on ALTER TABLE...IMPORT TABLESPACE, let us write
0 to FIL_PAGE_LSN instead of using log_sys.lsn.
buf_flush_init_for_writing(), buf_flush_update_zip_checksum(),
fil_encrypt_buf_for_full_crc32(), fil_encrypt_buf(),
fil_space_encrypt(): Remove the parameter lsn.
buf_page_get_newest_modification(): Merge with the only caller.
buf_tmp_reserve_compression_buf(), buf_tmp_page_encrypt(),
buf_page_encrypt(): Define static in the same compilation unit
with the only caller.
PageConverter::m_current_lsn: Remove. Write 0 to FIL_PAGE_LSN
on ALTER TABLE...IMPORT TABLESPACE.
6 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-18644: Support full_crc32 for page_compressed
This is a follow-up task to MDEV-12026, which introduced
innodb_checksum_algorithm=full_crc32 and a simpler page format.
MDEV-12026 did not enable full_crc32 for page_compressed tables,
which we will be doing now.
This is joint work with Thirunarayanan Balathandayuthapani.
For innodb_checksum_algorithm=full_crc32 we change the
page_compressed format as follows:
FIL_PAGE_TYPE: The most significant bit will be set to indicate
page_compressed format. The least significant bits will contain
the compressed page size, rounded up to a multiple of 256 bytes.
The checksum will be stored in the last 4 bytes of the page
(whether it is the full page or a page_compressed page whose
size is determined by FIL_PAGE_TYPE), covering all preceding
bytes of the page. If encryption is used, then the page will
be encrypted between compression and computing the checksum.
For page_compressed, FIL_PAGE_LSN will not be repeated at
the end of the page.
FSP_SPACE_FLAGS (already implemented as part of MDEV-12026):
We will store the innodb_compression_algorithm that may be used
to compress pages. Previously, the choice of algorithm was written
to each compressed data page separately, and one would be unable
to know in advance which compression algorithm(s) are used.
fil_space_t::full_crc32_page_compressed_len(): Determine if the
page_compressed algorithm of the tablespace needs to know the
exact length of the compressed data. If yes, we will reserve and
write an extra byte for this right before the checksum.
buf_page_is_compressed(): Determine if a page uses page_compressed
(in any innodb_checksum_algorithm).
fil_page_decompress(): Pass also fil_space_t::flags so that the
format can be determined.
buf_page_is_zeroes(): Check if a page is full of zero bytes.
buf_page_full_crc32_is_corrupted(): Renamed from
buf_encrypted_full_crc32_page_is_corrupted(). For full_crc32,
we always simply validate the checksum to the page contents,
while the physical page size is explicitly specified by an
unencrypted part of the page header.
buf_page_full_crc32_size(): Determine the size of a full_crc32 page.
buf_dblwr_check_page_lsn(): Make this a debug-only function, because
it involves potentially costly lookups of fil_space_t.
create_table_info_t::check_table_options(),
ha_innobase::check_if_supported_inplace_alter(): Do allow the creation
of SPATIAL INDEX with full_crc32 also when page_compressed is used.
commit_cache_norebuild(): Preserve the compression algorithm when
updating the page_compression_level.
dict_tf_to_fsp_flags(): Set the flags for page compression algorithm.
FIXME: Maybe there should be a table option page_compression_algorithm
and a session variable to back it?
7 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-18644: Support full_crc32 for page_compressed
This is a follow-up task to MDEV-12026, which introduced
innodb_checksum_algorithm=full_crc32 and a simpler page format.
MDEV-12026 did not enable full_crc32 for page_compressed tables,
which we will be doing now.
This is joint work with Thirunarayanan Balathandayuthapani.
For innodb_checksum_algorithm=full_crc32 we change the
page_compressed format as follows:
FIL_PAGE_TYPE: The most significant bit will be set to indicate
page_compressed format. The least significant bits will contain
the compressed page size, rounded up to a multiple of 256 bytes.
The checksum will be stored in the last 4 bytes of the page
(whether it is the full page or a page_compressed page whose
size is determined by FIL_PAGE_TYPE), covering all preceding
bytes of the page. If encryption is used, then the page will
be encrypted between compression and computing the checksum.
For page_compressed, FIL_PAGE_LSN will not be repeated at
the end of the page.
FSP_SPACE_FLAGS (already implemented as part of MDEV-12026):
We will store the innodb_compression_algorithm that may be used
to compress pages. Previously, the choice of algorithm was written
to each compressed data page separately, and one would be unable
to know in advance which compression algorithm(s) are used.
fil_space_t::full_crc32_page_compressed_len(): Determine if the
page_compressed algorithm of the tablespace needs to know the
exact length of the compressed data. If yes, we will reserve and
write an extra byte for this right before the checksum.
buf_page_is_compressed(): Determine if a page uses page_compressed
(in any innodb_checksum_algorithm).
fil_page_decompress(): Pass also fil_space_t::flags so that the
format can be determined.
buf_page_is_zeroes(): Check if a page is full of zero bytes.
buf_page_full_crc32_is_corrupted(): Renamed from
buf_encrypted_full_crc32_page_is_corrupted(). For full_crc32,
we always simply validate the checksum to the page contents,
while the physical page size is explicitly specified by an
unencrypted part of the page header.
buf_page_full_crc32_size(): Determine the size of a full_crc32 page.
buf_dblwr_check_page_lsn(): Make this a debug-only function, because
it involves potentially costly lookups of fil_space_t.
create_table_info_t::check_table_options(),
ha_innobase::check_if_supported_inplace_alter(): Do allow the creation
of SPATIAL INDEX with full_crc32 also when page_compressed is used.
commit_cache_norebuild(): Preserve the compression algorithm when
updating the page_compression_level.
dict_tf_to_fsp_flags(): Set the flags for page compression algorithm.
FIXME: Maybe there should be a table option page_compression_algorithm
and a session variable to back it?
7 years ago MDEV-21132 Remove buf_page_t::newest_modification
At each mini-transaction commit, the log sequence number of the
mini-transaction must be written to each modified page, so that
it will be available in the FIL_PAGE_LSN field when the page is
being read in crash recovery.
InnoDB was unnecessarily allocating redundant storage for the
field, in buf_page_t::newest_modification. Let us access
FIL_PAGE_LSN directly.
Furthermore, on ALTER TABLE...IMPORT TABLESPACE, let us write
0 to FIL_PAGE_LSN instead of using log_sys.lsn.
buf_flush_init_for_writing(), buf_flush_update_zip_checksum(),
fil_encrypt_buf_for_full_crc32(), fil_encrypt_buf(),
fil_space_encrypt(): Remove the parameter lsn.
buf_page_get_newest_modification(): Merge with the only caller.
buf_tmp_reserve_compression_buf(), buf_tmp_page_encrypt(),
buf_page_encrypt(): Define static in the same compilation unit
with the only caller.
PageConverter::m_current_lsn: Remove. Write 0 to FIL_PAGE_LSN
on ALTER TABLE...IMPORT TABLESPACE.
6 years ago MDEV-18644: Support full_crc32 for page_compressed
This is a follow-up task to MDEV-12026, which introduced
innodb_checksum_algorithm=full_crc32 and a simpler page format.
MDEV-12026 did not enable full_crc32 for page_compressed tables,
which we will be doing now.
This is joint work with Thirunarayanan Balathandayuthapani.
For innodb_checksum_algorithm=full_crc32 we change the
page_compressed format as follows:
FIL_PAGE_TYPE: The most significant bit will be set to indicate
page_compressed format. The least significant bits will contain
the compressed page size, rounded up to a multiple of 256 bytes.
The checksum will be stored in the last 4 bytes of the page
(whether it is the full page or a page_compressed page whose
size is determined by FIL_PAGE_TYPE), covering all preceding
bytes of the page. If encryption is used, then the page will
be encrypted between compression and computing the checksum.
For page_compressed, FIL_PAGE_LSN will not be repeated at
the end of the page.
FSP_SPACE_FLAGS (already implemented as part of MDEV-12026):
We will store the innodb_compression_algorithm that may be used
to compress pages. Previously, the choice of algorithm was written
to each compressed data page separately, and one would be unable
to know in advance which compression algorithm(s) are used.
fil_space_t::full_crc32_page_compressed_len(): Determine if the
page_compressed algorithm of the tablespace needs to know the
exact length of the compressed data. If yes, we will reserve and
write an extra byte for this right before the checksum.
buf_page_is_compressed(): Determine if a page uses page_compressed
(in any innodb_checksum_algorithm).
fil_page_decompress(): Pass also fil_space_t::flags so that the
format can be determined.
buf_page_is_zeroes(): Check if a page is full of zero bytes.
buf_page_full_crc32_is_corrupted(): Renamed from
buf_encrypted_full_crc32_page_is_corrupted(). For full_crc32,
we always simply validate the checksum to the page contents,
while the physical page size is explicitly specified by an
unencrypted part of the page header.
buf_page_full_crc32_size(): Determine the size of a full_crc32 page.
buf_dblwr_check_page_lsn(): Make this a debug-only function, because
it involves potentially costly lookups of fil_space_t.
create_table_info_t::check_table_options(),
ha_innobase::check_if_supported_inplace_alter(): Do allow the creation
of SPATIAL INDEX with full_crc32 also when page_compressed is used.
commit_cache_norebuild(): Preserve the compression algorithm when
updating the page_compression_level.
dict_tf_to_fsp_flags(): Set the flags for page compression algorithm.
FIXME: Maybe there should be a table option page_compression_algorithm
and a session variable to back it?
7 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago  MDEV-12026: Implement innodb_checksum_algorithm=full_crc32
MariaDB data-at-rest encryption (innodb_encrypt_tables)
had repurposed the same unused data field that was repurposed
in MySQL 5.7 (and MariaDB 10.2) for the Split Sequence Number (SSN)
field of SPATIAL INDEX. Because of this, MariaDB was unable to
support encryption on SPATIAL INDEX pages.
Furthermore, InnoDB page checksums skipped some bytes, and there
are multiple variations and checksum algorithms. By default,
InnoDB accepts all variations of all algorithms that ever existed.
This unnecessarily weakens the page checksums.
We hereby introduce two more innodb_checksum_algorithm variants
(full_crc32, strict_full_crc32) that are special in a way:
When either setting is active, newly created data files will
carry a flag (fil_space_t::full_crc32()) that indicates that
all pages of the file will use a full CRC-32C checksum over the
entire page contents (excluding the bytes where the checksum
is stored, at the very end of the page). Such files will always
use that checksum, no matter what the parameter
innodb_checksum_algorithm is assigned to.
For old files, the old checksum algorithms will continue to be
used. The value strict_full_crc32 will be equivalent to strict_crc32
and the value full_crc32 will be equivalent to crc32.
ROW_FORMAT=COMPRESSED tables will only use the old format.
These tables do not support new features, such as larger
innodb_page_size or instant ADD/DROP COLUMN. They may be
deprecated in the future. We do not want an unnecessary
file format change for them.
The new full_crc32() format also cleans up the MariaDB tablespace
flags. We will reserve flags to store the page_compressed
compression algorithm, and to store the compressed payload length,
so that checksum can be computed over the compressed (and
possibly encrypted) stream and can be validated without
decrypting or decompressing the page.
In the full_crc32 format, there no longer are separate before-encryption
and after-encryption checksums for pages. The single checksum is
computed on the page contents that is written to the file.
We do not make the new algorithm the default for two reasons.
First, MariaDB 10.4.2 was a beta release, and the default values
of parameters should not change after beta. Second, we did not
yet implement the full_crc32 format for page_compressed pages.
This will be fixed in MDEV-18644.
This is joint work with Marko Mäkelä.
7 years ago MDEV-21132 Remove buf_page_t::newest_modification
At each mini-transaction commit, the log sequence number of the
mini-transaction must be written to each modified page, so that
it will be available in the FIL_PAGE_LSN field when the page is
being read in crash recovery.
InnoDB was unnecessarily allocating redundant storage for the
field, in buf_page_t::newest_modification. Let us access
FIL_PAGE_LSN directly.
Furthermore, on ALTER TABLE...IMPORT TABLESPACE, let us write
0 to FIL_PAGE_LSN instead of using log_sys.lsn.
buf_flush_init_for_writing(), buf_flush_update_zip_checksum(),
fil_encrypt_buf_for_full_crc32(), fil_encrypt_buf(),
fil_space_encrypt(): Remove the parameter lsn.
buf_page_get_newest_modification(): Merge with the only caller.
buf_tmp_reserve_compression_buf(), buf_tmp_page_encrypt(),
buf_page_encrypt(): Define static in the same compilation unit
with the only caller.
PageConverter::m_current_lsn: Remove. Write 0 to FIL_PAGE_LSN
on ALTER TABLE...IMPORT TABLESPACE.
6 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-25105 Remove innodb_checksum_algorithm values none,innodb,...
Historically, InnoDB supported a buggy page checksum algorithm that did not
compute a checksum over the full page. Later, well before MySQL 4.1
introduced .ibd files and the innodb_file_per_table option, the algorithm
was corrected and the first 4 bytes of each page were redefined to be
a checksum.
The original checksum was so slow that an option to disable page checksum
was introduced for benchmarketing purposes.
The Intel Nehalem microarchitecture introduced the SSE4.2 instruction set
extension, which includes instructions for faster computation of CRC-32C.
In MySQL 5.6 (and MariaDB 10.0), innodb_checksum_algorithm=crc32 was
implemented to make of that. As that option was changed to be the default
in MySQL 5.7, a bug was found on big-endian platforms and some work-around
code was added to weaken that checksum further. MariaDB disables that
work-around by default since MDEV-17958.
Later, SIMD-accelerated CRC-32C has been implemented in MariaDB for POWER
and ARM and also for IA-32/AMD64, making use of carry-less multiplication
where available.
Long story short, innodb_checksum_algorithm=crc32 is faster and more secure
than the pre-MySQL 5.6 checksum, called innodb_checksum_algorithm=innodb.
It should have removed any need to use innodb_checksum_algorithm=none.
The setting innodb_checksum_algorithm=crc32 is the default in
MySQL 5.7 and MariaDB Server 10.2, 10.3, 10.4. In MariaDB 10.5,
MDEV-19534 made innodb_checksum_algorithm=full_crc32 the default.
It is even faster and more secure.
The default settings in MariaDB do allow old data files to be read,
no matter if a worse checksum algorithm had been used.
(Unfortunately, before innodb_checksum_algorithm=full_crc32,
the data files did not identify which checksum algorithm is being used.)
The non-default settings innodb_checksum_algorithm=strict_crc32 or
innodb_checksum_algorithm=strict_full_crc32 would only allow CRC-32C
checksums. The incompatibility with old data files is why they are
not the default.
The newest server not to support innodb_checksum_algorithm=crc32
were MySQL 5.5 and MariaDB 5.5. Both have reached their end of life.
A valid reason for using innodb_checksum_algorithm=innodb could have
been the ability to downgrade. If it is really needed, data files
can be converted with an older version of the innochecksum utility.
Because there is no good reason to allow data files to be written
with insecure checksums, we will reject those option values:
innodb_checksum_algorithm=none
innodb_checksum_algorithm=innodb
innodb_checksum_algorithm=strict_none
innodb_checksum_algorithm=strict_innodb
Furthermore, the following innochecksum options will be removed,
because only strict crc32 will be supported:
innochecksum --strict-check=crc32
innochecksum -C crc32
innochecksum --write=crc32
innochecksum -w crc32
If a user wishes to convert a data file to use a different checksum
(so that it might be used with the no-longer-supported
MySQL 5.5 or MariaDB 5.5, which do not support IMPORT TABLESPACE
nor system tablespace format changes that were made in MariaDB 10.3),
then the innochecksum tool from MariaDB 10.2, 10.3, 10.4, 10.5 or
MySQL 5.7 can be used.
Reviewed by: Thirunarayanan Balathandayuthapani
5 years ago MDEV-17958 Make bug-endian innodb_checksum_algorithm=crc32 optional
In MySQL 5.7, it was noticed that files are not portable between
big-endian and little-endian processor architectures
(such as SPARC and x86), because the original implementation of
innodb_checksum_algorithm=crc32 was not byte order agnostic.
A byte order agnostic implementation of innodb_checksum_algorithm=crc32
was only added to MySQL 5.7, not backported to 5.6. Consequently,
MariaDB Server versions 10.0 and 10.1 only contain the CRC-32C
implementation that works incorrectly on big-endian architectures,
and MariaDB Server 10.2.2 got the byte-order agnostic CRC-32C
implementation from MySQL 5.7.
MySQL 5.7 introduced a "legacy crc32" variant that is functionally
equivalent to the big-endian version of the original crc32 implementation.
Thanks to this variant, old data files can be transferred from big-endian
systems to newer versions.
Introducing new variants of checksum algorithms (without introducing
new names for them, or something on the pages themselves to identify
the algorithm) generally is a bad idea, because each checksum algorithm
is like a lottery ticket. The more algorithms you try, the more likely
it will be for the checksum to match on a corrupted page.
So, essentially MySQL 5.7 weakened innodb_checksum_algorithm=crc32,
and MariaDB 10.2.2 inherited this weakening.
We introduce a build option that together with MDEV-17957
makes innodb_checksum_algorithm=strict_crc32 strict again
by only allowing one variant of the checksum to match.
WITH_INNODB_BUG_ENDIAN_CRC32: A new cmake option for enabling the
bug-compatible "legacy crc32" checksum. This is only enabled on
big-endian systems by default, to facilitate an upgrade from
MariaDB 10.0 or 10.1. Checked by #ifdef INNODB_BUG_ENDIAN_CRC32.
ut_crc32_byte_by_byte: Remove (unused function).
legacy_big_endian_checksum: Remove. This variable seems to have
unnecessarily complicated the logic. When the weakening is enabled,
we must always fall back to the buggy checksum.
buf_page_check_crc32(): A helper function to compute one or
two CRC-32C variants.
7 years ago MDEV-25105 Remove innodb_checksum_algorithm values none,innodb,...
Historically, InnoDB supported a buggy page checksum algorithm that did not
compute a checksum over the full page. Later, well before MySQL 4.1
introduced .ibd files and the innodb_file_per_table option, the algorithm
was corrected and the first 4 bytes of each page were redefined to be
a checksum.
The original checksum was so slow that an option to disable page checksum
was introduced for benchmarketing purposes.
The Intel Nehalem microarchitecture introduced the SSE4.2 instruction set
extension, which includes instructions for faster computation of CRC-32C.
In MySQL 5.6 (and MariaDB 10.0), innodb_checksum_algorithm=crc32 was
implemented to make of that. As that option was changed to be the default
in MySQL 5.7, a bug was found on big-endian platforms and some work-around
code was added to weaken that checksum further. MariaDB disables that
work-around by default since MDEV-17958.
Later, SIMD-accelerated CRC-32C has been implemented in MariaDB for POWER
and ARM and also for IA-32/AMD64, making use of carry-less multiplication
where available.
Long story short, innodb_checksum_algorithm=crc32 is faster and more secure
than the pre-MySQL 5.6 checksum, called innodb_checksum_algorithm=innodb.
It should have removed any need to use innodb_checksum_algorithm=none.
The setting innodb_checksum_algorithm=crc32 is the default in
MySQL 5.7 and MariaDB Server 10.2, 10.3, 10.4. In MariaDB 10.5,
MDEV-19534 made innodb_checksum_algorithm=full_crc32 the default.
It is even faster and more secure.
The default settings in MariaDB do allow old data files to be read,
no matter if a worse checksum algorithm had been used.
(Unfortunately, before innodb_checksum_algorithm=full_crc32,
the data files did not identify which checksum algorithm is being used.)
The non-default settings innodb_checksum_algorithm=strict_crc32 or
innodb_checksum_algorithm=strict_full_crc32 would only allow CRC-32C
checksums. The incompatibility with old data files is why they are
not the default.
The newest server not to support innodb_checksum_algorithm=crc32
were MySQL 5.5 and MariaDB 5.5. Both have reached their end of life.
A valid reason for using innodb_checksum_algorithm=innodb could have
been the ability to downgrade. If it is really needed, data files
can be converted with an older version of the innochecksum utility.
Because there is no good reason to allow data files to be written
with insecure checksums, we will reject those option values:
innodb_checksum_algorithm=none
innodb_checksum_algorithm=innodb
innodb_checksum_algorithm=strict_none
innodb_checksum_algorithm=strict_innodb
Furthermore, the following innochecksum options will be removed,
because only strict crc32 will be supported:
innochecksum --strict-check=crc32
innochecksum -C crc32
innochecksum --write=crc32
innochecksum -w crc32
If a user wishes to convert a data file to use a different checksum
(so that it might be used with the no-longer-supported
MySQL 5.5 or MariaDB 5.5, which do not support IMPORT TABLESPACE
nor system tablespace format changes that were made in MariaDB 10.3),
then the innochecksum tool from MariaDB 10.2, 10.3, 10.4, 10.5 or
MySQL 5.7 can be used.
Reviewed by: Thirunarayanan Balathandayuthapani
5 years ago MDEV-25105 Remove innodb_checksum_algorithm values none,innodb,...
Historically, InnoDB supported a buggy page checksum algorithm that did not
compute a checksum over the full page. Later, well before MySQL 4.1
introduced .ibd files and the innodb_file_per_table option, the algorithm
was corrected and the first 4 bytes of each page were redefined to be
a checksum.
The original checksum was so slow that an option to disable page checksum
was introduced for benchmarketing purposes.
The Intel Nehalem microarchitecture introduced the SSE4.2 instruction set
extension, which includes instructions for faster computation of CRC-32C.
In MySQL 5.6 (and MariaDB 10.0), innodb_checksum_algorithm=crc32 was
implemented to make of that. As that option was changed to be the default
in MySQL 5.7, a bug was found on big-endian platforms and some work-around
code was added to weaken that checksum further. MariaDB disables that
work-around by default since MDEV-17958.
Later, SIMD-accelerated CRC-32C has been implemented in MariaDB for POWER
and ARM and also for IA-32/AMD64, making use of carry-less multiplication
where available.
Long story short, innodb_checksum_algorithm=crc32 is faster and more secure
than the pre-MySQL 5.6 checksum, called innodb_checksum_algorithm=innodb.
It should have removed any need to use innodb_checksum_algorithm=none.
The setting innodb_checksum_algorithm=crc32 is the default in
MySQL 5.7 and MariaDB Server 10.2, 10.3, 10.4. In MariaDB 10.5,
MDEV-19534 made innodb_checksum_algorithm=full_crc32 the default.
It is even faster and more secure.
The default settings in MariaDB do allow old data files to be read,
no matter if a worse checksum algorithm had been used.
(Unfortunately, before innodb_checksum_algorithm=full_crc32,
the data files did not identify which checksum algorithm is being used.)
The non-default settings innodb_checksum_algorithm=strict_crc32 or
innodb_checksum_algorithm=strict_full_crc32 would only allow CRC-32C
checksums. The incompatibility with old data files is why they are
not the default.
The newest server not to support innodb_checksum_algorithm=crc32
were MySQL 5.5 and MariaDB 5.5. Both have reached their end of life.
A valid reason for using innodb_checksum_algorithm=innodb could have
been the ability to downgrade. If it is really needed, data files
can be converted with an older version of the innochecksum utility.
Because there is no good reason to allow data files to be written
with insecure checksums, we will reject those option values:
innodb_checksum_algorithm=none
innodb_checksum_algorithm=innodb
innodb_checksum_algorithm=strict_none
innodb_checksum_algorithm=strict_innodb
Furthermore, the following innochecksum options will be removed,
because only strict crc32 will be supported:
innochecksum --strict-check=crc32
innochecksum -C crc32
innochecksum --write=crc32
innochecksum -w crc32
If a user wishes to convert a data file to use a different checksum
(so that it might be used with the no-longer-supported
MySQL 5.5 or MariaDB 5.5, which do not support IMPORT TABLESPACE
nor system tablespace format changes that were made in MariaDB 10.3),
then the innochecksum tool from MariaDB 10.2, 10.3, 10.4, 10.5 or
MySQL 5.7 can be used.
Reviewed by: Thirunarayanan Balathandayuthapani
5 years ago MDEV-34830: LSN in the future is not being treated as serious corruption
The invariant of write-ahead logging is that before any change to a
page is written to the data file, the corresponding log record must
must first have been durably written.
In crash recovery, there were some sloppy checks for this. Let us
implement accurate checks and flag an inconsistency as a hard error,
so that we can avoid further corruption of a corrupted database.
For data extraction from the corrupted database, innodb_force_recovery
can be used.
Before recovery is reading any data pages or invoking
buf_dblwr_t::recover() to recover torn pages from the
doublewrite buffer, InnoDB will have parsed the log until the
final LSN and updated log_sys.lsn to that. So, we can rely on
log_sys.lsn at all times. The doublewrite buffer recovery has been
refactored in such a way that the recv_sys.dblwr.pages may be consulted
while discovering files and their page sizes, but nothing will be
written back to data files before buf_dblwr_t::recover() is invoked.
recv_max_page_lsn, recv_lsn_checks_on: Remove.
recv_sys_t::validate_checkpoint(): Validate the write-ahead-logging
condition at the end of the recovery.
recv_dblwr_t::validate_page(): Keep track of the maximum LSN
(if we are checking a non-doublewrite copy of a page) but
do not complain LSN being in the future. The doublewrite buffer
is a special case, because it will be read early during recovery.
Besides, starting with commit 762bcb81b5bf9bbde61fed59afb26417f4ce1e86
the dblwr=true copies of pages may legitimately be "too new".
recv_dblwr_t::find_page(): Find a valid page with the smallest
FIL_PAGE_LSN that is in the valid range for recovery.
recv_dblwr_t::restore_first_page(): Replaced by find_page().
Only buf_dblwr_t::recover() will write to data files.
buf_dblwr_t::recover(): Simplify the message output. Do attempt
doublewrite recovery on user page read error. Ignore doublewrite
pages whose FIL_PAGE_LSN is outside the usable bounds. Previously,
we could wrongly recover a too new page from the doublewrite buffer.
It is unlikely that this could have lead to an actual error.
Write back all recovered pages from the doublewrite buffer here,
including for the first page of any tablespace.
buf_page_is_corrupted(): Distinguish the return values
CORRUPTED_FUTURE_LSN and CORRUPTED_OTHER.
buf_page_check_corrupt(): Return the error code DB_CORRUPTION
in case the LSN is in the future.
Datafile::read_first_page_flags(): Split from read_first_page().
Take a copy of the first page as a parameter.
recv_sys_t::free_corrupted_page(): Take the file as a parameter
and return whether a message was displayed. This avoids some duplicated
and incomplete error messages.
buf_page_t::read_complete(): Remove some redundant output and always
display the name of the corrupted file. Never return DB_FAIL;
use it only in internal error handling.
IORequest::read_complete(): Assume that buf_page_t::read_complete()
will have reported any error.
fil_space_t::set_corrupted(): Return whether this is the first time
the tablespace had been flagged as corrupted.
Datafile::validate_first_page(), fil_node_open_file_low(),
fil_node_open_file(), fil_space_t::read_page0(),
fil_node_t::read_page0(): Add a parameter for a copy of the
first page, and a parameter to indicate whether the FIL_PAGE_LSN
check should be suppressed. Before buf_dblwr_t::recover() is
invoked, we cannot validate the FIL_PAGE_LSN, but we can trust the
FSP_SPACE_FLAGS and the tablespace ID that may be present in a
potentially too new copy of a page.
Reviewed by: Debarun Banerjee
12 months ago MDEV-21923: LSN allocation is a bottleneck
The parameter innodb_log_spin_wait_delay will be deprecated and
ignored, because there is no spin loop anymore.
Thanks to commit 685d958e38b825ad9829be311f26729cccf37c46
and commit a635c40648519fd6c3729c9657872a16a0a20821
multiple mtr_t::commit() can concurrently copy their slice of
mtr_t::m_log to the shared log_sys.buf. Each writer would allocate
their own log sequence number by invoking log_t::append_prepare()
while holding a shared log_sys.latch. This function was too heavy,
because it would invoke a minimum of 4 atomic read-modify-write
operations as well as system calls in the supposedly fast code path.
It turns out that with a simpler data structure, instead of having
several data fields that needed to be kept consistent with each other,
we only need one Atomic_relaxed<uint64_t> write_lsn_offset, on which
we can operate using fetch_add(), fetch_sub() as well as a single-bit
fetch_or(), which reasonably modern compilers (GCC 7, Clang 15 or later)
can translate into loop-free code on AMD64.
Before anything can be written to the log, log_sys.clear_mmap()
must be invoked.
log_t::base_lsn: The LSN of the last write_buf() or persist().
This is a rough approximation of log_sys.lsn, which will be removed.
log_t::write_lsn_offset: An Atomic_relaxed<uint64_t> that buffers
updates of write_to_buf and base_lsn.
log_t::buf_free, log_t::max_buf_free, log_t::lsn. Remove.
Replaced by base_lsn and write_lsn_offset.
log_t::buf_size: Always reflects the usable size in append_prepare().
log_t::lsn_lock: Remove. For the memory-mapped log in resize_write(),
there will be a resize_wrap_mutex.
log_t::get_lsn_approx(): Return a lower bound of get_lsn().
This should be exact unless append_prepare_wait() is pending.
log_get_lsn(): A wrapper for log_sys.get_lsn(), which must be invoked
while holding an exclusive log_sys.latch.
recv_recovery_from_checkpoint_start(): Do not invoke fil_names_clear();
it would seem to be unnecessary.
In many places, references to log_sys.get_lsn() are replaced with
log_sys.get_flushed_lsn(), which remains a simple std::atomic::load().
Reviewed by: Debarun Banerjee
6 months ago MDEV-34830: LSN in the future is not being treated as serious corruption
The invariant of write-ahead logging is that before any change to a
page is written to the data file, the corresponding log record must
must first have been durably written.
In crash recovery, there were some sloppy checks for this. Let us
implement accurate checks and flag an inconsistency as a hard error,
so that we can avoid further corruption of a corrupted database.
For data extraction from the corrupted database, innodb_force_recovery
can be used.
Before recovery is reading any data pages or invoking
buf_dblwr_t::recover() to recover torn pages from the
doublewrite buffer, InnoDB will have parsed the log until the
final LSN and updated log_sys.lsn to that. So, we can rely on
log_sys.lsn at all times. The doublewrite buffer recovery has been
refactored in such a way that the recv_sys.dblwr.pages may be consulted
while discovering files and their page sizes, but nothing will be
written back to data files before buf_dblwr_t::recover() is invoked.
recv_max_page_lsn, recv_lsn_checks_on: Remove.
recv_sys_t::validate_checkpoint(): Validate the write-ahead-logging
condition at the end of the recovery.
recv_dblwr_t::validate_page(): Keep track of the maximum LSN
(if we are checking a non-doublewrite copy of a page) but
do not complain LSN being in the future. The doublewrite buffer
is a special case, because it will be read early during recovery.
Besides, starting with commit 762bcb81b5bf9bbde61fed59afb26417f4ce1e86
the dblwr=true copies of pages may legitimately be "too new".
recv_dblwr_t::find_page(): Find a valid page with the smallest
FIL_PAGE_LSN that is in the valid range for recovery.
recv_dblwr_t::restore_first_page(): Replaced by find_page().
Only buf_dblwr_t::recover() will write to data files.
buf_dblwr_t::recover(): Simplify the message output. Do attempt
doublewrite recovery on user page read error. Ignore doublewrite
pages whose FIL_PAGE_LSN is outside the usable bounds. Previously,
we could wrongly recover a too new page from the doublewrite buffer.
It is unlikely that this could have lead to an actual error.
Write back all recovered pages from the doublewrite buffer here,
including for the first page of any tablespace.
buf_page_is_corrupted(): Distinguish the return values
CORRUPTED_FUTURE_LSN and CORRUPTED_OTHER.
buf_page_check_corrupt(): Return the error code DB_CORRUPTION
in case the LSN is in the future.
Datafile::read_first_page_flags(): Split from read_first_page().
Take a copy of the first page as a parameter.
recv_sys_t::free_corrupted_page(): Take the file as a parameter
and return whether a message was displayed. This avoids some duplicated
and incomplete error messages.
buf_page_t::read_complete(): Remove some redundant output and always
display the name of the corrupted file. Never return DB_FAIL;
use it only in internal error handling.
IORequest::read_complete(): Assume that buf_page_t::read_complete()
will have reported any error.
fil_space_t::set_corrupted(): Return whether this is the first time
the tablespace had been flagged as corrupted.
Datafile::validate_first_page(), fil_node_open_file_low(),
fil_node_open_file(), fil_space_t::read_page0(),
fil_node_t::read_page0(): Add a parameter for a copy of the
first page, and a parameter to indicate whether the FIL_PAGE_LSN
check should be suppressed. Before buf_dblwr_t::recover() is
invoked, we cannot validate the FIL_PAGE_LSN, but we can trust the
FSP_SPACE_FLAGS and the tablespace ID that may be present in a
potentially too new copy of a page.
Reviewed by: Debarun Banerjee
12 months ago MDEV-21923: LSN allocation is a bottleneck
The parameter innodb_log_spin_wait_delay will be deprecated and
ignored, because there is no spin loop anymore.
Thanks to commit 685d958e38b825ad9829be311f26729cccf37c46
and commit a635c40648519fd6c3729c9657872a16a0a20821
multiple mtr_t::commit() can concurrently copy their slice of
mtr_t::m_log to the shared log_sys.buf. Each writer would allocate
their own log sequence number by invoking log_t::append_prepare()
while holding a shared log_sys.latch. This function was too heavy,
because it would invoke a minimum of 4 atomic read-modify-write
operations as well as system calls in the supposedly fast code path.
It turns out that with a simpler data structure, instead of having
several data fields that needed to be kept consistent with each other,
we only need one Atomic_relaxed<uint64_t> write_lsn_offset, on which
we can operate using fetch_add(), fetch_sub() as well as a single-bit
fetch_or(), which reasonably modern compilers (GCC 7, Clang 15 or later)
can translate into loop-free code on AMD64.
Before anything can be written to the log, log_sys.clear_mmap()
must be invoked.
log_t::base_lsn: The LSN of the last write_buf() or persist().
This is a rough approximation of log_sys.lsn, which will be removed.
log_t::write_lsn_offset: An Atomic_relaxed<uint64_t> that buffers
updates of write_to_buf and base_lsn.
log_t::buf_free, log_t::max_buf_free, log_t::lsn. Remove.
Replaced by base_lsn and write_lsn_offset.
log_t::buf_size: Always reflects the usable size in append_prepare().
log_t::lsn_lock: Remove. For the memory-mapped log in resize_write(),
there will be a resize_wrap_mutex.
log_t::get_lsn_approx(): Return a lower bound of get_lsn().
This should be exact unless append_prepare_wait() is pending.
log_get_lsn(): A wrapper for log_sys.get_lsn(), which must be invoked
while holding an exclusive log_sys.latch.
recv_recovery_from_checkpoint_start(): Do not invoke fil_names_clear();
it would seem to be unnecessary.
In many places, references to log_sys.get_lsn() are replaced with
log_sys.get_flushed_lsn(), which remains a simple std::atomic::load().
Reviewed by: Debarun Banerjee
6 months ago MDEV-12026: Implement innodb_checksum_algorithm=full_crc32
MariaDB data-at-rest encryption (innodb_encrypt_tables)
had repurposed the same unused data field that was repurposed
in MySQL 5.7 (and MariaDB 10.2) for the Split Sequence Number (SSN)
field of SPATIAL INDEX. Because of this, MariaDB was unable to
support encryption on SPATIAL INDEX pages.
Furthermore, InnoDB page checksums skipped some bytes, and there
are multiple variations and checksum algorithms. By default,
InnoDB accepts all variations of all algorithms that ever existed.
This unnecessarily weakens the page checksums.
We hereby introduce two more innodb_checksum_algorithm variants
(full_crc32, strict_full_crc32) that are special in a way:
When either setting is active, newly created data files will
carry a flag (fil_space_t::full_crc32()) that indicates that
all pages of the file will use a full CRC-32C checksum over the
entire page contents (excluding the bytes where the checksum
is stored, at the very end of the page). Such files will always
use that checksum, no matter what the parameter
innodb_checksum_algorithm is assigned to.
For old files, the old checksum algorithms will continue to be
used. The value strict_full_crc32 will be equivalent to strict_crc32
and the value full_crc32 will be equivalent to crc32.
ROW_FORMAT=COMPRESSED tables will only use the old format.
These tables do not support new features, such as larger
innodb_page_size or instant ADD/DROP COLUMN. They may be
deprecated in the future. We do not want an unnecessary
file format change for them.
The new full_crc32() format also cleans up the MariaDB tablespace
flags. We will reserve flags to store the page_compressed
compression algorithm, and to store the compressed payload length,
so that checksum can be computed over the compressed (and
possibly encrypted) stream and can be validated without
decrypting or decompressing the page.
In the full_crc32 format, there no longer are separate before-encryption
and after-encryption checksums for pages. The single checksum is
computed on the page contents that is written to the file.
We do not make the new algorithm the default for two reasons.
First, MariaDB 10.4.2 was a beta release, and the default values
of parameters should not change after beta. Second, we did not
yet implement the full_crc32 format for page_compressed pages.
This will be fixed in MDEV-18644.
This is joint work with Marko Mäkelä.
7 years ago MDEV-34830: LSN in the future is not being treated as serious corruption
The invariant of write-ahead logging is that before any change to a
page is written to the data file, the corresponding log record must
must first have been durably written.
In crash recovery, there were some sloppy checks for this. Let us
implement accurate checks and flag an inconsistency as a hard error,
so that we can avoid further corruption of a corrupted database.
For data extraction from the corrupted database, innodb_force_recovery
can be used.
Before recovery is reading any data pages or invoking
buf_dblwr_t::recover() to recover torn pages from the
doublewrite buffer, InnoDB will have parsed the log until the
final LSN and updated log_sys.lsn to that. So, we can rely on
log_sys.lsn at all times. The doublewrite buffer recovery has been
refactored in such a way that the recv_sys.dblwr.pages may be consulted
while discovering files and their page sizes, but nothing will be
written back to data files before buf_dblwr_t::recover() is invoked.
recv_max_page_lsn, recv_lsn_checks_on: Remove.
recv_sys_t::validate_checkpoint(): Validate the write-ahead-logging
condition at the end of the recovery.
recv_dblwr_t::validate_page(): Keep track of the maximum LSN
(if we are checking a non-doublewrite copy of a page) but
do not complain LSN being in the future. The doublewrite buffer
is a special case, because it will be read early during recovery.
Besides, starting with commit 762bcb81b5bf9bbde61fed59afb26417f4ce1e86
the dblwr=true copies of pages may legitimately be "too new".
recv_dblwr_t::find_page(): Find a valid page with the smallest
FIL_PAGE_LSN that is in the valid range for recovery.
recv_dblwr_t::restore_first_page(): Replaced by find_page().
Only buf_dblwr_t::recover() will write to data files.
buf_dblwr_t::recover(): Simplify the message output. Do attempt
doublewrite recovery on user page read error. Ignore doublewrite
pages whose FIL_PAGE_LSN is outside the usable bounds. Previously,
we could wrongly recover a too new page from the doublewrite buffer.
It is unlikely that this could have lead to an actual error.
Write back all recovered pages from the doublewrite buffer here,
including for the first page of any tablespace.
buf_page_is_corrupted(): Distinguish the return values
CORRUPTED_FUTURE_LSN and CORRUPTED_OTHER.
buf_page_check_corrupt(): Return the error code DB_CORRUPTION
in case the LSN is in the future.
Datafile::read_first_page_flags(): Split from read_first_page().
Take a copy of the first page as a parameter.
recv_sys_t::free_corrupted_page(): Take the file as a parameter
and return whether a message was displayed. This avoids some duplicated
and incomplete error messages.
buf_page_t::read_complete(): Remove some redundant output and always
display the name of the corrupted file. Never return DB_FAIL;
use it only in internal error handling.
IORequest::read_complete(): Assume that buf_page_t::read_complete()
will have reported any error.
fil_space_t::set_corrupted(): Return whether this is the first time
the tablespace had been flagged as corrupted.
Datafile::validate_first_page(), fil_node_open_file_low(),
fil_node_open_file(), fil_space_t::read_page0(),
fil_node_t::read_page0(): Add a parameter for a copy of the
first page, and a parameter to indicate whether the FIL_PAGE_LSN
check should be suppressed. Before buf_dblwr_t::recover() is
invoked, we cannot validate the FIL_PAGE_LSN, but we can trust the
FSP_SPACE_FLAGS and the tablespace ID that may be present in a
potentially too new copy of a page.
Reviewed by: Debarun Banerjee
12 months ago MDEV-21923: LSN allocation is a bottleneck
The parameter innodb_log_spin_wait_delay will be deprecated and
ignored, because there is no spin loop anymore.
Thanks to commit 685d958e38b825ad9829be311f26729cccf37c46
and commit a635c40648519fd6c3729c9657872a16a0a20821
multiple mtr_t::commit() can concurrently copy their slice of
mtr_t::m_log to the shared log_sys.buf. Each writer would allocate
their own log sequence number by invoking log_t::append_prepare()
while holding a shared log_sys.latch. This function was too heavy,
because it would invoke a minimum of 4 atomic read-modify-write
operations as well as system calls in the supposedly fast code path.
It turns out that with a simpler data structure, instead of having
several data fields that needed to be kept consistent with each other,
we only need one Atomic_relaxed<uint64_t> write_lsn_offset, on which
we can operate using fetch_add(), fetch_sub() as well as a single-bit
fetch_or(), which reasonably modern compilers (GCC 7, Clang 15 or later)
can translate into loop-free code on AMD64.
Before anything can be written to the log, log_sys.clear_mmap()
must be invoked.
log_t::base_lsn: The LSN of the last write_buf() or persist().
This is a rough approximation of log_sys.lsn, which will be removed.
log_t::write_lsn_offset: An Atomic_relaxed<uint64_t> that buffers
updates of write_to_buf and base_lsn.
log_t::buf_free, log_t::max_buf_free, log_t::lsn. Remove.
Replaced by base_lsn and write_lsn_offset.
log_t::buf_size: Always reflects the usable size in append_prepare().
log_t::lsn_lock: Remove. For the memory-mapped log in resize_write(),
there will be a resize_wrap_mutex.
log_t::get_lsn_approx(): Return a lower bound of get_lsn().
This should be exact unless append_prepare_wait() is pending.
log_get_lsn(): A wrapper for log_sys.get_lsn(), which must be invoked
while holding an exclusive log_sys.latch.
recv_recovery_from_checkpoint_start(): Do not invoke fil_names_clear();
it would seem to be unnecessary.
In many places, references to log_sys.get_lsn() are replaced with
log_sys.get_flushed_lsn(), which remains a simple std::atomic::load().
Reviewed by: Debarun Banerjee
6 months ago MDEV-34830: LSN in the future is not being treated as serious corruption
The invariant of write-ahead logging is that before any change to a
page is written to the data file, the corresponding log record must
must first have been durably written.
In crash recovery, there were some sloppy checks for this. Let us
implement accurate checks and flag an inconsistency as a hard error,
so that we can avoid further corruption of a corrupted database.
For data extraction from the corrupted database, innodb_force_recovery
can be used.
Before recovery is reading any data pages or invoking
buf_dblwr_t::recover() to recover torn pages from the
doublewrite buffer, InnoDB will have parsed the log until the
final LSN and updated log_sys.lsn to that. So, we can rely on
log_sys.lsn at all times. The doublewrite buffer recovery has been
refactored in such a way that the recv_sys.dblwr.pages may be consulted
while discovering files and their page sizes, but nothing will be
written back to data files before buf_dblwr_t::recover() is invoked.
recv_max_page_lsn, recv_lsn_checks_on: Remove.
recv_sys_t::validate_checkpoint(): Validate the write-ahead-logging
condition at the end of the recovery.
recv_dblwr_t::validate_page(): Keep track of the maximum LSN
(if we are checking a non-doublewrite copy of a page) but
do not complain LSN being in the future. The doublewrite buffer
is a special case, because it will be read early during recovery.
Besides, starting with commit 762bcb81b5bf9bbde61fed59afb26417f4ce1e86
the dblwr=true copies of pages may legitimately be "too new".
recv_dblwr_t::find_page(): Find a valid page with the smallest
FIL_PAGE_LSN that is in the valid range for recovery.
recv_dblwr_t::restore_first_page(): Replaced by find_page().
Only buf_dblwr_t::recover() will write to data files.
buf_dblwr_t::recover(): Simplify the message output. Do attempt
doublewrite recovery on user page read error. Ignore doublewrite
pages whose FIL_PAGE_LSN is outside the usable bounds. Previously,
we could wrongly recover a too new page from the doublewrite buffer.
It is unlikely that this could have lead to an actual error.
Write back all recovered pages from the doublewrite buffer here,
including for the first page of any tablespace.
buf_page_is_corrupted(): Distinguish the return values
CORRUPTED_FUTURE_LSN and CORRUPTED_OTHER.
buf_page_check_corrupt(): Return the error code DB_CORRUPTION
in case the LSN is in the future.
Datafile::read_first_page_flags(): Split from read_first_page().
Take a copy of the first page as a parameter.
recv_sys_t::free_corrupted_page(): Take the file as a parameter
and return whether a message was displayed. This avoids some duplicated
and incomplete error messages.
buf_page_t::read_complete(): Remove some redundant output and always
display the name of the corrupted file. Never return DB_FAIL;
use it only in internal error handling.
IORequest::read_complete(): Assume that buf_page_t::read_complete()
will have reported any error.
fil_space_t::set_corrupted(): Return whether this is the first time
the tablespace had been flagged as corrupted.
Datafile::validate_first_page(), fil_node_open_file_low(),
fil_node_open_file(), fil_space_t::read_page0(),
fil_node_t::read_page0(): Add a parameter for a copy of the
first page, and a parameter to indicate whether the FIL_PAGE_LSN
check should be suppressed. Before buf_dblwr_t::recover() is
invoked, we cannot validate the FIL_PAGE_LSN, but we can trust the
FSP_SPACE_FLAGS and the tablespace ID that may be present in a
potentially too new copy of a page.
Reviewed by: Debarun Banerjee
12 months ago MDEV-12026: Implement innodb_checksum_algorithm=full_crc32
MariaDB data-at-rest encryption (innodb_encrypt_tables)
had repurposed the same unused data field that was repurposed
in MySQL 5.7 (and MariaDB 10.2) for the Split Sequence Number (SSN)
field of SPATIAL INDEX. Because of this, MariaDB was unable to
support encryption on SPATIAL INDEX pages.
Furthermore, InnoDB page checksums skipped some bytes, and there
are multiple variations and checksum algorithms. By default,
InnoDB accepts all variations of all algorithms that ever existed.
This unnecessarily weakens the page checksums.
We hereby introduce two more innodb_checksum_algorithm variants
(full_crc32, strict_full_crc32) that are special in a way:
When either setting is active, newly created data files will
carry a flag (fil_space_t::full_crc32()) that indicates that
all pages of the file will use a full CRC-32C checksum over the
entire page contents (excluding the bytes where the checksum
is stored, at the very end of the page). Such files will always
use that checksum, no matter what the parameter
innodb_checksum_algorithm is assigned to.
For old files, the old checksum algorithms will continue to be
used. The value strict_full_crc32 will be equivalent to strict_crc32
and the value full_crc32 will be equivalent to crc32.
ROW_FORMAT=COMPRESSED tables will only use the old format.
These tables do not support new features, such as larger
innodb_page_size or instant ADD/DROP COLUMN. They may be
deprecated in the future. We do not want an unnecessary
file format change for them.
The new full_crc32() format also cleans up the MariaDB tablespace
flags. We will reserve flags to store the page_compressed
compression algorithm, and to store the compressed payload length,
so that checksum can be computed over the compressed (and
possibly encrypted) stream and can be validated without
decrypting or decompressing the page.
In the full_crc32 format, there no longer are separate before-encryption
and after-encryption checksums for pages. The single checksum is
computed on the page contents that is written to the file.
We do not make the new algorithm the default for two reasons.
First, MariaDB 10.4.2 was a beta release, and the default values
of parameters should not change after beta. Second, we did not
yet implement the full_crc32 format for page_compressed pages.
This will be fixed in MDEV-18644.
This is joint work with Marko Mäkelä.
7 years ago MDEV-34830: LSN in the future is not being treated as serious corruption
The invariant of write-ahead logging is that before any change to a
page is written to the data file, the corresponding log record must
must first have been durably written.
In crash recovery, there were some sloppy checks for this. Let us
implement accurate checks and flag an inconsistency as a hard error,
so that we can avoid further corruption of a corrupted database.
For data extraction from the corrupted database, innodb_force_recovery
can be used.
Before recovery is reading any data pages or invoking
buf_dblwr_t::recover() to recover torn pages from the
doublewrite buffer, InnoDB will have parsed the log until the
final LSN and updated log_sys.lsn to that. So, we can rely on
log_sys.lsn at all times. The doublewrite buffer recovery has been
refactored in such a way that the recv_sys.dblwr.pages may be consulted
while discovering files and their page sizes, but nothing will be
written back to data files before buf_dblwr_t::recover() is invoked.
recv_max_page_lsn, recv_lsn_checks_on: Remove.
recv_sys_t::validate_checkpoint(): Validate the write-ahead-logging
condition at the end of the recovery.
recv_dblwr_t::validate_page(): Keep track of the maximum LSN
(if we are checking a non-doublewrite copy of a page) but
do not complain LSN being in the future. The doublewrite buffer
is a special case, because it will be read early during recovery.
Besides, starting with commit 762bcb81b5bf9bbde61fed59afb26417f4ce1e86
the dblwr=true copies of pages may legitimately be "too new".
recv_dblwr_t::find_page(): Find a valid page with the smallest
FIL_PAGE_LSN that is in the valid range for recovery.
recv_dblwr_t::restore_first_page(): Replaced by find_page().
Only buf_dblwr_t::recover() will write to data files.
buf_dblwr_t::recover(): Simplify the message output. Do attempt
doublewrite recovery on user page read error. Ignore doublewrite
pages whose FIL_PAGE_LSN is outside the usable bounds. Previously,
we could wrongly recover a too new page from the doublewrite buffer.
It is unlikely that this could have lead to an actual error.
Write back all recovered pages from the doublewrite buffer here,
including for the first page of any tablespace.
buf_page_is_corrupted(): Distinguish the return values
CORRUPTED_FUTURE_LSN and CORRUPTED_OTHER.
buf_page_check_corrupt(): Return the error code DB_CORRUPTION
in case the LSN is in the future.
Datafile::read_first_page_flags(): Split from read_first_page().
Take a copy of the first page as a parameter.
recv_sys_t::free_corrupted_page(): Take the file as a parameter
and return whether a message was displayed. This avoids some duplicated
and incomplete error messages.
buf_page_t::read_complete(): Remove some redundant output and always
display the name of the corrupted file. Never return DB_FAIL;
use it only in internal error handling.
IORequest::read_complete(): Assume that buf_page_t::read_complete()
will have reported any error.
fil_space_t::set_corrupted(): Return whether this is the first time
the tablespace had been flagged as corrupted.
Datafile::validate_first_page(), fil_node_open_file_low(),
fil_node_open_file(), fil_space_t::read_page0(),
fil_node_t::read_page0(): Add a parameter for a copy of the
first page, and a parameter to indicate whether the FIL_PAGE_LSN
check should be suppressed. Before buf_dblwr_t::recover() is
invoked, we cannot validate the FIL_PAGE_LSN, but we can trust the
FSP_SPACE_FLAGS and the tablespace ID that may be present in a
potentially too new copy of a page.
Reviewed by: Debarun Banerjee
12 months ago MDEV-12026: Implement innodb_checksum_algorithm=full_crc32
MariaDB data-at-rest encryption (innodb_encrypt_tables)
had repurposed the same unused data field that was repurposed
in MySQL 5.7 (and MariaDB 10.2) for the Split Sequence Number (SSN)
field of SPATIAL INDEX. Because of this, MariaDB was unable to
support encryption on SPATIAL INDEX pages.
Furthermore, InnoDB page checksums skipped some bytes, and there
are multiple variations and checksum algorithms. By default,
InnoDB accepts all variations of all algorithms that ever existed.
This unnecessarily weakens the page checksums.
We hereby introduce two more innodb_checksum_algorithm variants
(full_crc32, strict_full_crc32) that are special in a way:
When either setting is active, newly created data files will
carry a flag (fil_space_t::full_crc32()) that indicates that
all pages of the file will use a full CRC-32C checksum over the
entire page contents (excluding the bytes where the checksum
is stored, at the very end of the page). Such files will always
use that checksum, no matter what the parameter
innodb_checksum_algorithm is assigned to.
For old files, the old checksum algorithms will continue to be
used. The value strict_full_crc32 will be equivalent to strict_crc32
and the value full_crc32 will be equivalent to crc32.
ROW_FORMAT=COMPRESSED tables will only use the old format.
These tables do not support new features, such as larger
innodb_page_size or instant ADD/DROP COLUMN. They may be
deprecated in the future. We do not want an unnecessary
file format change for them.
The new full_crc32() format also cleans up the MariaDB tablespace
flags. We will reserve flags to store the page_compressed
compression algorithm, and to store the compressed payload length,
so that checksum can be computed over the compressed (and
possibly encrypted) stream and can be validated without
decrypting or decompressing the page.
In the full_crc32 format, there no longer are separate before-encryption
and after-encryption checksums for pages. The single checksum is
computed on the page contents that is written to the file.
We do not make the new algorithm the default for two reasons.
First, MariaDB 10.4.2 was a beta release, and the default values
of parameters should not change after beta. Second, we did not
yet implement the full_crc32 format for page_compressed pages.
This will be fixed in MDEV-18644.
This is joint work with Marko Mäkelä.
7 years ago MDEV-34830: LSN in the future is not being treated as serious corruption
The invariant of write-ahead logging is that before any change to a
page is written to the data file, the corresponding log record must
must first have been durably written.
In crash recovery, there were some sloppy checks for this. Let us
implement accurate checks and flag an inconsistency as a hard error,
so that we can avoid further corruption of a corrupted database.
For data extraction from the corrupted database, innodb_force_recovery
can be used.
Before recovery is reading any data pages or invoking
buf_dblwr_t::recover() to recover torn pages from the
doublewrite buffer, InnoDB will have parsed the log until the
final LSN and updated log_sys.lsn to that. So, we can rely on
log_sys.lsn at all times. The doublewrite buffer recovery has been
refactored in such a way that the recv_sys.dblwr.pages may be consulted
while discovering files and their page sizes, but nothing will be
written back to data files before buf_dblwr_t::recover() is invoked.
recv_max_page_lsn, recv_lsn_checks_on: Remove.
recv_sys_t::validate_checkpoint(): Validate the write-ahead-logging
condition at the end of the recovery.
recv_dblwr_t::validate_page(): Keep track of the maximum LSN
(if we are checking a non-doublewrite copy of a page) but
do not complain LSN being in the future. The doublewrite buffer
is a special case, because it will be read early during recovery.
Besides, starting with commit 762bcb81b5bf9bbde61fed59afb26417f4ce1e86
the dblwr=true copies of pages may legitimately be "too new".
recv_dblwr_t::find_page(): Find a valid page with the smallest
FIL_PAGE_LSN that is in the valid range for recovery.
recv_dblwr_t::restore_first_page(): Replaced by find_page().
Only buf_dblwr_t::recover() will write to data files.
buf_dblwr_t::recover(): Simplify the message output. Do attempt
doublewrite recovery on user page read error. Ignore doublewrite
pages whose FIL_PAGE_LSN is outside the usable bounds. Previously,
we could wrongly recover a too new page from the doublewrite buffer.
It is unlikely that this could have lead to an actual error.
Write back all recovered pages from the doublewrite buffer here,
including for the first page of any tablespace.
buf_page_is_corrupted(): Distinguish the return values
CORRUPTED_FUTURE_LSN and CORRUPTED_OTHER.
buf_page_check_corrupt(): Return the error code DB_CORRUPTION
in case the LSN is in the future.
Datafile::read_first_page_flags(): Split from read_first_page().
Take a copy of the first page as a parameter.
recv_sys_t::free_corrupted_page(): Take the file as a parameter
and return whether a message was displayed. This avoids some duplicated
and incomplete error messages.
buf_page_t::read_complete(): Remove some redundant output and always
display the name of the corrupted file. Never return DB_FAIL;
use it only in internal error handling.
IORequest::read_complete(): Assume that buf_page_t::read_complete()
will have reported any error.
fil_space_t::set_corrupted(): Return whether this is the first time
the tablespace had been flagged as corrupted.
Datafile::validate_first_page(), fil_node_open_file_low(),
fil_node_open_file(), fil_space_t::read_page0(),
fil_node_t::read_page0(): Add a parameter for a copy of the
first page, and a parameter to indicate whether the FIL_PAGE_LSN
check should be suppressed. Before buf_dblwr_t::recover() is
invoked, we cannot validate the FIL_PAGE_LSN, but we can trust the
FSP_SPACE_FLAGS and the tablespace ID that may be present in a
potentially too new copy of a page.
Reviewed by: Debarun Banerjee
12 months ago MDEV-12026: Implement innodb_checksum_algorithm=full_crc32
MariaDB data-at-rest encryption (innodb_encrypt_tables)
had repurposed the same unused data field that was repurposed
in MySQL 5.7 (and MariaDB 10.2) for the Split Sequence Number (SSN)
field of SPATIAL INDEX. Because of this, MariaDB was unable to
support encryption on SPATIAL INDEX pages.
Furthermore, InnoDB page checksums skipped some bytes, and there
are multiple variations and checksum algorithms. By default,
InnoDB accepts all variations of all algorithms that ever existed.
This unnecessarily weakens the page checksums.
We hereby introduce two more innodb_checksum_algorithm variants
(full_crc32, strict_full_crc32) that are special in a way:
When either setting is active, newly created data files will
carry a flag (fil_space_t::full_crc32()) that indicates that
all pages of the file will use a full CRC-32C checksum over the
entire page contents (excluding the bytes where the checksum
is stored, at the very end of the page). Such files will always
use that checksum, no matter what the parameter
innodb_checksum_algorithm is assigned to.
For old files, the old checksum algorithms will continue to be
used. The value strict_full_crc32 will be equivalent to strict_crc32
and the value full_crc32 will be equivalent to crc32.
ROW_FORMAT=COMPRESSED tables will only use the old format.
These tables do not support new features, such as larger
innodb_page_size or instant ADD/DROP COLUMN. They may be
deprecated in the future. We do not want an unnecessary
file format change for them.
The new full_crc32() format also cleans up the MariaDB tablespace
flags. We will reserve flags to store the page_compressed
compression algorithm, and to store the compressed payload length,
so that checksum can be computed over the compressed (and
possibly encrypted) stream and can be validated without
decrypting or decompressing the page.
In the full_crc32 format, there no longer are separate before-encryption
and after-encryption checksums for pages. The single checksum is
computed on the page contents that is written to the file.
We do not make the new algorithm the default for two reasons.
First, MariaDB 10.4.2 was a beta release, and the default values
of parameters should not change after beta. Second, we did not
yet implement the full_crc32 format for page_compressed pages.
This will be fixed in MDEV-18644.
This is joint work with Marko Mäkelä.
7 years ago MDEV-18644: Support full_crc32 for page_compressed
This is a follow-up task to MDEV-12026, which introduced
innodb_checksum_algorithm=full_crc32 and a simpler page format.
MDEV-12026 did not enable full_crc32 for page_compressed tables,
which we will be doing now.
This is joint work with Thirunarayanan Balathandayuthapani.
For innodb_checksum_algorithm=full_crc32 we change the
page_compressed format as follows:
FIL_PAGE_TYPE: The most significant bit will be set to indicate
page_compressed format. The least significant bits will contain
the compressed page size, rounded up to a multiple of 256 bytes.
The checksum will be stored in the last 4 bytes of the page
(whether it is the full page or a page_compressed page whose
size is determined by FIL_PAGE_TYPE), covering all preceding
bytes of the page. If encryption is used, then the page will
be encrypted between compression and computing the checksum.
For page_compressed, FIL_PAGE_LSN will not be repeated at
the end of the page.
FSP_SPACE_FLAGS (already implemented as part of MDEV-12026):
We will store the innodb_compression_algorithm that may be used
to compress pages. Previously, the choice of algorithm was written
to each compressed data page separately, and one would be unable
to know in advance which compression algorithm(s) are used.
fil_space_t::full_crc32_page_compressed_len(): Determine if the
page_compressed algorithm of the tablespace needs to know the
exact length of the compressed data. If yes, we will reserve and
write an extra byte for this right before the checksum.
buf_page_is_compressed(): Determine if a page uses page_compressed
(in any innodb_checksum_algorithm).
fil_page_decompress(): Pass also fil_space_t::flags so that the
format can be determined.
buf_page_is_zeroes(): Check if a page is full of zero bytes.
buf_page_full_crc32_is_corrupted(): Renamed from
buf_encrypted_full_crc32_page_is_corrupted(). For full_crc32,
we always simply validate the checksum to the page contents,
while the physical page size is explicitly specified by an
unencrypted part of the page header.
buf_page_full_crc32_size(): Determine the size of a full_crc32 page.
buf_dblwr_check_page_lsn(): Make this a debug-only function, because
it involves potentially costly lookups of fil_space_t.
create_table_info_t::check_table_options(),
ha_innobase::check_if_supported_inplace_alter(): Do allow the creation
of SPATIAL INDEX with full_crc32 also when page_compressed is used.
commit_cache_norebuild(): Preserve the compression algorithm when
updating the page_compression_level.
dict_tf_to_fsp_flags(): Set the flags for page compression algorithm.
FIXME: Maybe there should be a table option page_compression_algorithm
and a session variable to back it?
7 years ago MDEV-18644: Support full_crc32 for page_compressed
This is a follow-up task to MDEV-12026, which introduced
innodb_checksum_algorithm=full_crc32 and a simpler page format.
MDEV-12026 did not enable full_crc32 for page_compressed tables,
which we will be doing now.
This is joint work with Thirunarayanan Balathandayuthapani.
For innodb_checksum_algorithm=full_crc32 we change the
page_compressed format as follows:
FIL_PAGE_TYPE: The most significant bit will be set to indicate
page_compressed format. The least significant bits will contain
the compressed page size, rounded up to a multiple of 256 bytes.
The checksum will be stored in the last 4 bytes of the page
(whether it is the full page or a page_compressed page whose
size is determined by FIL_PAGE_TYPE), covering all preceding
bytes of the page. If encryption is used, then the page will
be encrypted between compression and computing the checksum.
For page_compressed, FIL_PAGE_LSN will not be repeated at
the end of the page.
FSP_SPACE_FLAGS (already implemented as part of MDEV-12026):
We will store the innodb_compression_algorithm that may be used
to compress pages. Previously, the choice of algorithm was written
to each compressed data page separately, and one would be unable
to know in advance which compression algorithm(s) are used.
fil_space_t::full_crc32_page_compressed_len(): Determine if the
page_compressed algorithm of the tablespace needs to know the
exact length of the compressed data. If yes, we will reserve and
write an extra byte for this right before the checksum.
buf_page_is_compressed(): Determine if a page uses page_compressed
(in any innodb_checksum_algorithm).
fil_page_decompress(): Pass also fil_space_t::flags so that the
format can be determined.
buf_page_is_zeroes(): Check if a page is full of zero bytes.
buf_page_full_crc32_is_corrupted(): Renamed from
buf_encrypted_full_crc32_page_is_corrupted(). For full_crc32,
we always simply validate the checksum to the page contents,
while the physical page size is explicitly specified by an
unencrypted part of the page header.
buf_page_full_crc32_size(): Determine the size of a full_crc32 page.
buf_dblwr_check_page_lsn(): Make this a debug-only function, because
it involves potentially costly lookups of fil_space_t.
create_table_info_t::check_table_options(),
ha_innobase::check_if_supported_inplace_alter(): Do allow the creation
of SPATIAL INDEX with full_crc32 also when page_compressed is used.
commit_cache_norebuild(): Preserve the compression algorithm when
updating the page_compression_level.
dict_tf_to_fsp_flags(): Set the flags for page compression algorithm.
FIXME: Maybe there should be a table option page_compression_algorithm
and a session variable to back it?
7 years ago MDEV-12602 InnoDB: Failing assertion: space->n_pending_ops == 0
This fixes a regression caused by MDEV-12428.
When we introduced a variant of fil_space_acquire() that could
increment space->n_pending_ops after space->stop_new_ops was set,
the logic of fil_check_pending_operations() was broken.
fil_space_t::n_pending_ios: A new field to track read or write
access from the buffer pool routines immediately before a block
write or after a block read in the file system.
fil_space_acquire_for_io(), fil_space_release_for_io(): Similar
to fil_space_acquire_silent() and fil_space_release(), but
modify fil_space_t::n_pending_ios instead of fil_space_t::n_pending_ops.
Adjust a number of places accordingly, and remove some redundant
tablespace lookups.
The following parts of this fix differ from the 10.2 version of this fix:
buf_page_get_corrupt(): Add a tablespace parameter.
In 10.2, we already had a two-phase process of freeing fil_space objects
(first, fil_space_detach(), then release fil_system->mutex, and finally
free the fil_space and fil_node objects).
fil_space_free_and_mutex_exit(): Renamed from fil_space_free().
Detach the tablespace from the fil_system cache, release the
fil_system->mutex, and then wait for space->n_pending_ios to reach 0,
to avoid accessing freed data in a concurrent thread.
During the wait, future calls to fil_space_acquire_for_io() will
not find this tablespace, and the count can only be decremented to 0,
at which point it is safe to free the objects.
fil_node_free_part1(), fil_node_free_part2(): Refactored from
fil_node_free().
9 years ago MDEV-34830: LSN in the future is not being treated as serious corruption
The invariant of write-ahead logging is that before any change to a
page is written to the data file, the corresponding log record must
must first have been durably written.
In crash recovery, there were some sloppy checks for this. Let us
implement accurate checks and flag an inconsistency as a hard error,
so that we can avoid further corruption of a corrupted database.
For data extraction from the corrupted database, innodb_force_recovery
can be used.
Before recovery is reading any data pages or invoking
buf_dblwr_t::recover() to recover torn pages from the
doublewrite buffer, InnoDB will have parsed the log until the
final LSN and updated log_sys.lsn to that. So, we can rely on
log_sys.lsn at all times. The doublewrite buffer recovery has been
refactored in such a way that the recv_sys.dblwr.pages may be consulted
while discovering files and their page sizes, but nothing will be
written back to data files before buf_dblwr_t::recover() is invoked.
recv_max_page_lsn, recv_lsn_checks_on: Remove.
recv_sys_t::validate_checkpoint(): Validate the write-ahead-logging
condition at the end of the recovery.
recv_dblwr_t::validate_page(): Keep track of the maximum LSN
(if we are checking a non-doublewrite copy of a page) but
do not complain LSN being in the future. The doublewrite buffer
is a special case, because it will be read early during recovery.
Besides, starting with commit 762bcb81b5bf9bbde61fed59afb26417f4ce1e86
the dblwr=true copies of pages may legitimately be "too new".
recv_dblwr_t::find_page(): Find a valid page with the smallest
FIL_PAGE_LSN that is in the valid range for recovery.
recv_dblwr_t::restore_first_page(): Replaced by find_page().
Only buf_dblwr_t::recover() will write to data files.
buf_dblwr_t::recover(): Simplify the message output. Do attempt
doublewrite recovery on user page read error. Ignore doublewrite
pages whose FIL_PAGE_LSN is outside the usable bounds. Previously,
we could wrongly recover a too new page from the doublewrite buffer.
It is unlikely that this could have lead to an actual error.
Write back all recovered pages from the doublewrite buffer here,
including for the first page of any tablespace.
buf_page_is_corrupted(): Distinguish the return values
CORRUPTED_FUTURE_LSN and CORRUPTED_OTHER.
buf_page_check_corrupt(): Return the error code DB_CORRUPTION
in case the LSN is in the future.
Datafile::read_first_page_flags(): Split from read_first_page().
Take a copy of the first page as a parameter.
recv_sys_t::free_corrupted_page(): Take the file as a parameter
and return whether a message was displayed. This avoids some duplicated
and incomplete error messages.
buf_page_t::read_complete(): Remove some redundant output and always
display the name of the corrupted file. Never return DB_FAIL;
use it only in internal error handling.
IORequest::read_complete(): Assume that buf_page_t::read_complete()
will have reported any error.
fil_space_t::set_corrupted(): Return whether this is the first time
the tablespace had been flagged as corrupted.
Datafile::validate_first_page(), fil_node_open_file_low(),
fil_node_open_file(), fil_space_t::read_page0(),
fil_node_t::read_page0(): Add a parameter for a copy of the
first page, and a parameter to indicate whether the FIL_PAGE_LSN
check should be suppressed. Before buf_dblwr_t::recover() is
invoked, we cannot validate the FIL_PAGE_LSN, but we can trust the
FSP_SPACE_FLAGS and the tablespace ID that may be present in a
potentially too new copy of a page.
Reviewed by: Debarun Banerjee
12 months ago MDEV-18644: Support full_crc32 for page_compressed
This is a follow-up task to MDEV-12026, which introduced
innodb_checksum_algorithm=full_crc32 and a simpler page format.
MDEV-12026 did not enable full_crc32 for page_compressed tables,
which we will be doing now.
This is joint work with Thirunarayanan Balathandayuthapani.
For innodb_checksum_algorithm=full_crc32 we change the
page_compressed format as follows:
FIL_PAGE_TYPE: The most significant bit will be set to indicate
page_compressed format. The least significant bits will contain
the compressed page size, rounded up to a multiple of 256 bytes.
The checksum will be stored in the last 4 bytes of the page
(whether it is the full page or a page_compressed page whose
size is determined by FIL_PAGE_TYPE), covering all preceding
bytes of the page. If encryption is used, then the page will
be encrypted between compression and computing the checksum.
For page_compressed, FIL_PAGE_LSN will not be repeated at
the end of the page.
FSP_SPACE_FLAGS (already implemented as part of MDEV-12026):
We will store the innodb_compression_algorithm that may be used
to compress pages. Previously, the choice of algorithm was written
to each compressed data page separately, and one would be unable
to know in advance which compression algorithm(s) are used.
fil_space_t::full_crc32_page_compressed_len(): Determine if the
page_compressed algorithm of the tablespace needs to know the
exact length of the compressed data. If yes, we will reserve and
write an extra byte for this right before the checksum.
buf_page_is_compressed(): Determine if a page uses page_compressed
(in any innodb_checksum_algorithm).
fil_page_decompress(): Pass also fil_space_t::flags so that the
format can be determined.
buf_page_is_zeroes(): Check if a page is full of zero bytes.
buf_page_full_crc32_is_corrupted(): Renamed from
buf_encrypted_full_crc32_page_is_corrupted(). For full_crc32,
we always simply validate the checksum to the page contents,
while the physical page size is explicitly specified by an
unencrypted part of the page header.
buf_page_full_crc32_size(): Determine the size of a full_crc32 page.
buf_dblwr_check_page_lsn(): Make this a debug-only function, because
it involves potentially costly lookups of fil_space_t.
create_table_info_t::check_table_options(),
ha_innobase::check_if_supported_inplace_alter(): Do allow the creation
of SPATIAL INDEX with full_crc32 also when page_compressed is used.
commit_cache_norebuild(): Preserve the compression algorithm when
updating the page_compression_level.
dict_tf_to_fsp_flags(): Set the flags for page compression algorithm.
FIXME: Maybe there should be a table option page_compression_algorithm
and a session variable to back it?
7 years ago MDEV-34830: LSN in the future is not being treated as serious corruption
The invariant of write-ahead logging is that before any change to a
page is written to the data file, the corresponding log record must
must first have been durably written.
In crash recovery, there were some sloppy checks for this. Let us
implement accurate checks and flag an inconsistency as a hard error,
so that we can avoid further corruption of a corrupted database.
For data extraction from the corrupted database, innodb_force_recovery
can be used.
Before recovery is reading any data pages or invoking
buf_dblwr_t::recover() to recover torn pages from the
doublewrite buffer, InnoDB will have parsed the log until the
final LSN and updated log_sys.lsn to that. So, we can rely on
log_sys.lsn at all times. The doublewrite buffer recovery has been
refactored in such a way that the recv_sys.dblwr.pages may be consulted
while discovering files and their page sizes, but nothing will be
written back to data files before buf_dblwr_t::recover() is invoked.
recv_max_page_lsn, recv_lsn_checks_on: Remove.
recv_sys_t::validate_checkpoint(): Validate the write-ahead-logging
condition at the end of the recovery.
recv_dblwr_t::validate_page(): Keep track of the maximum LSN
(if we are checking a non-doublewrite copy of a page) but
do not complain LSN being in the future. The doublewrite buffer
is a special case, because it will be read early during recovery.
Besides, starting with commit 762bcb81b5bf9bbde61fed59afb26417f4ce1e86
the dblwr=true copies of pages may legitimately be "too new".
recv_dblwr_t::find_page(): Find a valid page with the smallest
FIL_PAGE_LSN that is in the valid range for recovery.
recv_dblwr_t::restore_first_page(): Replaced by find_page().
Only buf_dblwr_t::recover() will write to data files.
buf_dblwr_t::recover(): Simplify the message output. Do attempt
doublewrite recovery on user page read error. Ignore doublewrite
pages whose FIL_PAGE_LSN is outside the usable bounds. Previously,
we could wrongly recover a too new page from the doublewrite buffer.
It is unlikely that this could have lead to an actual error.
Write back all recovered pages from the doublewrite buffer here,
including for the first page of any tablespace.
buf_page_is_corrupted(): Distinguish the return values
CORRUPTED_FUTURE_LSN and CORRUPTED_OTHER.
buf_page_check_corrupt(): Return the error code DB_CORRUPTION
in case the LSN is in the future.
Datafile::read_first_page_flags(): Split from read_first_page().
Take a copy of the first page as a parameter.
recv_sys_t::free_corrupted_page(): Take the file as a parameter
and return whether a message was displayed. This avoids some duplicated
and incomplete error messages.
buf_page_t::read_complete(): Remove some redundant output and always
display the name of the corrupted file. Never return DB_FAIL;
use it only in internal error handling.
IORequest::read_complete(): Assume that buf_page_t::read_complete()
will have reported any error.
fil_space_t::set_corrupted(): Return whether this is the first time
the tablespace had been flagged as corrupted.
Datafile::validate_first_page(), fil_node_open_file_low(),
fil_node_open_file(), fil_space_t::read_page0(),
fil_node_t::read_page0(): Add a parameter for a copy of the
first page, and a parameter to indicate whether the FIL_PAGE_LSN
check should be suppressed. Before buf_dblwr_t::recover() is
invoked, we cannot validate the FIL_PAGE_LSN, but we can trust the
FSP_SPACE_FLAGS and the tablespace ID that may be present in a
potentially too new copy of a page.
Reviewed by: Debarun Banerjee
12 months ago MDEV-18644: Support full_crc32 for page_compressed
This is a follow-up task to MDEV-12026, which introduced
innodb_checksum_algorithm=full_crc32 and a simpler page format.
MDEV-12026 did not enable full_crc32 for page_compressed tables,
which we will be doing now.
This is joint work with Thirunarayanan Balathandayuthapani.
For innodb_checksum_algorithm=full_crc32 we change the
page_compressed format as follows:
FIL_PAGE_TYPE: The most significant bit will be set to indicate
page_compressed format. The least significant bits will contain
the compressed page size, rounded up to a multiple of 256 bytes.
The checksum will be stored in the last 4 bytes of the page
(whether it is the full page or a page_compressed page whose
size is determined by FIL_PAGE_TYPE), covering all preceding
bytes of the page. If encryption is used, then the page will
be encrypted between compression and computing the checksum.
For page_compressed, FIL_PAGE_LSN will not be repeated at
the end of the page.
FSP_SPACE_FLAGS (already implemented as part of MDEV-12026):
We will store the innodb_compression_algorithm that may be used
to compress pages. Previously, the choice of algorithm was written
to each compressed data page separately, and one would be unable
to know in advance which compression algorithm(s) are used.
fil_space_t::full_crc32_page_compressed_len(): Determine if the
page_compressed algorithm of the tablespace needs to know the
exact length of the compressed data. If yes, we will reserve and
write an extra byte for this right before the checksum.
buf_page_is_compressed(): Determine if a page uses page_compressed
(in any innodb_checksum_algorithm).
fil_page_decompress(): Pass also fil_space_t::flags so that the
format can be determined.
buf_page_is_zeroes(): Check if a page is full of zero bytes.
buf_page_full_crc32_is_corrupted(): Renamed from
buf_encrypted_full_crc32_page_is_corrupted(). For full_crc32,
we always simply validate the checksum to the page contents,
while the physical page size is explicitly specified by an
unencrypted part of the page header.
buf_page_full_crc32_size(): Determine the size of a full_crc32 page.
buf_dblwr_check_page_lsn(): Make this a debug-only function, because
it involves potentially costly lookups of fil_space_t.
create_table_info_t::check_table_options(),
ha_innobase::check_if_supported_inplace_alter(): Do allow the creation
of SPATIAL INDEX with full_crc32 also when page_compressed is used.
commit_cache_norebuild(): Preserve the compression algorithm when
updating the page_compression_level.
dict_tf_to_fsp_flags(): Set the flags for page compression algorithm.
FIXME: Maybe there should be a table option page_compression_algorithm
and a session variable to back it?
7 years ago MDEV-34830: LSN in the future is not being treated as serious corruption
The invariant of write-ahead logging is that before any change to a
page is written to the data file, the corresponding log record must
must first have been durably written.
In crash recovery, there were some sloppy checks for this. Let us
implement accurate checks and flag an inconsistency as a hard error,
so that we can avoid further corruption of a corrupted database.
For data extraction from the corrupted database, innodb_force_recovery
can be used.
Before recovery is reading any data pages or invoking
buf_dblwr_t::recover() to recover torn pages from the
doublewrite buffer, InnoDB will have parsed the log until the
final LSN and updated log_sys.lsn to that. So, we can rely on
log_sys.lsn at all times. The doublewrite buffer recovery has been
refactored in such a way that the recv_sys.dblwr.pages may be consulted
while discovering files and their page sizes, but nothing will be
written back to data files before buf_dblwr_t::recover() is invoked.
recv_max_page_lsn, recv_lsn_checks_on: Remove.
recv_sys_t::validate_checkpoint(): Validate the write-ahead-logging
condition at the end of the recovery.
recv_dblwr_t::validate_page(): Keep track of the maximum LSN
(if we are checking a non-doublewrite copy of a page) but
do not complain LSN being in the future. The doublewrite buffer
is a special case, because it will be read early during recovery.
Besides, starting with commit 762bcb81b5bf9bbde61fed59afb26417f4ce1e86
the dblwr=true copies of pages may legitimately be "too new".
recv_dblwr_t::find_page(): Find a valid page with the smallest
FIL_PAGE_LSN that is in the valid range for recovery.
recv_dblwr_t::restore_first_page(): Replaced by find_page().
Only buf_dblwr_t::recover() will write to data files.
buf_dblwr_t::recover(): Simplify the message output. Do attempt
doublewrite recovery on user page read error. Ignore doublewrite
pages whose FIL_PAGE_LSN is outside the usable bounds. Previously,
we could wrongly recover a too new page from the doublewrite buffer.
It is unlikely that this could have lead to an actual error.
Write back all recovered pages from the doublewrite buffer here,
including for the first page of any tablespace.
buf_page_is_corrupted(): Distinguish the return values
CORRUPTED_FUTURE_LSN and CORRUPTED_OTHER.
buf_page_check_corrupt(): Return the error code DB_CORRUPTION
in case the LSN is in the future.
Datafile::read_first_page_flags(): Split from read_first_page().
Take a copy of the first page as a parameter.
recv_sys_t::free_corrupted_page(): Take the file as a parameter
and return whether a message was displayed. This avoids some duplicated
and incomplete error messages.
buf_page_t::read_complete(): Remove some redundant output and always
display the name of the corrupted file. Never return DB_FAIL;
use it only in internal error handling.
IORequest::read_complete(): Assume that buf_page_t::read_complete()
will have reported any error.
fil_space_t::set_corrupted(): Return whether this is the first time
the tablespace had been flagged as corrupted.
Datafile::validate_first_page(), fil_node_open_file_low(),
fil_node_open_file(), fil_space_t::read_page0(),
fil_node_t::read_page0(): Add a parameter for a copy of the
first page, and a parameter to indicate whether the FIL_PAGE_LSN
check should be suppressed. Before buf_dblwr_t::recover() is
invoked, we cannot validate the FIL_PAGE_LSN, but we can trust the
FSP_SPACE_FLAGS and the tablespace ID that may be present in a
potentially too new copy of a page.
Reviewed by: Debarun Banerjee
12 months ago MDEV-12026: Implement innodb_checksum_algorithm=full_crc32
MariaDB data-at-rest encryption (innodb_encrypt_tables)
had repurposed the same unused data field that was repurposed
in MySQL 5.7 (and MariaDB 10.2) for the Split Sequence Number (SSN)
field of SPATIAL INDEX. Because of this, MariaDB was unable to
support encryption on SPATIAL INDEX pages.
Furthermore, InnoDB page checksums skipped some bytes, and there
are multiple variations and checksum algorithms. By default,
InnoDB accepts all variations of all algorithms that ever existed.
This unnecessarily weakens the page checksums.
We hereby introduce two more innodb_checksum_algorithm variants
(full_crc32, strict_full_crc32) that are special in a way:
When either setting is active, newly created data files will
carry a flag (fil_space_t::full_crc32()) that indicates that
all pages of the file will use a full CRC-32C checksum over the
entire page contents (excluding the bytes where the checksum
is stored, at the very end of the page). Such files will always
use that checksum, no matter what the parameter
innodb_checksum_algorithm is assigned to.
For old files, the old checksum algorithms will continue to be
used. The value strict_full_crc32 will be equivalent to strict_crc32
and the value full_crc32 will be equivalent to crc32.
ROW_FORMAT=COMPRESSED tables will only use the old format.
These tables do not support new features, such as larger
innodb_page_size or instant ADD/DROP COLUMN. They may be
deprecated in the future. We do not want an unnecessary
file format change for them.
The new full_crc32() format also cleans up the MariaDB tablespace
flags. We will reserve flags to store the page_compressed
compression algorithm, and to store the compressed payload length,
so that checksum can be computed over the compressed (and
possibly encrypted) stream and can be validated without
decrypting or decompressing the page.
In the full_crc32 format, there no longer are separate before-encryption
and after-encryption checksums for pages. The single checksum is
computed on the page contents that is written to the file.
We do not make the new algorithm the default for two reasons.
First, MariaDB 10.4.2 was a beta release, and the default values
of parameters should not change after beta. Second, we did not
yet implement the full_crc32 format for page_compressed pages.
This will be fixed in MDEV-18644.
This is joint work with Marko Mäkelä.
7 years ago MDEV-34830: LSN in the future is not being treated as serious corruption
The invariant of write-ahead logging is that before any change to a
page is written to the data file, the corresponding log record must
must first have been durably written.
In crash recovery, there were some sloppy checks for this. Let us
implement accurate checks and flag an inconsistency as a hard error,
so that we can avoid further corruption of a corrupted database.
For data extraction from the corrupted database, innodb_force_recovery
can be used.
Before recovery is reading any data pages or invoking
buf_dblwr_t::recover() to recover torn pages from the
doublewrite buffer, InnoDB will have parsed the log until the
final LSN and updated log_sys.lsn to that. So, we can rely on
log_sys.lsn at all times. The doublewrite buffer recovery has been
refactored in such a way that the recv_sys.dblwr.pages may be consulted
while discovering files and their page sizes, but nothing will be
written back to data files before buf_dblwr_t::recover() is invoked.
recv_max_page_lsn, recv_lsn_checks_on: Remove.
recv_sys_t::validate_checkpoint(): Validate the write-ahead-logging
condition at the end of the recovery.
recv_dblwr_t::validate_page(): Keep track of the maximum LSN
(if we are checking a non-doublewrite copy of a page) but
do not complain LSN being in the future. The doublewrite buffer
is a special case, because it will be read early during recovery.
Besides, starting with commit 762bcb81b5bf9bbde61fed59afb26417f4ce1e86
the dblwr=true copies of pages may legitimately be "too new".
recv_dblwr_t::find_page(): Find a valid page with the smallest
FIL_PAGE_LSN that is in the valid range for recovery.
recv_dblwr_t::restore_first_page(): Replaced by find_page().
Only buf_dblwr_t::recover() will write to data files.
buf_dblwr_t::recover(): Simplify the message output. Do attempt
doublewrite recovery on user page read error. Ignore doublewrite
pages whose FIL_PAGE_LSN is outside the usable bounds. Previously,
we could wrongly recover a too new page from the doublewrite buffer.
It is unlikely that this could have lead to an actual error.
Write back all recovered pages from the doublewrite buffer here,
including for the first page of any tablespace.
buf_page_is_corrupted(): Distinguish the return values
CORRUPTED_FUTURE_LSN and CORRUPTED_OTHER.
buf_page_check_corrupt(): Return the error code DB_CORRUPTION
in case the LSN is in the future.
Datafile::read_first_page_flags(): Split from read_first_page().
Take a copy of the first page as a parameter.
recv_sys_t::free_corrupted_page(): Take the file as a parameter
and return whether a message was displayed. This avoids some duplicated
and incomplete error messages.
buf_page_t::read_complete(): Remove some redundant output and always
display the name of the corrupted file. Never return DB_FAIL;
use it only in internal error handling.
IORequest::read_complete(): Assume that buf_page_t::read_complete()
will have reported any error.
fil_space_t::set_corrupted(): Return whether this is the first time
the tablespace had been flagged as corrupted.
Datafile::validate_first_page(), fil_node_open_file_low(),
fil_node_open_file(), fil_space_t::read_page0(),
fil_node_t::read_page0(): Add a parameter for a copy of the
first page, and a parameter to indicate whether the FIL_PAGE_LSN
check should be suppressed. Before buf_dblwr_t::recover() is
invoked, we cannot validate the FIL_PAGE_LSN, but we can trust the
FSP_SPACE_FLAGS and the tablespace ID that may be present in a
potentially too new copy of a page.
Reviewed by: Debarun Banerjee
12 months ago MDEV-12026: Implement innodb_checksum_algorithm=full_crc32
MariaDB data-at-rest encryption (innodb_encrypt_tables)
had repurposed the same unused data field that was repurposed
in MySQL 5.7 (and MariaDB 10.2) for the Split Sequence Number (SSN)
field of SPATIAL INDEX. Because of this, MariaDB was unable to
support encryption on SPATIAL INDEX pages.
Furthermore, InnoDB page checksums skipped some bytes, and there
are multiple variations and checksum algorithms. By default,
InnoDB accepts all variations of all algorithms that ever existed.
This unnecessarily weakens the page checksums.
We hereby introduce two more innodb_checksum_algorithm variants
(full_crc32, strict_full_crc32) that are special in a way:
When either setting is active, newly created data files will
carry a flag (fil_space_t::full_crc32()) that indicates that
all pages of the file will use a full CRC-32C checksum over the
entire page contents (excluding the bytes where the checksum
is stored, at the very end of the page). Such files will always
use that checksum, no matter what the parameter
innodb_checksum_algorithm is assigned to.
For old files, the old checksum algorithms will continue to be
used. The value strict_full_crc32 will be equivalent to strict_crc32
and the value full_crc32 will be equivalent to crc32.
ROW_FORMAT=COMPRESSED tables will only use the old format.
These tables do not support new features, such as larger
innodb_page_size or instant ADD/DROP COLUMN. They may be
deprecated in the future. We do not want an unnecessary
file format change for them.
The new full_crc32() format also cleans up the MariaDB tablespace
flags. We will reserve flags to store the page_compressed
compression algorithm, and to store the compressed payload length,
so that checksum can be computed over the compressed (and
possibly encrypted) stream and can be validated without
decrypting or decompressing the page.
In the full_crc32 format, there no longer are separate before-encryption
and after-encryption checksums for pages. The single checksum is
computed on the page contents that is written to the file.
We do not make the new algorithm the default for two reasons.
First, MariaDB 10.4.2 was a beta release, and the default values
of parameters should not change after beta. Second, we did not
yet implement the full_crc32 format for page_compressed pages.
This will be fixed in MDEV-18644.
This is joint work with Marko Mäkelä.
7 years ago MDEV-18644: Support full_crc32 for page_compressed
This is a follow-up task to MDEV-12026, which introduced
innodb_checksum_algorithm=full_crc32 and a simpler page format.
MDEV-12026 did not enable full_crc32 for page_compressed tables,
which we will be doing now.
This is joint work with Thirunarayanan Balathandayuthapani.
For innodb_checksum_algorithm=full_crc32 we change the
page_compressed format as follows:
FIL_PAGE_TYPE: The most significant bit will be set to indicate
page_compressed format. The least significant bits will contain
the compressed page size, rounded up to a multiple of 256 bytes.
The checksum will be stored in the last 4 bytes of the page
(whether it is the full page or a page_compressed page whose
size is determined by FIL_PAGE_TYPE), covering all preceding
bytes of the page. If encryption is used, then the page will
be encrypted between compression and computing the checksum.
For page_compressed, FIL_PAGE_LSN will not be repeated at
the end of the page.
FSP_SPACE_FLAGS (already implemented as part of MDEV-12026):
We will store the innodb_compression_algorithm that may be used
to compress pages. Previously, the choice of algorithm was written
to each compressed data page separately, and one would be unable
to know in advance which compression algorithm(s) are used.
fil_space_t::full_crc32_page_compressed_len(): Determine if the
page_compressed algorithm of the tablespace needs to know the
exact length of the compressed data. If yes, we will reserve and
write an extra byte for this right before the checksum.
buf_page_is_compressed(): Determine if a page uses page_compressed
(in any innodb_checksum_algorithm).
fil_page_decompress(): Pass also fil_space_t::flags so that the
format can be determined.
buf_page_is_zeroes(): Check if a page is full of zero bytes.
buf_page_full_crc32_is_corrupted(): Renamed from
buf_encrypted_full_crc32_page_is_corrupted(). For full_crc32,
we always simply validate the checksum to the page contents,
while the physical page size is explicitly specified by an
unencrypted part of the page header.
buf_page_full_crc32_size(): Determine the size of a full_crc32 page.
buf_dblwr_check_page_lsn(): Make this a debug-only function, because
it involves potentially costly lookups of fil_space_t.
create_table_info_t::check_table_options(),
ha_innobase::check_if_supported_inplace_alter(): Do allow the creation
of SPATIAL INDEX with full_crc32 also when page_compressed is used.
commit_cache_norebuild(): Preserve the compression algorithm when
updating the page_compression_level.
dict_tf_to_fsp_flags(): Set the flags for page compression algorithm.
FIXME: Maybe there should be a table option page_compression_algorithm
and a session variable to back it?
7 years ago MDEV-34830: LSN in the future is not being treated as serious corruption
The invariant of write-ahead logging is that before any change to a
page is written to the data file, the corresponding log record must
must first have been durably written.
In crash recovery, there were some sloppy checks for this. Let us
implement accurate checks and flag an inconsistency as a hard error,
so that we can avoid further corruption of a corrupted database.
For data extraction from the corrupted database, innodb_force_recovery
can be used.
Before recovery is reading any data pages or invoking
buf_dblwr_t::recover() to recover torn pages from the
doublewrite buffer, InnoDB will have parsed the log until the
final LSN and updated log_sys.lsn to that. So, we can rely on
log_sys.lsn at all times. The doublewrite buffer recovery has been
refactored in such a way that the recv_sys.dblwr.pages may be consulted
while discovering files and their page sizes, but nothing will be
written back to data files before buf_dblwr_t::recover() is invoked.
recv_max_page_lsn, recv_lsn_checks_on: Remove.
recv_sys_t::validate_checkpoint(): Validate the write-ahead-logging
condition at the end of the recovery.
recv_dblwr_t::validate_page(): Keep track of the maximum LSN
(if we are checking a non-doublewrite copy of a page) but
do not complain LSN being in the future. The doublewrite buffer
is a special case, because it will be read early during recovery.
Besides, starting with commit 762bcb81b5bf9bbde61fed59afb26417f4ce1e86
the dblwr=true copies of pages may legitimately be "too new".
recv_dblwr_t::find_page(): Find a valid page with the smallest
FIL_PAGE_LSN that is in the valid range for recovery.
recv_dblwr_t::restore_first_page(): Replaced by find_page().
Only buf_dblwr_t::recover() will write to data files.
buf_dblwr_t::recover(): Simplify the message output. Do attempt
doublewrite recovery on user page read error. Ignore doublewrite
pages whose FIL_PAGE_LSN is outside the usable bounds. Previously,
we could wrongly recover a too new page from the doublewrite buffer.
It is unlikely that this could have lead to an actual error.
Write back all recovered pages from the doublewrite buffer here,
including for the first page of any tablespace.
buf_page_is_corrupted(): Distinguish the return values
CORRUPTED_FUTURE_LSN and CORRUPTED_OTHER.
buf_page_check_corrupt(): Return the error code DB_CORRUPTION
in case the LSN is in the future.
Datafile::read_first_page_flags(): Split from read_first_page().
Take a copy of the first page as a parameter.
recv_sys_t::free_corrupted_page(): Take the file as a parameter
and return whether a message was displayed. This avoids some duplicated
and incomplete error messages.
buf_page_t::read_complete(): Remove some redundant output and always
display the name of the corrupted file. Never return DB_FAIL;
use it only in internal error handling.
IORequest::read_complete(): Assume that buf_page_t::read_complete()
will have reported any error.
fil_space_t::set_corrupted(): Return whether this is the first time
the tablespace had been flagged as corrupted.
Datafile::validate_first_page(), fil_node_open_file_low(),
fil_node_open_file(), fil_space_t::read_page0(),
fil_node_t::read_page0(): Add a parameter for a copy of the
first page, and a parameter to indicate whether the FIL_PAGE_LSN
check should be suppressed. Before buf_dblwr_t::recover() is
invoked, we cannot validate the FIL_PAGE_LSN, but we can trust the
FSP_SPACE_FLAGS and the tablespace ID that may be present in a
potentially too new copy of a page.
Reviewed by: Debarun Banerjee
12 months ago MDEV-12026: Implement innodb_checksum_algorithm=full_crc32
MariaDB data-at-rest encryption (innodb_encrypt_tables)
had repurposed the same unused data field that was repurposed
in MySQL 5.7 (and MariaDB 10.2) for the Split Sequence Number (SSN)
field of SPATIAL INDEX. Because of this, MariaDB was unable to
support encryption on SPATIAL INDEX pages.
Furthermore, InnoDB page checksums skipped some bytes, and there
are multiple variations and checksum algorithms. By default,
InnoDB accepts all variations of all algorithms that ever existed.
This unnecessarily weakens the page checksums.
We hereby introduce two more innodb_checksum_algorithm variants
(full_crc32, strict_full_crc32) that are special in a way:
When either setting is active, newly created data files will
carry a flag (fil_space_t::full_crc32()) that indicates that
all pages of the file will use a full CRC-32C checksum over the
entire page contents (excluding the bytes where the checksum
is stored, at the very end of the page). Such files will always
use that checksum, no matter what the parameter
innodb_checksum_algorithm is assigned to.
For old files, the old checksum algorithms will continue to be
used. The value strict_full_crc32 will be equivalent to strict_crc32
and the value full_crc32 will be equivalent to crc32.
ROW_FORMAT=COMPRESSED tables will only use the old format.
These tables do not support new features, such as larger
innodb_page_size or instant ADD/DROP COLUMN. They may be
deprecated in the future. We do not want an unnecessary
file format change for them.
The new full_crc32() format also cleans up the MariaDB tablespace
flags. We will reserve flags to store the page_compressed
compression algorithm, and to store the compressed payload length,
so that checksum can be computed over the compressed (and
possibly encrypted) stream and can be validated without
decrypting or decompressing the page.
In the full_crc32 format, there no longer are separate before-encryption
and after-encryption checksums for pages. The single checksum is
computed on the page contents that is written to the file.
We do not make the new algorithm the default for two reasons.
First, MariaDB 10.4.2 was a beta release, and the default values
of parameters should not change after beta. Second, we did not
yet implement the full_crc32 format for page_compressed pages.
This will be fixed in MDEV-18644.
This is joint work with Marko Mäkelä.
7 years ago MDEV-18644: Support full_crc32 for page_compressed
This is a follow-up task to MDEV-12026, which introduced
innodb_checksum_algorithm=full_crc32 and a simpler page format.
MDEV-12026 did not enable full_crc32 for page_compressed tables,
which we will be doing now.
This is joint work with Thirunarayanan Balathandayuthapani.
For innodb_checksum_algorithm=full_crc32 we change the
page_compressed format as follows:
FIL_PAGE_TYPE: The most significant bit will be set to indicate
page_compressed format. The least significant bits will contain
the compressed page size, rounded up to a multiple of 256 bytes.
The checksum will be stored in the last 4 bytes of the page
(whether it is the full page or a page_compressed page whose
size is determined by FIL_PAGE_TYPE), covering all preceding
bytes of the page. If encryption is used, then the page will
be encrypted between compression and computing the checksum.
For page_compressed, FIL_PAGE_LSN will not be repeated at
the end of the page.
FSP_SPACE_FLAGS (already implemented as part of MDEV-12026):
We will store the innodb_compression_algorithm that may be used
to compress pages. Previously, the choice of algorithm was written
to each compressed data page separately, and one would be unable
to know in advance which compression algorithm(s) are used.
fil_space_t::full_crc32_page_compressed_len(): Determine if the
page_compressed algorithm of the tablespace needs to know the
exact length of the compressed data. If yes, we will reserve and
write an extra byte for this right before the checksum.
buf_page_is_compressed(): Determine if a page uses page_compressed
(in any innodb_checksum_algorithm).
fil_page_decompress(): Pass also fil_space_t::flags so that the
format can be determined.
buf_page_is_zeroes(): Check if a page is full of zero bytes.
buf_page_full_crc32_is_corrupted(): Renamed from
buf_encrypted_full_crc32_page_is_corrupted(). For full_crc32,
we always simply validate the checksum to the page contents,
while the physical page size is explicitly specified by an
unencrypted part of the page header.
buf_page_full_crc32_size(): Determine the size of a full_crc32 page.
buf_dblwr_check_page_lsn(): Make this a debug-only function, because
it involves potentially costly lookups of fil_space_t.
create_table_info_t::check_table_options(),
ha_innobase::check_if_supported_inplace_alter(): Do allow the creation
of SPATIAL INDEX with full_crc32 also when page_compressed is used.
commit_cache_norebuild(): Preserve the compression algorithm when
updating the page_compression_level.
dict_tf_to_fsp_flags(): Set the flags for page compression algorithm.
FIXME: Maybe there should be a table option page_compression_algorithm
and a session variable to back it?
7 years ago MDEV-34830: LSN in the future is not being treated as serious corruption
The invariant of write-ahead logging is that before any change to a
page is written to the data file, the corresponding log record must
must first have been durably written.
In crash recovery, there were some sloppy checks for this. Let us
implement accurate checks and flag an inconsistency as a hard error,
so that we can avoid further corruption of a corrupted database.
For data extraction from the corrupted database, innodb_force_recovery
can be used.
Before recovery is reading any data pages or invoking
buf_dblwr_t::recover() to recover torn pages from the
doublewrite buffer, InnoDB will have parsed the log until the
final LSN and updated log_sys.lsn to that. So, we can rely on
log_sys.lsn at all times. The doublewrite buffer recovery has been
refactored in such a way that the recv_sys.dblwr.pages may be consulted
while discovering files and their page sizes, but nothing will be
written back to data files before buf_dblwr_t::recover() is invoked.
recv_max_page_lsn, recv_lsn_checks_on: Remove.
recv_sys_t::validate_checkpoint(): Validate the write-ahead-logging
condition at the end of the recovery.
recv_dblwr_t::validate_page(): Keep track of the maximum LSN
(if we are checking a non-doublewrite copy of a page) but
do not complain LSN being in the future. The doublewrite buffer
is a special case, because it will be read early during recovery.
Besides, starting with commit 762bcb81b5bf9bbde61fed59afb26417f4ce1e86
the dblwr=true copies of pages may legitimately be "too new".
recv_dblwr_t::find_page(): Find a valid page with the smallest
FIL_PAGE_LSN that is in the valid range for recovery.
recv_dblwr_t::restore_first_page(): Replaced by find_page().
Only buf_dblwr_t::recover() will write to data files.
buf_dblwr_t::recover(): Simplify the message output. Do attempt
doublewrite recovery on user page read error. Ignore doublewrite
pages whose FIL_PAGE_LSN is outside the usable bounds. Previously,
we could wrongly recover a too new page from the doublewrite buffer.
It is unlikely that this could have lead to an actual error.
Write back all recovered pages from the doublewrite buffer here,
including for the first page of any tablespace.
buf_page_is_corrupted(): Distinguish the return values
CORRUPTED_FUTURE_LSN and CORRUPTED_OTHER.
buf_page_check_corrupt(): Return the error code DB_CORRUPTION
in case the LSN is in the future.
Datafile::read_first_page_flags(): Split from read_first_page().
Take a copy of the first page as a parameter.
recv_sys_t::free_corrupted_page(): Take the file as a parameter
and return whether a message was displayed. This avoids some duplicated
and incomplete error messages.
buf_page_t::read_complete(): Remove some redundant output and always
display the name of the corrupted file. Never return DB_FAIL;
use it only in internal error handling.
IORequest::read_complete(): Assume that buf_page_t::read_complete()
will have reported any error.
fil_space_t::set_corrupted(): Return whether this is the first time
the tablespace had been flagged as corrupted.
Datafile::validate_first_page(), fil_node_open_file_low(),
fil_node_open_file(), fil_space_t::read_page0(),
fil_node_t::read_page0(): Add a parameter for a copy of the
first page, and a parameter to indicate whether the FIL_PAGE_LSN
check should be suppressed. Before buf_dblwr_t::recover() is
invoked, we cannot validate the FIL_PAGE_LSN, but we can trust the
FSP_SPACE_FLAGS and the tablespace ID that may be present in a
potentially too new copy of a page.
Reviewed by: Debarun Banerjee
12 months ago MDEV-12026: Implement innodb_checksum_algorithm=full_crc32
MariaDB data-at-rest encryption (innodb_encrypt_tables)
had repurposed the same unused data field that was repurposed
in MySQL 5.7 (and MariaDB 10.2) for the Split Sequence Number (SSN)
field of SPATIAL INDEX. Because of this, MariaDB was unable to
support encryption on SPATIAL INDEX pages.
Furthermore, InnoDB page checksums skipped some bytes, and there
are multiple variations and checksum algorithms. By default,
InnoDB accepts all variations of all algorithms that ever existed.
This unnecessarily weakens the page checksums.
We hereby introduce two more innodb_checksum_algorithm variants
(full_crc32, strict_full_crc32) that are special in a way:
When either setting is active, newly created data files will
carry a flag (fil_space_t::full_crc32()) that indicates that
all pages of the file will use a full CRC-32C checksum over the
entire page contents (excluding the bytes where the checksum
is stored, at the very end of the page). Such files will always
use that checksum, no matter what the parameter
innodb_checksum_algorithm is assigned to.
For old files, the old checksum algorithms will continue to be
used. The value strict_full_crc32 will be equivalent to strict_crc32
and the value full_crc32 will be equivalent to crc32.
ROW_FORMAT=COMPRESSED tables will only use the old format.
These tables do not support new features, such as larger
innodb_page_size or instant ADD/DROP COLUMN. They may be
deprecated in the future. We do not want an unnecessary
file format change for them.
The new full_crc32() format also cleans up the MariaDB tablespace
flags. We will reserve flags to store the page_compressed
compression algorithm, and to store the compressed payload length,
so that checksum can be computed over the compressed (and
possibly encrypted) stream and can be validated without
decrypting or decompressing the page.
In the full_crc32 format, there no longer are separate before-encryption
and after-encryption checksums for pages. The single checksum is
computed on the page contents that is written to the file.
We do not make the new algorithm the default for two reasons.
First, MariaDB 10.4.2 was a beta release, and the default values
of parameters should not change after beta. Second, we did not
yet implement the full_crc32 format for page_compressed pages.
This will be fixed in MDEV-18644.
This is joint work with Marko Mäkelä.
7 years ago MDEV-34830: LSN in the future is not being treated as serious corruption
The invariant of write-ahead logging is that before any change to a
page is written to the data file, the corresponding log record must
must first have been durably written.
In crash recovery, there were some sloppy checks for this. Let us
implement accurate checks and flag an inconsistency as a hard error,
so that we can avoid further corruption of a corrupted database.
For data extraction from the corrupted database, innodb_force_recovery
can be used.
Before recovery is reading any data pages or invoking
buf_dblwr_t::recover() to recover torn pages from the
doublewrite buffer, InnoDB will have parsed the log until the
final LSN and updated log_sys.lsn to that. So, we can rely on
log_sys.lsn at all times. The doublewrite buffer recovery has been
refactored in such a way that the recv_sys.dblwr.pages may be consulted
while discovering files and their page sizes, but nothing will be
written back to data files before buf_dblwr_t::recover() is invoked.
recv_max_page_lsn, recv_lsn_checks_on: Remove.
recv_sys_t::validate_checkpoint(): Validate the write-ahead-logging
condition at the end of the recovery.
recv_dblwr_t::validate_page(): Keep track of the maximum LSN
(if we are checking a non-doublewrite copy of a page) but
do not complain LSN being in the future. The doublewrite buffer
is a special case, because it will be read early during recovery.
Besides, starting with commit 762bcb81b5bf9bbde61fed59afb26417f4ce1e86
the dblwr=true copies of pages may legitimately be "too new".
recv_dblwr_t::find_page(): Find a valid page with the smallest
FIL_PAGE_LSN that is in the valid range for recovery.
recv_dblwr_t::restore_first_page(): Replaced by find_page().
Only buf_dblwr_t::recover() will write to data files.
buf_dblwr_t::recover(): Simplify the message output. Do attempt
doublewrite recovery on user page read error. Ignore doublewrite
pages whose FIL_PAGE_LSN is outside the usable bounds. Previously,
we could wrongly recover a too new page from the doublewrite buffer.
It is unlikely that this could have lead to an actual error.
Write back all recovered pages from the doublewrite buffer here,
including for the first page of any tablespace.
buf_page_is_corrupted(): Distinguish the return values
CORRUPTED_FUTURE_LSN and CORRUPTED_OTHER.
buf_page_check_corrupt(): Return the error code DB_CORRUPTION
in case the LSN is in the future.
Datafile::read_first_page_flags(): Split from read_first_page().
Take a copy of the first page as a parameter.
recv_sys_t::free_corrupted_page(): Take the file as a parameter
and return whether a message was displayed. This avoids some duplicated
and incomplete error messages.
buf_page_t::read_complete(): Remove some redundant output and always
display the name of the corrupted file. Never return DB_FAIL;
use it only in internal error handling.
IORequest::read_complete(): Assume that buf_page_t::read_complete()
will have reported any error.
fil_space_t::set_corrupted(): Return whether this is the first time
the tablespace had been flagged as corrupted.
Datafile::validate_first_page(), fil_node_open_file_low(),
fil_node_open_file(), fil_space_t::read_page0(),
fil_node_t::read_page0(): Add a parameter for a copy of the
first page, and a parameter to indicate whether the FIL_PAGE_LSN
check should be suppressed. Before buf_dblwr_t::recover() is
invoked, we cannot validate the FIL_PAGE_LSN, but we can trust the
FSP_SPACE_FLAGS and the tablespace ID that may be present in a
potentially too new copy of a page.
Reviewed by: Debarun Banerjee
12 months ago MDEV-12026: Implement innodb_checksum_algorithm=full_crc32
MariaDB data-at-rest encryption (innodb_encrypt_tables)
had repurposed the same unused data field that was repurposed
in MySQL 5.7 (and MariaDB 10.2) for the Split Sequence Number (SSN)
field of SPATIAL INDEX. Because of this, MariaDB was unable to
support encryption on SPATIAL INDEX pages.
Furthermore, InnoDB page checksums skipped some bytes, and there
are multiple variations and checksum algorithms. By default,
InnoDB accepts all variations of all algorithms that ever existed.
This unnecessarily weakens the page checksums.
We hereby introduce two more innodb_checksum_algorithm variants
(full_crc32, strict_full_crc32) that are special in a way:
When either setting is active, newly created data files will
carry a flag (fil_space_t::full_crc32()) that indicates that
all pages of the file will use a full CRC-32C checksum over the
entire page contents (excluding the bytes where the checksum
is stored, at the very end of the page). Such files will always
use that checksum, no matter what the parameter
innodb_checksum_algorithm is assigned to.
For old files, the old checksum algorithms will continue to be
used. The value strict_full_crc32 will be equivalent to strict_crc32
and the value full_crc32 will be equivalent to crc32.
ROW_FORMAT=COMPRESSED tables will only use the old format.
These tables do not support new features, such as larger
innodb_page_size or instant ADD/DROP COLUMN. They may be
deprecated in the future. We do not want an unnecessary
file format change for them.
The new full_crc32() format also cleans up the MariaDB tablespace
flags. We will reserve flags to store the page_compressed
compression algorithm, and to store the compressed payload length,
so that checksum can be computed over the compressed (and
possibly encrypted) stream and can be validated without
decrypting or decompressing the page.
In the full_crc32 format, there no longer are separate before-encryption
and after-encryption checksums for pages. The single checksum is
computed on the page contents that is written to the file.
We do not make the new algorithm the default for two reasons.
First, MariaDB 10.4.2 was a beta release, and the default values
of parameters should not change after beta. Second, we did not
yet implement the full_crc32 format for page_compressed pages.
This will be fixed in MDEV-18644.
This is joint work with Marko Mäkelä.
7 years ago  MDEV-11759: Encryption code in MariaDB 10.1/10.2 causes
compatibility problems
Pages that are encrypted contain post encryption checksum on
different location that normal checksum fields. Therefore,
we should before decryption check this checksum to avoid
unencrypting corrupted pages. After decryption we can use
traditional checksum check to detect if page is corrupted
or unencryption was done using incorrect key.
Pages that are page compressed do not contain any checksum,
here we need to fist unencrypt, decompress and finally
use tradional checksum check to detect page corruption
or that we used incorrect key in unencryption.
buf0buf.cc: buf_page_is_corrupted() mofified so that
compressed pages are skipped.
buf0buf.h, buf_block_init(), buf_page_init_low():
removed unnecessary page_encrypted, page_compressed,
stored_checksum, valculated_checksum fields from
buf_page_t
buf_page_get_gen(): use new buf_page_check_corrupt() function
to detect corrupted pages.
buf_page_check_corrupt(): If page was not yet decrypted
check if post encryption checksum still matches.
If page is not anymore encrypted, use buf_page_is_corrupted()
traditional checksum method.
If page is detected as corrupted and it is not encrypted
we print corruption message to error log.
If page is still encrypted or it was encrypted and now
corrupted, we will print message that page is
encrypted to error log.
buf_page_io_complete(): use new buf_page_check_corrupt()
function to detect corrupted pages.
buf_page_decrypt_after_read(): Verify post encryption
checksum before tring to decrypt.
fil0crypt.cc: fil_encrypt_buf() verify post encryption
checksum and ind fil_space_decrypt() return true
if we really decrypted the page.
fil_space_verify_crypt_checksum(): rewrite to use
the method used when calculating post encryption
checksum. We also check if post encryption checksum
matches that traditional checksum check does not
match.
fil0fil.ic: Add missed page type encrypted and page
compressed to fil_get_page_type_name()
Note that this change does not yet fix innochecksum tool,
that will be done in separate MDEV.
Fix test failures caused by buf page corruption injection.
9 years ago MDEV-11759: Encryption code in MariaDB 10.1/10.2 causes
compatibility problems
Pages that are encrypted contain post encryption checksum on
different location that normal checksum fields. Therefore,
we should before decryption check this checksum to avoid
unencrypting corrupted pages. After decryption we can use
traditional checksum check to detect if page is corrupted
or unencryption was done using incorrect key.
Pages that are page compressed do not contain any checksum,
here we need to fist unencrypt, decompress and finally
use tradional checksum check to detect page corruption
or that we used incorrect key in unencryption.
buf0buf.cc: buf_page_is_corrupted() mofified so that
compressed pages are skipped.
buf0buf.h, buf_block_init(), buf_page_init_low():
removed unnecessary page_encrypted, page_compressed,
stored_checksum, valculated_checksum fields from
buf_page_t
buf_page_get_gen(): use new buf_page_check_corrupt() function
to detect corrupted pages.
buf_page_check_corrupt(): If page was not yet decrypted
check if post encryption checksum still matches.
If page is not anymore encrypted, use buf_page_is_corrupted()
traditional checksum method.
If page is detected as corrupted and it is not encrypted
we print corruption message to error log.
If page is still encrypted or it was encrypted and now
corrupted, we will print message that page is
encrypted to error log.
buf_page_io_complete(): use new buf_page_check_corrupt()
function to detect corrupted pages.
buf_page_decrypt_after_read(): Verify post encryption
checksum before tring to decrypt.
fil0crypt.cc: fil_encrypt_buf() verify post encryption
checksum and ind fil_space_decrypt() return true
if we really decrypted the page.
fil_space_verify_crypt_checksum(): rewrite to use
the method used when calculating post encryption
checksum. We also check if post encryption checksum
matches that traditional checksum check does not
match.
fil0fil.ic: Add missed page type encrypted and page
compressed to fil_get_page_type_name()
Note that this change does not yet fix innochecksum tool,
that will be done in separate MDEV.
Fix test failures caused by buf page corruption injection.
9 years ago MDEV-11759: Encryption code in MariaDB 10.1/10.2 causes
compatibility problems
Pages that are encrypted contain post encryption checksum on
different location that normal checksum fields. Therefore,
we should before decryption check this checksum to avoid
unencrypting corrupted pages. After decryption we can use
traditional checksum check to detect if page is corrupted
or unencryption was done using incorrect key.
Pages that are page compressed do not contain any checksum,
here we need to fist unencrypt, decompress and finally
use tradional checksum check to detect page corruption
or that we used incorrect key in unencryption.
buf0buf.cc: buf_page_is_corrupted() mofified so that
compressed pages are skipped.
buf0buf.h, buf_block_init(), buf_page_init_low():
removed unnecessary page_encrypted, page_compressed,
stored_checksum, valculated_checksum fields from
buf_page_t
buf_page_get_gen(): use new buf_page_check_corrupt() function
to detect corrupted pages.
buf_page_check_corrupt(): If page was not yet decrypted
check if post encryption checksum still matches.
If page is not anymore encrypted, use buf_page_is_corrupted()
traditional checksum method.
If page is detected as corrupted and it is not encrypted
we print corruption message to error log.
If page is still encrypted or it was encrypted and now
corrupted, we will print message that page is
encrypted to error log.
buf_page_io_complete(): use new buf_page_check_corrupt()
function to detect corrupted pages.
buf_page_decrypt_after_read(): Verify post encryption
checksum before tring to decrypt.
fil0crypt.cc: fil_encrypt_buf() verify post encryption
checksum and ind fil_space_decrypt() return true
if we really decrypted the page.
fil_space_verify_crypt_checksum(): rewrite to use
the method used when calculating post encryption
checksum. We also check if post encryption checksum
matches that traditional checksum check does not
match.
fil0fil.ic: Add missed page type encrypted and page
compressed to fil_get_page_type_name()
Note that this change does not yet fix innochecksum tool,
that will be done in separate MDEV.
Fix test failures caused by buf page corruption injection.
9 years ago MDEV-12026: Implement innodb_checksum_algorithm=full_crc32
MariaDB data-at-rest encryption (innodb_encrypt_tables)
had repurposed the same unused data field that was repurposed
in MySQL 5.7 (and MariaDB 10.2) for the Split Sequence Number (SSN)
field of SPATIAL INDEX. Because of this, MariaDB was unable to
support encryption on SPATIAL INDEX pages.
Furthermore, InnoDB page checksums skipped some bytes, and there
are multiple variations and checksum algorithms. By default,
InnoDB accepts all variations of all algorithms that ever existed.
This unnecessarily weakens the page checksums.
We hereby introduce two more innodb_checksum_algorithm variants
(full_crc32, strict_full_crc32) that are special in a way:
When either setting is active, newly created data files will
carry a flag (fil_space_t::full_crc32()) that indicates that
all pages of the file will use a full CRC-32C checksum over the
entire page contents (excluding the bytes where the checksum
is stored, at the very end of the page). Such files will always
use that checksum, no matter what the parameter
innodb_checksum_algorithm is assigned to.
For old files, the old checksum algorithms will continue to be
used. The value strict_full_crc32 will be equivalent to strict_crc32
and the value full_crc32 will be equivalent to crc32.
ROW_FORMAT=COMPRESSED tables will only use the old format.
These tables do not support new features, such as larger
innodb_page_size or instant ADD/DROP COLUMN. They may be
deprecated in the future. We do not want an unnecessary
file format change for them.
The new full_crc32() format also cleans up the MariaDB tablespace
flags. We will reserve flags to store the page_compressed
compression algorithm, and to store the compressed payload length,
so that checksum can be computed over the compressed (and
possibly encrypted) stream and can be validated without
decrypting or decompressing the page.
In the full_crc32 format, there no longer are separate before-encryption
and after-encryption checksums for pages. The single checksum is
computed on the page contents that is written to the file.
We do not make the new algorithm the default for two reasons.
First, MariaDB 10.4.2 was a beta release, and the default values
of parameters should not change after beta. Second, we did not
yet implement the full_crc32 format for page_compressed pages.
This will be fixed in MDEV-18644.
This is joint work with Marko Mäkelä.
7 years ago MDEV-34830: LSN in the future is not being treated as serious corruption
The invariant of write-ahead logging is that before any change to a
page is written to the data file, the corresponding log record must
must first have been durably written.
In crash recovery, there were some sloppy checks for this. Let us
implement accurate checks and flag an inconsistency as a hard error,
so that we can avoid further corruption of a corrupted database.
For data extraction from the corrupted database, innodb_force_recovery
can be used.
Before recovery is reading any data pages or invoking
buf_dblwr_t::recover() to recover torn pages from the
doublewrite buffer, InnoDB will have parsed the log until the
final LSN and updated log_sys.lsn to that. So, we can rely on
log_sys.lsn at all times. The doublewrite buffer recovery has been
refactored in such a way that the recv_sys.dblwr.pages may be consulted
while discovering files and their page sizes, but nothing will be
written back to data files before buf_dblwr_t::recover() is invoked.
recv_max_page_lsn, recv_lsn_checks_on: Remove.
recv_sys_t::validate_checkpoint(): Validate the write-ahead-logging
condition at the end of the recovery.
recv_dblwr_t::validate_page(): Keep track of the maximum LSN
(if we are checking a non-doublewrite copy of a page) but
do not complain LSN being in the future. The doublewrite buffer
is a special case, because it will be read early during recovery.
Besides, starting with commit 762bcb81b5bf9bbde61fed59afb26417f4ce1e86
the dblwr=true copies of pages may legitimately be "too new".
recv_dblwr_t::find_page(): Find a valid page with the smallest
FIL_PAGE_LSN that is in the valid range for recovery.
recv_dblwr_t::restore_first_page(): Replaced by find_page().
Only buf_dblwr_t::recover() will write to data files.
buf_dblwr_t::recover(): Simplify the message output. Do attempt
doublewrite recovery on user page read error. Ignore doublewrite
pages whose FIL_PAGE_LSN is outside the usable bounds. Previously,
we could wrongly recover a too new page from the doublewrite buffer.
It is unlikely that this could have lead to an actual error.
Write back all recovered pages from the doublewrite buffer here,
including for the first page of any tablespace.
buf_page_is_corrupted(): Distinguish the return values
CORRUPTED_FUTURE_LSN and CORRUPTED_OTHER.
buf_page_check_corrupt(): Return the error code DB_CORRUPTION
in case the LSN is in the future.
Datafile::read_first_page_flags(): Split from read_first_page().
Take a copy of the first page as a parameter.
recv_sys_t::free_corrupted_page(): Take the file as a parameter
and return whether a message was displayed. This avoids some duplicated
and incomplete error messages.
buf_page_t::read_complete(): Remove some redundant output and always
display the name of the corrupted file. Never return DB_FAIL;
use it only in internal error handling.
IORequest::read_complete(): Assume that buf_page_t::read_complete()
will have reported any error.
fil_space_t::set_corrupted(): Return whether this is the first time
the tablespace had been flagged as corrupted.
Datafile::validate_first_page(), fil_node_open_file_low(),
fil_node_open_file(), fil_space_t::read_page0(),
fil_node_t::read_page0(): Add a parameter for a copy of the
first page, and a parameter to indicate whether the FIL_PAGE_LSN
check should be suppressed. Before buf_dblwr_t::recover() is
invoked, we cannot validate the FIL_PAGE_LSN, but we can trust the
FSP_SPACE_FLAGS and the tablespace ID that may be present in a
potentially too new copy of a page.
Reviewed by: Debarun Banerjee
12 months ago MDEV-34830: LSN in the future is not being treated as serious corruption
The invariant of write-ahead logging is that before any change to a
page is written to the data file, the corresponding log record must
must first have been durably written.
In crash recovery, there were some sloppy checks for this. Let us
implement accurate checks and flag an inconsistency as a hard error,
so that we can avoid further corruption of a corrupted database.
For data extraction from the corrupted database, innodb_force_recovery
can be used.
Before recovery is reading any data pages or invoking
buf_dblwr_t::recover() to recover torn pages from the
doublewrite buffer, InnoDB will have parsed the log until the
final LSN and updated log_sys.lsn to that. So, we can rely on
log_sys.lsn at all times. The doublewrite buffer recovery has been
refactored in such a way that the recv_sys.dblwr.pages may be consulted
while discovering files and their page sizes, but nothing will be
written back to data files before buf_dblwr_t::recover() is invoked.
recv_max_page_lsn, recv_lsn_checks_on: Remove.
recv_sys_t::validate_checkpoint(): Validate the write-ahead-logging
condition at the end of the recovery.
recv_dblwr_t::validate_page(): Keep track of the maximum LSN
(if we are checking a non-doublewrite copy of a page) but
do not complain LSN being in the future. The doublewrite buffer
is a special case, because it will be read early during recovery.
Besides, starting with commit 762bcb81b5bf9bbde61fed59afb26417f4ce1e86
the dblwr=true copies of pages may legitimately be "too new".
recv_dblwr_t::find_page(): Find a valid page with the smallest
FIL_PAGE_LSN that is in the valid range for recovery.
recv_dblwr_t::restore_first_page(): Replaced by find_page().
Only buf_dblwr_t::recover() will write to data files.
buf_dblwr_t::recover(): Simplify the message output. Do attempt
doublewrite recovery on user page read error. Ignore doublewrite
pages whose FIL_PAGE_LSN is outside the usable bounds. Previously,
we could wrongly recover a too new page from the doublewrite buffer.
It is unlikely that this could have lead to an actual error.
Write back all recovered pages from the doublewrite buffer here,
including for the first page of any tablespace.
buf_page_is_corrupted(): Distinguish the return values
CORRUPTED_FUTURE_LSN and CORRUPTED_OTHER.
buf_page_check_corrupt(): Return the error code DB_CORRUPTION
in case the LSN is in the future.
Datafile::read_first_page_flags(): Split from read_first_page().
Take a copy of the first page as a parameter.
recv_sys_t::free_corrupted_page(): Take the file as a parameter
and return whether a message was displayed. This avoids some duplicated
and incomplete error messages.
buf_page_t::read_complete(): Remove some redundant output and always
display the name of the corrupted file. Never return DB_FAIL;
use it only in internal error handling.
IORequest::read_complete(): Assume that buf_page_t::read_complete()
will have reported any error.
fil_space_t::set_corrupted(): Return whether this is the first time
the tablespace had been flagged as corrupted.
Datafile::validate_first_page(), fil_node_open_file_low(),
fil_node_open_file(), fil_space_t::read_page0(),
fil_node_t::read_page0(): Add a parameter for a copy of the
first page, and a parameter to indicate whether the FIL_PAGE_LSN
check should be suppressed. Before buf_dblwr_t::recover() is
invoked, we cannot validate the FIL_PAGE_LSN, but we can trust the
FSP_SPACE_FLAGS and the tablespace ID that may be present in a
potentially too new copy of a page.
Reviewed by: Debarun Banerjee
12 months ago MDEV-34830: LSN in the future is not being treated as serious corruption
The invariant of write-ahead logging is that before any change to a
page is written to the data file, the corresponding log record must
must first have been durably written.
In crash recovery, there were some sloppy checks for this. Let us
implement accurate checks and flag an inconsistency as a hard error,
so that we can avoid further corruption of a corrupted database.
For data extraction from the corrupted database, innodb_force_recovery
can be used.
Before recovery is reading any data pages or invoking
buf_dblwr_t::recover() to recover torn pages from the
doublewrite buffer, InnoDB will have parsed the log until the
final LSN and updated log_sys.lsn to that. So, we can rely on
log_sys.lsn at all times. The doublewrite buffer recovery has been
refactored in such a way that the recv_sys.dblwr.pages may be consulted
while discovering files and their page sizes, but nothing will be
written back to data files before buf_dblwr_t::recover() is invoked.
recv_max_page_lsn, recv_lsn_checks_on: Remove.
recv_sys_t::validate_checkpoint(): Validate the write-ahead-logging
condition at the end of the recovery.
recv_dblwr_t::validate_page(): Keep track of the maximum LSN
(if we are checking a non-doublewrite copy of a page) but
do not complain LSN being in the future. The doublewrite buffer
is a special case, because it will be read early during recovery.
Besides, starting with commit 762bcb81b5bf9bbde61fed59afb26417f4ce1e86
the dblwr=true copies of pages may legitimately be "too new".
recv_dblwr_t::find_page(): Find a valid page with the smallest
FIL_PAGE_LSN that is in the valid range for recovery.
recv_dblwr_t::restore_first_page(): Replaced by find_page().
Only buf_dblwr_t::recover() will write to data files.
buf_dblwr_t::recover(): Simplify the message output. Do attempt
doublewrite recovery on user page read error. Ignore doublewrite
pages whose FIL_PAGE_LSN is outside the usable bounds. Previously,
we could wrongly recover a too new page from the doublewrite buffer.
It is unlikely that this could have lead to an actual error.
Write back all recovered pages from the doublewrite buffer here,
including for the first page of any tablespace.
buf_page_is_corrupted(): Distinguish the return values
CORRUPTED_FUTURE_LSN and CORRUPTED_OTHER.
buf_page_check_corrupt(): Return the error code DB_CORRUPTION
in case the LSN is in the future.
Datafile::read_first_page_flags(): Split from read_first_page().
Take a copy of the first page as a parameter.
recv_sys_t::free_corrupted_page(): Take the file as a parameter
and return whether a message was displayed. This avoids some duplicated
and incomplete error messages.
buf_page_t::read_complete(): Remove some redundant output and always
display the name of the corrupted file. Never return DB_FAIL;
use it only in internal error handling.
IORequest::read_complete(): Assume that buf_page_t::read_complete()
will have reported any error.
fil_space_t::set_corrupted(): Return whether this is the first time
the tablespace had been flagged as corrupted.
Datafile::validate_first_page(), fil_node_open_file_low(),
fil_node_open_file(), fil_space_t::read_page0(),
fil_node_t::read_page0(): Add a parameter for a copy of the
first page, and a parameter to indicate whether the FIL_PAGE_LSN
check should be suppressed. Before buf_dblwr_t::recover() is
invoked, we cannot validate the FIL_PAGE_LSN, but we can trust the
FSP_SPACE_FLAGS and the tablespace ID that may be present in a
potentially too new copy of a page.
Reviewed by: Debarun Banerjee
12 months ago MDEV-25105 Remove innodb_checksum_algorithm values none,innodb,...
Historically, InnoDB supported a buggy page checksum algorithm that did not
compute a checksum over the full page. Later, well before MySQL 4.1
introduced .ibd files and the innodb_file_per_table option, the algorithm
was corrected and the first 4 bytes of each page were redefined to be
a checksum.
The original checksum was so slow that an option to disable page checksum
was introduced for benchmarketing purposes.
The Intel Nehalem microarchitecture introduced the SSE4.2 instruction set
extension, which includes instructions for faster computation of CRC-32C.
In MySQL 5.6 (and MariaDB 10.0), innodb_checksum_algorithm=crc32 was
implemented to make of that. As that option was changed to be the default
in MySQL 5.7, a bug was found on big-endian platforms and some work-around
code was added to weaken that checksum further. MariaDB disables that
work-around by default since MDEV-17958.
Later, SIMD-accelerated CRC-32C has been implemented in MariaDB for POWER
and ARM and also for IA-32/AMD64, making use of carry-less multiplication
where available.
Long story short, innodb_checksum_algorithm=crc32 is faster and more secure
than the pre-MySQL 5.6 checksum, called innodb_checksum_algorithm=innodb.
It should have removed any need to use innodb_checksum_algorithm=none.
The setting innodb_checksum_algorithm=crc32 is the default in
MySQL 5.7 and MariaDB Server 10.2, 10.3, 10.4. In MariaDB 10.5,
MDEV-19534 made innodb_checksum_algorithm=full_crc32 the default.
It is even faster and more secure.
The default settings in MariaDB do allow old data files to be read,
no matter if a worse checksum algorithm had been used.
(Unfortunately, before innodb_checksum_algorithm=full_crc32,
the data files did not identify which checksum algorithm is being used.)
The non-default settings innodb_checksum_algorithm=strict_crc32 or
innodb_checksum_algorithm=strict_full_crc32 would only allow CRC-32C
checksums. The incompatibility with old data files is why they are
not the default.
The newest server not to support innodb_checksum_algorithm=crc32
were MySQL 5.5 and MariaDB 5.5. Both have reached their end of life.
A valid reason for using innodb_checksum_algorithm=innodb could have
been the ability to downgrade. If it is really needed, data files
can be converted with an older version of the innochecksum utility.
Because there is no good reason to allow data files to be written
with insecure checksums, we will reject those option values:
innodb_checksum_algorithm=none
innodb_checksum_algorithm=innodb
innodb_checksum_algorithm=strict_none
innodb_checksum_algorithm=strict_innodb
Furthermore, the following innochecksum options will be removed,
because only strict crc32 will be supported:
innochecksum --strict-check=crc32
innochecksum -C crc32
innochecksum --write=crc32
innochecksum -w crc32
If a user wishes to convert a data file to use a different checksum
(so that it might be used with the no-longer-supported
MySQL 5.5 or MariaDB 5.5, which do not support IMPORT TABLESPACE
nor system tablespace format changes that were made in MariaDB 10.3),
then the innochecksum tool from MariaDB 10.2, 10.3, 10.4, 10.5 or
MySQL 5.7 can be used.
Reviewed by: Thirunarayanan Balathandayuthapani
5 years ago MDEV-25105 Remove innodb_checksum_algorithm values none,innodb,...
Historically, InnoDB supported a buggy page checksum algorithm that did not
compute a checksum over the full page. Later, well before MySQL 4.1
introduced .ibd files and the innodb_file_per_table option, the algorithm
was corrected and the first 4 bytes of each page were redefined to be
a checksum.
The original checksum was so slow that an option to disable page checksum
was introduced for benchmarketing purposes.
The Intel Nehalem microarchitecture introduced the SSE4.2 instruction set
extension, which includes instructions for faster computation of CRC-32C.
In MySQL 5.6 (and MariaDB 10.0), innodb_checksum_algorithm=crc32 was
implemented to make of that. As that option was changed to be the default
in MySQL 5.7, a bug was found on big-endian platforms and some work-around
code was added to weaken that checksum further. MariaDB disables that
work-around by default since MDEV-17958.
Later, SIMD-accelerated CRC-32C has been implemented in MariaDB for POWER
and ARM and also for IA-32/AMD64, making use of carry-less multiplication
where available.
Long story short, innodb_checksum_algorithm=crc32 is faster and more secure
than the pre-MySQL 5.6 checksum, called innodb_checksum_algorithm=innodb.
It should have removed any need to use innodb_checksum_algorithm=none.
The setting innodb_checksum_algorithm=crc32 is the default in
MySQL 5.7 and MariaDB Server 10.2, 10.3, 10.4. In MariaDB 10.5,
MDEV-19534 made innodb_checksum_algorithm=full_crc32 the default.
It is even faster and more secure.
The default settings in MariaDB do allow old data files to be read,
no matter if a worse checksum algorithm had been used.
(Unfortunately, before innodb_checksum_algorithm=full_crc32,
the data files did not identify which checksum algorithm is being used.)
The non-default settings innodb_checksum_algorithm=strict_crc32 or
innodb_checksum_algorithm=strict_full_crc32 would only allow CRC-32C
checksums. The incompatibility with old data files is why they are
not the default.
The newest server not to support innodb_checksum_algorithm=crc32
were MySQL 5.5 and MariaDB 5.5. Both have reached their end of life.
A valid reason for using innodb_checksum_algorithm=innodb could have
been the ability to downgrade. If it is really needed, data files
can be converted with an older version of the innochecksum utility.
Because there is no good reason to allow data files to be written
with insecure checksums, we will reject those option values:
innodb_checksum_algorithm=none
innodb_checksum_algorithm=innodb
innodb_checksum_algorithm=strict_none
innodb_checksum_algorithm=strict_innodb
Furthermore, the following innochecksum options will be removed,
because only strict crc32 will be supported:
innochecksum --strict-check=crc32
innochecksum -C crc32
innochecksum --write=crc32
innochecksum -w crc32
If a user wishes to convert a data file to use a different checksum
(so that it might be used with the no-longer-supported
MySQL 5.5 or MariaDB 5.5, which do not support IMPORT TABLESPACE
nor system tablespace format changes that were made in MariaDB 10.3),
then the innochecksum tool from MariaDB 10.2, 10.3, 10.4, 10.5 or
MySQL 5.7 can be used.
Reviewed by: Thirunarayanan Balathandayuthapani
5 years ago MDEV-34830: LSN in the future is not being treated as serious corruption
The invariant of write-ahead logging is that before any change to a
page is written to the data file, the corresponding log record must
must first have been durably written.
In crash recovery, there were some sloppy checks for this. Let us
implement accurate checks and flag an inconsistency as a hard error,
so that we can avoid further corruption of a corrupted database.
For data extraction from the corrupted database, innodb_force_recovery
can be used.
Before recovery is reading any data pages or invoking
buf_dblwr_t::recover() to recover torn pages from the
doublewrite buffer, InnoDB will have parsed the log until the
final LSN and updated log_sys.lsn to that. So, we can rely on
log_sys.lsn at all times. The doublewrite buffer recovery has been
refactored in such a way that the recv_sys.dblwr.pages may be consulted
while discovering files and their page sizes, but nothing will be
written back to data files before buf_dblwr_t::recover() is invoked.
recv_max_page_lsn, recv_lsn_checks_on: Remove.
recv_sys_t::validate_checkpoint(): Validate the write-ahead-logging
condition at the end of the recovery.
recv_dblwr_t::validate_page(): Keep track of the maximum LSN
(if we are checking a non-doublewrite copy of a page) but
do not complain LSN being in the future. The doublewrite buffer
is a special case, because it will be read early during recovery.
Besides, starting with commit 762bcb81b5bf9bbde61fed59afb26417f4ce1e86
the dblwr=true copies of pages may legitimately be "too new".
recv_dblwr_t::find_page(): Find a valid page with the smallest
FIL_PAGE_LSN that is in the valid range for recovery.
recv_dblwr_t::restore_first_page(): Replaced by find_page().
Only buf_dblwr_t::recover() will write to data files.
buf_dblwr_t::recover(): Simplify the message output. Do attempt
doublewrite recovery on user page read error. Ignore doublewrite
pages whose FIL_PAGE_LSN is outside the usable bounds. Previously,
we could wrongly recover a too new page from the doublewrite buffer.
It is unlikely that this could have lead to an actual error.
Write back all recovered pages from the doublewrite buffer here,
including for the first page of any tablespace.
buf_page_is_corrupted(): Distinguish the return values
CORRUPTED_FUTURE_LSN and CORRUPTED_OTHER.
buf_page_check_corrupt(): Return the error code DB_CORRUPTION
in case the LSN is in the future.
Datafile::read_first_page_flags(): Split from read_first_page().
Take a copy of the first page as a parameter.
recv_sys_t::free_corrupted_page(): Take the file as a parameter
and return whether a message was displayed. This avoids some duplicated
and incomplete error messages.
buf_page_t::read_complete(): Remove some redundant output and always
display the name of the corrupted file. Never return DB_FAIL;
use it only in internal error handling.
IORequest::read_complete(): Assume that buf_page_t::read_complete()
will have reported any error.
fil_space_t::set_corrupted(): Return whether this is the first time
the tablespace had been flagged as corrupted.
Datafile::validate_first_page(), fil_node_open_file_low(),
fil_node_open_file(), fil_space_t::read_page0(),
fil_node_t::read_page0(): Add a parameter for a copy of the
first page, and a parameter to indicate whether the FIL_PAGE_LSN
check should be suppressed. Before buf_dblwr_t::recover() is
invoked, we cannot validate the FIL_PAGE_LSN, but we can trust the
FSP_SPACE_FLAGS and the tablespace ID that may be present in a
potentially too new copy of a page.
Reviewed by: Debarun Banerjee
12 months ago MDEV-25105 Remove innodb_checksum_algorithm values none,innodb,...
Historically, InnoDB supported a buggy page checksum algorithm that did not
compute a checksum over the full page. Later, well before MySQL 4.1
introduced .ibd files and the innodb_file_per_table option, the algorithm
was corrected and the first 4 bytes of each page were redefined to be
a checksum.
The original checksum was so slow that an option to disable page checksum
was introduced for benchmarketing purposes.
The Intel Nehalem microarchitecture introduced the SSE4.2 instruction set
extension, which includes instructions for faster computation of CRC-32C.
In MySQL 5.6 (and MariaDB 10.0), innodb_checksum_algorithm=crc32 was
implemented to make of that. As that option was changed to be the default
in MySQL 5.7, a bug was found on big-endian platforms and some work-around
code was added to weaken that checksum further. MariaDB disables that
work-around by default since MDEV-17958.
Later, SIMD-accelerated CRC-32C has been implemented in MariaDB for POWER
and ARM and also for IA-32/AMD64, making use of carry-less multiplication
where available.
Long story short, innodb_checksum_algorithm=crc32 is faster and more secure
than the pre-MySQL 5.6 checksum, called innodb_checksum_algorithm=innodb.
It should have removed any need to use innodb_checksum_algorithm=none.
The setting innodb_checksum_algorithm=crc32 is the default in
MySQL 5.7 and MariaDB Server 10.2, 10.3, 10.4. In MariaDB 10.5,
MDEV-19534 made innodb_checksum_algorithm=full_crc32 the default.
It is even faster and more secure.
The default settings in MariaDB do allow old data files to be read,
no matter if a worse checksum algorithm had been used.
(Unfortunately, before innodb_checksum_algorithm=full_crc32,
the data files did not identify which checksum algorithm is being used.)
The non-default settings innodb_checksum_algorithm=strict_crc32 or
innodb_checksum_algorithm=strict_full_crc32 would only allow CRC-32C
checksums. The incompatibility with old data files is why they are
not the default.
The newest server not to support innodb_checksum_algorithm=crc32
were MySQL 5.5 and MariaDB 5.5. Both have reached their end of life.
A valid reason for using innodb_checksum_algorithm=innodb could have
been the ability to downgrade. If it is really needed, data files
can be converted with an older version of the innochecksum utility.
Because there is no good reason to allow data files to be written
with insecure checksums, we will reject those option values:
innodb_checksum_algorithm=none
innodb_checksum_algorithm=innodb
innodb_checksum_algorithm=strict_none
innodb_checksum_algorithm=strict_innodb
Furthermore, the following innochecksum options will be removed,
because only strict crc32 will be supported:
innochecksum --strict-check=crc32
innochecksum -C crc32
innochecksum --write=crc32
innochecksum -w crc32
If a user wishes to convert a data file to use a different checksum
(so that it might be used with the no-longer-supported
MySQL 5.5 or MariaDB 5.5, which do not support IMPORT TABLESPACE
nor system tablespace format changes that were made in MariaDB 10.3),
then the innochecksum tool from MariaDB 10.2, 10.3, 10.4, 10.5 or
MySQL 5.7 can be used.
Reviewed by: Thirunarayanan Balathandayuthapani
5 years ago MDEV-12026: Implement innodb_checksum_algorithm=full_crc32
MariaDB data-at-rest encryption (innodb_encrypt_tables)
had repurposed the same unused data field that was repurposed
in MySQL 5.7 (and MariaDB 10.2) for the Split Sequence Number (SSN)
field of SPATIAL INDEX. Because of this, MariaDB was unable to
support encryption on SPATIAL INDEX pages.
Furthermore, InnoDB page checksums skipped some bytes, and there
are multiple variations and checksum algorithms. By default,
InnoDB accepts all variations of all algorithms that ever existed.
This unnecessarily weakens the page checksums.
We hereby introduce two more innodb_checksum_algorithm variants
(full_crc32, strict_full_crc32) that are special in a way:
When either setting is active, newly created data files will
carry a flag (fil_space_t::full_crc32()) that indicates that
all pages of the file will use a full CRC-32C checksum over the
entire page contents (excluding the bytes where the checksum
is stored, at the very end of the page). Such files will always
use that checksum, no matter what the parameter
innodb_checksum_algorithm is assigned to.
For old files, the old checksum algorithms will continue to be
used. The value strict_full_crc32 will be equivalent to strict_crc32
and the value full_crc32 will be equivalent to crc32.
ROW_FORMAT=COMPRESSED tables will only use the old format.
These tables do not support new features, such as larger
innodb_page_size or instant ADD/DROP COLUMN. They may be
deprecated in the future. We do not want an unnecessary
file format change for them.
The new full_crc32() format also cleans up the MariaDB tablespace
flags. We will reserve flags to store the page_compressed
compression algorithm, and to store the compressed payload length,
so that checksum can be computed over the compressed (and
possibly encrypted) stream and can be validated without
decrypting or decompressing the page.
In the full_crc32 format, there no longer are separate before-encryption
and after-encryption checksums for pages. The single checksum is
computed on the page contents that is written to the file.
We do not make the new algorithm the default for two reasons.
First, MariaDB 10.4.2 was a beta release, and the default values
of parameters should not change after beta. Second, we did not
yet implement the full_crc32 format for page_compressed pages.
This will be fixed in MDEV-18644.
This is joint work with Marko Mäkelä.
7 years ago MDEV-25105 Remove innodb_checksum_algorithm values none,innodb,...
Historically, InnoDB supported a buggy page checksum algorithm that did not
compute a checksum over the full page. Later, well before MySQL 4.1
introduced .ibd files and the innodb_file_per_table option, the algorithm
was corrected and the first 4 bytes of each page were redefined to be
a checksum.
The original checksum was so slow that an option to disable page checksum
was introduced for benchmarketing purposes.
The Intel Nehalem microarchitecture introduced the SSE4.2 instruction set
extension, which includes instructions for faster computation of CRC-32C.
In MySQL 5.6 (and MariaDB 10.0), innodb_checksum_algorithm=crc32 was
implemented to make of that. As that option was changed to be the default
in MySQL 5.7, a bug was found on big-endian platforms and some work-around
code was added to weaken that checksum further. MariaDB disables that
work-around by default since MDEV-17958.
Later, SIMD-accelerated CRC-32C has been implemented in MariaDB for POWER
and ARM and also for IA-32/AMD64, making use of carry-less multiplication
where available.
Long story short, innodb_checksum_algorithm=crc32 is faster and more secure
than the pre-MySQL 5.6 checksum, called innodb_checksum_algorithm=innodb.
It should have removed any need to use innodb_checksum_algorithm=none.
The setting innodb_checksum_algorithm=crc32 is the default in
MySQL 5.7 and MariaDB Server 10.2, 10.3, 10.4. In MariaDB 10.5,
MDEV-19534 made innodb_checksum_algorithm=full_crc32 the default.
It is even faster and more secure.
The default settings in MariaDB do allow old data files to be read,
no matter if a worse checksum algorithm had been used.
(Unfortunately, before innodb_checksum_algorithm=full_crc32,
the data files did not identify which checksum algorithm is being used.)
The non-default settings innodb_checksum_algorithm=strict_crc32 or
innodb_checksum_algorithm=strict_full_crc32 would only allow CRC-32C
checksums. The incompatibility with old data files is why they are
not the default.
The newest server not to support innodb_checksum_algorithm=crc32
were MySQL 5.5 and MariaDB 5.5. Both have reached their end of life.
A valid reason for using innodb_checksum_algorithm=innodb could have
been the ability to downgrade. If it is really needed, data files
can be converted with an older version of the innochecksum utility.
Because there is no good reason to allow data files to be written
with insecure checksums, we will reject those option values:
innodb_checksum_algorithm=none
innodb_checksum_algorithm=innodb
innodb_checksum_algorithm=strict_none
innodb_checksum_algorithm=strict_innodb
Furthermore, the following innochecksum options will be removed,
because only strict crc32 will be supported:
innochecksum --strict-check=crc32
innochecksum -C crc32
innochecksum --write=crc32
innochecksum -w crc32
If a user wishes to convert a data file to use a different checksum
(so that it might be used with the no-longer-supported
MySQL 5.5 or MariaDB 5.5, which do not support IMPORT TABLESPACE
nor system tablespace format changes that were made in MariaDB 10.3),
then the innochecksum tool from MariaDB 10.2, 10.3, 10.4, 10.5 or
MySQL 5.7 can be used.
Reviewed by: Thirunarayanan Balathandayuthapani
5 years ago MDEV-34830: LSN in the future is not being treated as serious corruption
The invariant of write-ahead logging is that before any change to a
page is written to the data file, the corresponding log record must
must first have been durably written.
In crash recovery, there were some sloppy checks for this. Let us
implement accurate checks and flag an inconsistency as a hard error,
so that we can avoid further corruption of a corrupted database.
For data extraction from the corrupted database, innodb_force_recovery
can be used.
Before recovery is reading any data pages or invoking
buf_dblwr_t::recover() to recover torn pages from the
doublewrite buffer, InnoDB will have parsed the log until the
final LSN and updated log_sys.lsn to that. So, we can rely on
log_sys.lsn at all times. The doublewrite buffer recovery has been
refactored in such a way that the recv_sys.dblwr.pages may be consulted
while discovering files and their page sizes, but nothing will be
written back to data files before buf_dblwr_t::recover() is invoked.
recv_max_page_lsn, recv_lsn_checks_on: Remove.
recv_sys_t::validate_checkpoint(): Validate the write-ahead-logging
condition at the end of the recovery.
recv_dblwr_t::validate_page(): Keep track of the maximum LSN
(if we are checking a non-doublewrite copy of a page) but
do not complain LSN being in the future. The doublewrite buffer
is a special case, because it will be read early during recovery.
Besides, starting with commit 762bcb81b5bf9bbde61fed59afb26417f4ce1e86
the dblwr=true copies of pages may legitimately be "too new".
recv_dblwr_t::find_page(): Find a valid page with the smallest
FIL_PAGE_LSN that is in the valid range for recovery.
recv_dblwr_t::restore_first_page(): Replaced by find_page().
Only buf_dblwr_t::recover() will write to data files.
buf_dblwr_t::recover(): Simplify the message output. Do attempt
doublewrite recovery on user page read error. Ignore doublewrite
pages whose FIL_PAGE_LSN is outside the usable bounds. Previously,
we could wrongly recover a too new page from the doublewrite buffer.
It is unlikely that this could have lead to an actual error.
Write back all recovered pages from the doublewrite buffer here,
including for the first page of any tablespace.
buf_page_is_corrupted(): Distinguish the return values
CORRUPTED_FUTURE_LSN and CORRUPTED_OTHER.
buf_page_check_corrupt(): Return the error code DB_CORRUPTION
in case the LSN is in the future.
Datafile::read_first_page_flags(): Split from read_first_page().
Take a copy of the first page as a parameter.
recv_sys_t::free_corrupted_page(): Take the file as a parameter
and return whether a message was displayed. This avoids some duplicated
and incomplete error messages.
buf_page_t::read_complete(): Remove some redundant output and always
display the name of the corrupted file. Never return DB_FAIL;
use it only in internal error handling.
IORequest::read_complete(): Assume that buf_page_t::read_complete()
will have reported any error.
fil_space_t::set_corrupted(): Return whether this is the first time
the tablespace had been flagged as corrupted.
Datafile::validate_first_page(), fil_node_open_file_low(),
fil_node_open_file(), fil_space_t::read_page0(),
fil_node_t::read_page0(): Add a parameter for a copy of the
first page, and a parameter to indicate whether the FIL_PAGE_LSN
check should be suppressed. Before buf_dblwr_t::recover() is
invoked, we cannot validate the FIL_PAGE_LSN, but we can trust the
FSP_SPACE_FLAGS and the tablespace ID that may be present in a
potentially too new copy of a page.
Reviewed by: Debarun Banerjee
12 months ago MDEV-25105 Remove innodb_checksum_algorithm values none,innodb,...
Historically, InnoDB supported a buggy page checksum algorithm that did not
compute a checksum over the full page. Later, well before MySQL 4.1
introduced .ibd files and the innodb_file_per_table option, the algorithm
was corrected and the first 4 bytes of each page were redefined to be
a checksum.
The original checksum was so slow that an option to disable page checksum
was introduced for benchmarketing purposes.
The Intel Nehalem microarchitecture introduced the SSE4.2 instruction set
extension, which includes instructions for faster computation of CRC-32C.
In MySQL 5.6 (and MariaDB 10.0), innodb_checksum_algorithm=crc32 was
implemented to make of that. As that option was changed to be the default
in MySQL 5.7, a bug was found on big-endian platforms and some work-around
code was added to weaken that checksum further. MariaDB disables that
work-around by default since MDEV-17958.
Later, SIMD-accelerated CRC-32C has been implemented in MariaDB for POWER
and ARM and also for IA-32/AMD64, making use of carry-less multiplication
where available.
Long story short, innodb_checksum_algorithm=crc32 is faster and more secure
than the pre-MySQL 5.6 checksum, called innodb_checksum_algorithm=innodb.
It should have removed any need to use innodb_checksum_algorithm=none.
The setting innodb_checksum_algorithm=crc32 is the default in
MySQL 5.7 and MariaDB Server 10.2, 10.3, 10.4. In MariaDB 10.5,
MDEV-19534 made innodb_checksum_algorithm=full_crc32 the default.
It is even faster and more secure.
The default settings in MariaDB do allow old data files to be read,
no matter if a worse checksum algorithm had been used.
(Unfortunately, before innodb_checksum_algorithm=full_crc32,
the data files did not identify which checksum algorithm is being used.)
The non-default settings innodb_checksum_algorithm=strict_crc32 or
innodb_checksum_algorithm=strict_full_crc32 would only allow CRC-32C
checksums. The incompatibility with old data files is why they are
not the default.
The newest server not to support innodb_checksum_algorithm=crc32
were MySQL 5.5 and MariaDB 5.5. Both have reached their end of life.
A valid reason for using innodb_checksum_algorithm=innodb could have
been the ability to downgrade. If it is really needed, data files
can be converted with an older version of the innochecksum utility.
Because there is no good reason to allow data files to be written
with insecure checksums, we will reject those option values:
innodb_checksum_algorithm=none
innodb_checksum_algorithm=innodb
innodb_checksum_algorithm=strict_none
innodb_checksum_algorithm=strict_innodb
Furthermore, the following innochecksum options will be removed,
because only strict crc32 will be supported:
innochecksum --strict-check=crc32
innochecksum -C crc32
innochecksum --write=crc32
innochecksum -w crc32
If a user wishes to convert a data file to use a different checksum
(so that it might be used with the no-longer-supported
MySQL 5.5 or MariaDB 5.5, which do not support IMPORT TABLESPACE
nor system tablespace format changes that were made in MariaDB 10.3),
then the innochecksum tool from MariaDB 10.2, 10.3, 10.4, 10.5 or
MySQL 5.7 can be used.
Reviewed by: Thirunarayanan Balathandayuthapani
5 years ago MDEV-34830: LSN in the future is not being treated as serious corruption
The invariant of write-ahead logging is that before any change to a
page is written to the data file, the corresponding log record must
must first have been durably written.
In crash recovery, there were some sloppy checks for this. Let us
implement accurate checks and flag an inconsistency as a hard error,
so that we can avoid further corruption of a corrupted database.
For data extraction from the corrupted database, innodb_force_recovery
can be used.
Before recovery is reading any data pages or invoking
buf_dblwr_t::recover() to recover torn pages from the
doublewrite buffer, InnoDB will have parsed the log until the
final LSN and updated log_sys.lsn to that. So, we can rely on
log_sys.lsn at all times. The doublewrite buffer recovery has been
refactored in such a way that the recv_sys.dblwr.pages may be consulted
while discovering files and their page sizes, but nothing will be
written back to data files before buf_dblwr_t::recover() is invoked.
recv_max_page_lsn, recv_lsn_checks_on: Remove.
recv_sys_t::validate_checkpoint(): Validate the write-ahead-logging
condition at the end of the recovery.
recv_dblwr_t::validate_page(): Keep track of the maximum LSN
(if we are checking a non-doublewrite copy of a page) but
do not complain LSN being in the future. The doublewrite buffer
is a special case, because it will be read early during recovery.
Besides, starting with commit 762bcb81b5bf9bbde61fed59afb26417f4ce1e86
the dblwr=true copies of pages may legitimately be "too new".
recv_dblwr_t::find_page(): Find a valid page with the smallest
FIL_PAGE_LSN that is in the valid range for recovery.
recv_dblwr_t::restore_first_page(): Replaced by find_page().
Only buf_dblwr_t::recover() will write to data files.
buf_dblwr_t::recover(): Simplify the message output. Do attempt
doublewrite recovery on user page read error. Ignore doublewrite
pages whose FIL_PAGE_LSN is outside the usable bounds. Previously,
we could wrongly recover a too new page from the doublewrite buffer.
It is unlikely that this could have lead to an actual error.
Write back all recovered pages from the doublewrite buffer here,
including for the first page of any tablespace.
buf_page_is_corrupted(): Distinguish the return values
CORRUPTED_FUTURE_LSN and CORRUPTED_OTHER.
buf_page_check_corrupt(): Return the error code DB_CORRUPTION
in case the LSN is in the future.
Datafile::read_first_page_flags(): Split from read_first_page().
Take a copy of the first page as a parameter.
recv_sys_t::free_corrupted_page(): Take the file as a parameter
and return whether a message was displayed. This avoids some duplicated
and incomplete error messages.
buf_page_t::read_complete(): Remove some redundant output and always
display the name of the corrupted file. Never return DB_FAIL;
use it only in internal error handling.
IORequest::read_complete(): Assume that buf_page_t::read_complete()
will have reported any error.
fil_space_t::set_corrupted(): Return whether this is the first time
the tablespace had been flagged as corrupted.
Datafile::validate_first_page(), fil_node_open_file_low(),
fil_node_open_file(), fil_space_t::read_page0(),
fil_node_t::read_page0(): Add a parameter for a copy of the
first page, and a parameter to indicate whether the FIL_PAGE_LSN
check should be suppressed. Before buf_dblwr_t::recover() is
invoked, we cannot validate the FIL_PAGE_LSN, but we can trust the
FSP_SPACE_FLAGS and the tablespace ID that may be present in a
potentially too new copy of a page.
Reviewed by: Debarun Banerjee
12 months ago MDEV-25105 Remove innodb_checksum_algorithm values none,innodb,...
Historically, InnoDB supported a buggy page checksum algorithm that did not
compute a checksum over the full page. Later, well before MySQL 4.1
introduced .ibd files and the innodb_file_per_table option, the algorithm
was corrected and the first 4 bytes of each page were redefined to be
a checksum.
The original checksum was so slow that an option to disable page checksum
was introduced for benchmarketing purposes.
The Intel Nehalem microarchitecture introduced the SSE4.2 instruction set
extension, which includes instructions for faster computation of CRC-32C.
In MySQL 5.6 (and MariaDB 10.0), innodb_checksum_algorithm=crc32 was
implemented to make of that. As that option was changed to be the default
in MySQL 5.7, a bug was found on big-endian platforms and some work-around
code was added to weaken that checksum further. MariaDB disables that
work-around by default since MDEV-17958.
Later, SIMD-accelerated CRC-32C has been implemented in MariaDB for POWER
and ARM and also for IA-32/AMD64, making use of carry-less multiplication
where available.
Long story short, innodb_checksum_algorithm=crc32 is faster and more secure
than the pre-MySQL 5.6 checksum, called innodb_checksum_algorithm=innodb.
It should have removed any need to use innodb_checksum_algorithm=none.
The setting innodb_checksum_algorithm=crc32 is the default in
MySQL 5.7 and MariaDB Server 10.2, 10.3, 10.4. In MariaDB 10.5,
MDEV-19534 made innodb_checksum_algorithm=full_crc32 the default.
It is even faster and more secure.
The default settings in MariaDB do allow old data files to be read,
no matter if a worse checksum algorithm had been used.
(Unfortunately, before innodb_checksum_algorithm=full_crc32,
the data files did not identify which checksum algorithm is being used.)
The non-default settings innodb_checksum_algorithm=strict_crc32 or
innodb_checksum_algorithm=strict_full_crc32 would only allow CRC-32C
checksums. The incompatibility with old data files is why they are
not the default.
The newest server not to support innodb_checksum_algorithm=crc32
were MySQL 5.5 and MariaDB 5.5. Both have reached their end of life.
A valid reason for using innodb_checksum_algorithm=innodb could have
been the ability to downgrade. If it is really needed, data files
can be converted with an older version of the innochecksum utility.
Because there is no good reason to allow data files to be written
with insecure checksums, we will reject those option values:
innodb_checksum_algorithm=none
innodb_checksum_algorithm=innodb
innodb_checksum_algorithm=strict_none
innodb_checksum_algorithm=strict_innodb
Furthermore, the following innochecksum options will be removed,
because only strict crc32 will be supported:
innochecksum --strict-check=crc32
innochecksum -C crc32
innochecksum --write=crc32
innochecksum -w crc32
If a user wishes to convert a data file to use a different checksum
(so that it might be used with the no-longer-supported
MySQL 5.5 or MariaDB 5.5, which do not support IMPORT TABLESPACE
nor system tablespace format changes that were made in MariaDB 10.3),
then the innochecksum tool from MariaDB 10.2, 10.3, 10.4, 10.5 or
MySQL 5.7 can be used.
Reviewed by: Thirunarayanan Balathandayuthapani
5 years ago MDEV-34830: LSN in the future is not being treated as serious corruption
The invariant of write-ahead logging is that before any change to a
page is written to the data file, the corresponding log record must
must first have been durably written.
In crash recovery, there were some sloppy checks for this. Let us
implement accurate checks and flag an inconsistency as a hard error,
so that we can avoid further corruption of a corrupted database.
For data extraction from the corrupted database, innodb_force_recovery
can be used.
Before recovery is reading any data pages or invoking
buf_dblwr_t::recover() to recover torn pages from the
doublewrite buffer, InnoDB will have parsed the log until the
final LSN and updated log_sys.lsn to that. So, we can rely on
log_sys.lsn at all times. The doublewrite buffer recovery has been
refactored in such a way that the recv_sys.dblwr.pages may be consulted
while discovering files and their page sizes, but nothing will be
written back to data files before buf_dblwr_t::recover() is invoked.
recv_max_page_lsn, recv_lsn_checks_on: Remove.
recv_sys_t::validate_checkpoint(): Validate the write-ahead-logging
condition at the end of the recovery.
recv_dblwr_t::validate_page(): Keep track of the maximum LSN
(if we are checking a non-doublewrite copy of a page) but
do not complain LSN being in the future. The doublewrite buffer
is a special case, because it will be read early during recovery.
Besides, starting with commit 762bcb81b5bf9bbde61fed59afb26417f4ce1e86
the dblwr=true copies of pages may legitimately be "too new".
recv_dblwr_t::find_page(): Find a valid page with the smallest
FIL_PAGE_LSN that is in the valid range for recovery.
recv_dblwr_t::restore_first_page(): Replaced by find_page().
Only buf_dblwr_t::recover() will write to data files.
buf_dblwr_t::recover(): Simplify the message output. Do attempt
doublewrite recovery on user page read error. Ignore doublewrite
pages whose FIL_PAGE_LSN is outside the usable bounds. Previously,
we could wrongly recover a too new page from the doublewrite buffer.
It is unlikely that this could have lead to an actual error.
Write back all recovered pages from the doublewrite buffer here,
including for the first page of any tablespace.
buf_page_is_corrupted(): Distinguish the return values
CORRUPTED_FUTURE_LSN and CORRUPTED_OTHER.
buf_page_check_corrupt(): Return the error code DB_CORRUPTION
in case the LSN is in the future.
Datafile::read_first_page_flags(): Split from read_first_page().
Take a copy of the first page as a parameter.
recv_sys_t::free_corrupted_page(): Take the file as a parameter
and return whether a message was displayed. This avoids some duplicated
and incomplete error messages.
buf_page_t::read_complete(): Remove some redundant output and always
display the name of the corrupted file. Never return DB_FAIL;
use it only in internal error handling.
IORequest::read_complete(): Assume that buf_page_t::read_complete()
will have reported any error.
fil_space_t::set_corrupted(): Return whether this is the first time
the tablespace had been flagged as corrupted.
Datafile::validate_first_page(), fil_node_open_file_low(),
fil_node_open_file(), fil_space_t::read_page0(),
fil_node_t::read_page0(): Add a parameter for a copy of the
first page, and a parameter to indicate whether the FIL_PAGE_LSN
check should be suppressed. Before buf_dblwr_t::recover() is
invoked, we cannot validate the FIL_PAGE_LSN, but we can trust the
FSP_SPACE_FLAGS and the tablespace ID that may be present in a
potentially too new copy of a page.
Reviewed by: Debarun Banerjee
12 months ago MDEV-25105 Remove innodb_checksum_algorithm values none,innodb,...
Historically, InnoDB supported a buggy page checksum algorithm that did not
compute a checksum over the full page. Later, well before MySQL 4.1
introduced .ibd files and the innodb_file_per_table option, the algorithm
was corrected and the first 4 bytes of each page were redefined to be
a checksum.
The original checksum was so slow that an option to disable page checksum
was introduced for benchmarketing purposes.
The Intel Nehalem microarchitecture introduced the SSE4.2 instruction set
extension, which includes instructions for faster computation of CRC-32C.
In MySQL 5.6 (and MariaDB 10.0), innodb_checksum_algorithm=crc32 was
implemented to make of that. As that option was changed to be the default
in MySQL 5.7, a bug was found on big-endian platforms and some work-around
code was added to weaken that checksum further. MariaDB disables that
work-around by default since MDEV-17958.
Later, SIMD-accelerated CRC-32C has been implemented in MariaDB for POWER
and ARM and also for IA-32/AMD64, making use of carry-less multiplication
where available.
Long story short, innodb_checksum_algorithm=crc32 is faster and more secure
than the pre-MySQL 5.6 checksum, called innodb_checksum_algorithm=innodb.
It should have removed any need to use innodb_checksum_algorithm=none.
The setting innodb_checksum_algorithm=crc32 is the default in
MySQL 5.7 and MariaDB Server 10.2, 10.3, 10.4. In MariaDB 10.5,
MDEV-19534 made innodb_checksum_algorithm=full_crc32 the default.
It is even faster and more secure.
The default settings in MariaDB do allow old data files to be read,
no matter if a worse checksum algorithm had been used.
(Unfortunately, before innodb_checksum_algorithm=full_crc32,
the data files did not identify which checksum algorithm is being used.)
The non-default settings innodb_checksum_algorithm=strict_crc32 or
innodb_checksum_algorithm=strict_full_crc32 would only allow CRC-32C
checksums. The incompatibility with old data files is why they are
not the default.
The newest server not to support innodb_checksum_algorithm=crc32
were MySQL 5.5 and MariaDB 5.5. Both have reached their end of life.
A valid reason for using innodb_checksum_algorithm=innodb could have
been the ability to downgrade. If it is really needed, data files
can be converted with an older version of the innochecksum utility.
Because there is no good reason to allow data files to be written
with insecure checksums, we will reject those option values:
innodb_checksum_algorithm=none
innodb_checksum_algorithm=innodb
innodb_checksum_algorithm=strict_none
innodb_checksum_algorithm=strict_innodb
Furthermore, the following innochecksum options will be removed,
because only strict crc32 will be supported:
innochecksum --strict-check=crc32
innochecksum -C crc32
innochecksum --write=crc32
innochecksum -w crc32
If a user wishes to convert a data file to use a different checksum
(so that it might be used with the no-longer-supported
MySQL 5.5 or MariaDB 5.5, which do not support IMPORT TABLESPACE
nor system tablespace format changes that were made in MariaDB 10.3),
then the innochecksum tool from MariaDB 10.2, 10.3, 10.4, 10.5 or
MySQL 5.7 can be used.
Reviewed by: Thirunarayanan Balathandayuthapani
5 years ago MDEV-34830: LSN in the future is not being treated as serious corruption
The invariant of write-ahead logging is that before any change to a
page is written to the data file, the corresponding log record must
must first have been durably written.
In crash recovery, there were some sloppy checks for this. Let us
implement accurate checks and flag an inconsistency as a hard error,
so that we can avoid further corruption of a corrupted database.
For data extraction from the corrupted database, innodb_force_recovery
can be used.
Before recovery is reading any data pages or invoking
buf_dblwr_t::recover() to recover torn pages from the
doublewrite buffer, InnoDB will have parsed the log until the
final LSN and updated log_sys.lsn to that. So, we can rely on
log_sys.lsn at all times. The doublewrite buffer recovery has been
refactored in such a way that the recv_sys.dblwr.pages may be consulted
while discovering files and their page sizes, but nothing will be
written back to data files before buf_dblwr_t::recover() is invoked.
recv_max_page_lsn, recv_lsn_checks_on: Remove.
recv_sys_t::validate_checkpoint(): Validate the write-ahead-logging
condition at the end of the recovery.
recv_dblwr_t::validate_page(): Keep track of the maximum LSN
(if we are checking a non-doublewrite copy of a page) but
do not complain LSN being in the future. The doublewrite buffer
is a special case, because it will be read early during recovery.
Besides, starting with commit 762bcb81b5bf9bbde61fed59afb26417f4ce1e86
the dblwr=true copies of pages may legitimately be "too new".
recv_dblwr_t::find_page(): Find a valid page with the smallest
FIL_PAGE_LSN that is in the valid range for recovery.
recv_dblwr_t::restore_first_page(): Replaced by find_page().
Only buf_dblwr_t::recover() will write to data files.
buf_dblwr_t::recover(): Simplify the message output. Do attempt
doublewrite recovery on user page read error. Ignore doublewrite
pages whose FIL_PAGE_LSN is outside the usable bounds. Previously,
we could wrongly recover a too new page from the doublewrite buffer.
It is unlikely that this could have lead to an actual error.
Write back all recovered pages from the doublewrite buffer here,
including for the first page of any tablespace.
buf_page_is_corrupted(): Distinguish the return values
CORRUPTED_FUTURE_LSN and CORRUPTED_OTHER.
buf_page_check_corrupt(): Return the error code DB_CORRUPTION
in case the LSN is in the future.
Datafile::read_first_page_flags(): Split from read_first_page().
Take a copy of the first page as a parameter.
recv_sys_t::free_corrupted_page(): Take the file as a parameter
and return whether a message was displayed. This avoids some duplicated
and incomplete error messages.
buf_page_t::read_complete(): Remove some redundant output and always
display the name of the corrupted file. Never return DB_FAIL;
use it only in internal error handling.
IORequest::read_complete(): Assume that buf_page_t::read_complete()
will have reported any error.
fil_space_t::set_corrupted(): Return whether this is the first time
the tablespace had been flagged as corrupted.
Datafile::validate_first_page(), fil_node_open_file_low(),
fil_node_open_file(), fil_space_t::read_page0(),
fil_node_t::read_page0(): Add a parameter for a copy of the
first page, and a parameter to indicate whether the FIL_PAGE_LSN
check should be suppressed. Before buf_dblwr_t::recover() is
invoked, we cannot validate the FIL_PAGE_LSN, but we can trust the
FSP_SPACE_FLAGS and the tablespace ID that may be present in a
potentially too new copy of a page.
Reviewed by: Debarun Banerjee
12 months ago MDEV-25105 Remove innodb_checksum_algorithm values none,innodb,...
Historically, InnoDB supported a buggy page checksum algorithm that did not
compute a checksum over the full page. Later, well before MySQL 4.1
introduced .ibd files and the innodb_file_per_table option, the algorithm
was corrected and the first 4 bytes of each page were redefined to be
a checksum.
The original checksum was so slow that an option to disable page checksum
was introduced for benchmarketing purposes.
The Intel Nehalem microarchitecture introduced the SSE4.2 instruction set
extension, which includes instructions for faster computation of CRC-32C.
In MySQL 5.6 (and MariaDB 10.0), innodb_checksum_algorithm=crc32 was
implemented to make of that. As that option was changed to be the default
in MySQL 5.7, a bug was found on big-endian platforms and some work-around
code was added to weaken that checksum further. MariaDB disables that
work-around by default since MDEV-17958.
Later, SIMD-accelerated CRC-32C has been implemented in MariaDB for POWER
and ARM and also for IA-32/AMD64, making use of carry-less multiplication
where available.
Long story short, innodb_checksum_algorithm=crc32 is faster and more secure
than the pre-MySQL 5.6 checksum, called innodb_checksum_algorithm=innodb.
It should have removed any need to use innodb_checksum_algorithm=none.
The setting innodb_checksum_algorithm=crc32 is the default in
MySQL 5.7 and MariaDB Server 10.2, 10.3, 10.4. In MariaDB 10.5,
MDEV-19534 made innodb_checksum_algorithm=full_crc32 the default.
It is even faster and more secure.
The default settings in MariaDB do allow old data files to be read,
no matter if a worse checksum algorithm had been used.
(Unfortunately, before innodb_checksum_algorithm=full_crc32,
the data files did not identify which checksum algorithm is being used.)
The non-default settings innodb_checksum_algorithm=strict_crc32 or
innodb_checksum_algorithm=strict_full_crc32 would only allow CRC-32C
checksums. The incompatibility with old data files is why they are
not the default.
The newest server not to support innodb_checksum_algorithm=crc32
were MySQL 5.5 and MariaDB 5.5. Both have reached their end of life.
A valid reason for using innodb_checksum_algorithm=innodb could have
been the ability to downgrade. If it is really needed, data files
can be converted with an older version of the innochecksum utility.
Because there is no good reason to allow data files to be written
with insecure checksums, we will reject those option values:
innodb_checksum_algorithm=none
innodb_checksum_algorithm=innodb
innodb_checksum_algorithm=strict_none
innodb_checksum_algorithm=strict_innodb
Furthermore, the following innochecksum options will be removed,
because only strict crc32 will be supported:
innochecksum --strict-check=crc32
innochecksum -C crc32
innochecksum --write=crc32
innochecksum -w crc32
If a user wishes to convert a data file to use a different checksum
(so that it might be used with the no-longer-supported
MySQL 5.5 or MariaDB 5.5, which do not support IMPORT TABLESPACE
nor system tablespace format changes that were made in MariaDB 10.3),
then the innochecksum tool from MariaDB 10.2, 10.3, 10.4, 10.5 or
MySQL 5.7 can be used.
Reviewed by: Thirunarayanan Balathandayuthapani
5 years ago MDEV-34830: LSN in the future is not being treated as serious corruption
The invariant of write-ahead logging is that before any change to a
page is written to the data file, the corresponding log record must
must first have been durably written.
In crash recovery, there were some sloppy checks for this. Let us
implement accurate checks and flag an inconsistency as a hard error,
so that we can avoid further corruption of a corrupted database.
For data extraction from the corrupted database, innodb_force_recovery
can be used.
Before recovery is reading any data pages or invoking
buf_dblwr_t::recover() to recover torn pages from the
doublewrite buffer, InnoDB will have parsed the log until the
final LSN and updated log_sys.lsn to that. So, we can rely on
log_sys.lsn at all times. The doublewrite buffer recovery has been
refactored in such a way that the recv_sys.dblwr.pages may be consulted
while discovering files and their page sizes, but nothing will be
written back to data files before buf_dblwr_t::recover() is invoked.
recv_max_page_lsn, recv_lsn_checks_on: Remove.
recv_sys_t::validate_checkpoint(): Validate the write-ahead-logging
condition at the end of the recovery.
recv_dblwr_t::validate_page(): Keep track of the maximum LSN
(if we are checking a non-doublewrite copy of a page) but
do not complain LSN being in the future. The doublewrite buffer
is a special case, because it will be read early during recovery.
Besides, starting with commit 762bcb81b5bf9bbde61fed59afb26417f4ce1e86
the dblwr=true copies of pages may legitimately be "too new".
recv_dblwr_t::find_page(): Find a valid page with the smallest
FIL_PAGE_LSN that is in the valid range for recovery.
recv_dblwr_t::restore_first_page(): Replaced by find_page().
Only buf_dblwr_t::recover() will write to data files.
buf_dblwr_t::recover(): Simplify the message output. Do attempt
doublewrite recovery on user page read error. Ignore doublewrite
pages whose FIL_PAGE_LSN is outside the usable bounds. Previously,
we could wrongly recover a too new page from the doublewrite buffer.
It is unlikely that this could have lead to an actual error.
Write back all recovered pages from the doublewrite buffer here,
including for the first page of any tablespace.
buf_page_is_corrupted(): Distinguish the return values
CORRUPTED_FUTURE_LSN and CORRUPTED_OTHER.
buf_page_check_corrupt(): Return the error code DB_CORRUPTION
in case the LSN is in the future.
Datafile::read_first_page_flags(): Split from read_first_page().
Take a copy of the first page as a parameter.
recv_sys_t::free_corrupted_page(): Take the file as a parameter
and return whether a message was displayed. This avoids some duplicated
and incomplete error messages.
buf_page_t::read_complete(): Remove some redundant output and always
display the name of the corrupted file. Never return DB_FAIL;
use it only in internal error handling.
IORequest::read_complete(): Assume that buf_page_t::read_complete()
will have reported any error.
fil_space_t::set_corrupted(): Return whether this is the first time
the tablespace had been flagged as corrupted.
Datafile::validate_first_page(), fil_node_open_file_low(),
fil_node_open_file(), fil_space_t::read_page0(),
fil_node_t::read_page0(): Add a parameter for a copy of the
first page, and a parameter to indicate whether the FIL_PAGE_LSN
check should be suppressed. Before buf_dblwr_t::recover() is
invoked, we cannot validate the FIL_PAGE_LSN, but we can trust the
FSP_SPACE_FLAGS and the tablespace ID that may be present in a
potentially too new copy of a page.
Reviewed by: Debarun Banerjee
12 months ago MDEV-25105 Remove innodb_checksum_algorithm values none,innodb,...
Historically, InnoDB supported a buggy page checksum algorithm that did not
compute a checksum over the full page. Later, well before MySQL 4.1
introduced .ibd files and the innodb_file_per_table option, the algorithm
was corrected and the first 4 bytes of each page were redefined to be
a checksum.
The original checksum was so slow that an option to disable page checksum
was introduced for benchmarketing purposes.
The Intel Nehalem microarchitecture introduced the SSE4.2 instruction set
extension, which includes instructions for faster computation of CRC-32C.
In MySQL 5.6 (and MariaDB 10.0), innodb_checksum_algorithm=crc32 was
implemented to make of that. As that option was changed to be the default
in MySQL 5.7, a bug was found on big-endian platforms and some work-around
code was added to weaken that checksum further. MariaDB disables that
work-around by default since MDEV-17958.
Later, SIMD-accelerated CRC-32C has been implemented in MariaDB for POWER
and ARM and also for IA-32/AMD64, making use of carry-less multiplication
where available.
Long story short, innodb_checksum_algorithm=crc32 is faster and more secure
than the pre-MySQL 5.6 checksum, called innodb_checksum_algorithm=innodb.
It should have removed any need to use innodb_checksum_algorithm=none.
The setting innodb_checksum_algorithm=crc32 is the default in
MySQL 5.7 and MariaDB Server 10.2, 10.3, 10.4. In MariaDB 10.5,
MDEV-19534 made innodb_checksum_algorithm=full_crc32 the default.
It is even faster and more secure.
The default settings in MariaDB do allow old data files to be read,
no matter if a worse checksum algorithm had been used.
(Unfortunately, before innodb_checksum_algorithm=full_crc32,
the data files did not identify which checksum algorithm is being used.)
The non-default settings innodb_checksum_algorithm=strict_crc32 or
innodb_checksum_algorithm=strict_full_crc32 would only allow CRC-32C
checksums. The incompatibility with old data files is why they are
not the default.
The newest server not to support innodb_checksum_algorithm=crc32
were MySQL 5.5 and MariaDB 5.5. Both have reached their end of life.
A valid reason for using innodb_checksum_algorithm=innodb could have
been the ability to downgrade. If it is really needed, data files
can be converted with an older version of the innochecksum utility.
Because there is no good reason to allow data files to be written
with insecure checksums, we will reject those option values:
innodb_checksum_algorithm=none
innodb_checksum_algorithm=innodb
innodb_checksum_algorithm=strict_none
innodb_checksum_algorithm=strict_innodb
Furthermore, the following innochecksum options will be removed,
because only strict crc32 will be supported:
innochecksum --strict-check=crc32
innochecksum -C crc32
innochecksum --write=crc32
innochecksum -w crc32
If a user wishes to convert a data file to use a different checksum
(so that it might be used with the no-longer-supported
MySQL 5.5 or MariaDB 5.5, which do not support IMPORT TABLESPACE
nor system tablespace format changes that were made in MariaDB 10.3),
then the innochecksum tool from MariaDB 10.2, 10.3, 10.4, 10.5 or
MySQL 5.7 can be used.
Reviewed by: Thirunarayanan Balathandayuthapani
5 years ago MDEV-34830: LSN in the future is not being treated as serious corruption
The invariant of write-ahead logging is that before any change to a
page is written to the data file, the corresponding log record must
must first have been durably written.
In crash recovery, there were some sloppy checks for this. Let us
implement accurate checks and flag an inconsistency as a hard error,
so that we can avoid further corruption of a corrupted database.
For data extraction from the corrupted database, innodb_force_recovery
can be used.
Before recovery is reading any data pages or invoking
buf_dblwr_t::recover() to recover torn pages from the
doublewrite buffer, InnoDB will have parsed the log until the
final LSN and updated log_sys.lsn to that. So, we can rely on
log_sys.lsn at all times. The doublewrite buffer recovery has been
refactored in such a way that the recv_sys.dblwr.pages may be consulted
while discovering files and their page sizes, but nothing will be
written back to data files before buf_dblwr_t::recover() is invoked.
recv_max_page_lsn, recv_lsn_checks_on: Remove.
recv_sys_t::validate_checkpoint(): Validate the write-ahead-logging
condition at the end of the recovery.
recv_dblwr_t::validate_page(): Keep track of the maximum LSN
(if we are checking a non-doublewrite copy of a page) but
do not complain LSN being in the future. The doublewrite buffer
is a special case, because it will be read early during recovery.
Besides, starting with commit 762bcb81b5bf9bbde61fed59afb26417f4ce1e86
the dblwr=true copies of pages may legitimately be "too new".
recv_dblwr_t::find_page(): Find a valid page with the smallest
FIL_PAGE_LSN that is in the valid range for recovery.
recv_dblwr_t::restore_first_page(): Replaced by find_page().
Only buf_dblwr_t::recover() will write to data files.
buf_dblwr_t::recover(): Simplify the message output. Do attempt
doublewrite recovery on user page read error. Ignore doublewrite
pages whose FIL_PAGE_LSN is outside the usable bounds. Previously,
we could wrongly recover a too new page from the doublewrite buffer.
It is unlikely that this could have lead to an actual error.
Write back all recovered pages from the doublewrite buffer here,
including for the first page of any tablespace.
buf_page_is_corrupted(): Distinguish the return values
CORRUPTED_FUTURE_LSN and CORRUPTED_OTHER.
buf_page_check_corrupt(): Return the error code DB_CORRUPTION
in case the LSN is in the future.
Datafile::read_first_page_flags(): Split from read_first_page().
Take a copy of the first page as a parameter.
recv_sys_t::free_corrupted_page(): Take the file as a parameter
and return whether a message was displayed. This avoids some duplicated
and incomplete error messages.
buf_page_t::read_complete(): Remove some redundant output and always
display the name of the corrupted file. Never return DB_FAIL;
use it only in internal error handling.
IORequest::read_complete(): Assume that buf_page_t::read_complete()
will have reported any error.
fil_space_t::set_corrupted(): Return whether this is the first time
the tablespace had been flagged as corrupted.
Datafile::validate_first_page(), fil_node_open_file_low(),
fil_node_open_file(), fil_space_t::read_page0(),
fil_node_t::read_page0(): Add a parameter for a copy of the
first page, and a parameter to indicate whether the FIL_PAGE_LSN
check should be suppressed. Before buf_dblwr_t::recover() is
invoked, we cannot validate the FIL_PAGE_LSN, but we can trust the
FSP_SPACE_FLAGS and the tablespace ID that may be present in a
potentially too new copy of a page.
Reviewed by: Debarun Banerjee
12 months ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-14425 Improve the redo log for concurrency
The InnoDB redo log used to be formatted in blocks of 512 bytes.
The log blocks were encrypted and the checksum was calculated while
holding log_sys.mutex, creating a serious scalability bottleneck.
We remove the fixed-size redo log block structure altogether and
essentially turn every mini-transaction into a log block of its own.
This allows encryption and checksum calculations to be performed
on local mtr_t::m_log buffers, before acquiring log_sys.mutex.
The mutex only protects a memcpy() of the data to the shared
log_sys.buf, as well as the padding of the log, in case the
to-be-written part of the log would not end in a block boundary of
the underlying storage. For now, the "padding" consists of writing
a single NUL byte, to allow recovery and mariadb-backup to detect
the end of the circular log faster.
Like the previous implementation, we will overwrite the last log block
over and over again, until it has been completely filled. It would be
possible to write only up to the last completed block (if no more
recent write was requested), or to write dummy FILE_CHECKPOINT records
to fill the incomplete block, by invoking the currently disabled
function log_pad(). This would require adjustments to some logic around
log checkpoints, page flushing, and shutdown.
An upgrade after a crash of any previous version is not supported.
Logically empty log files from a previous version will be upgraded.
An attempt to start up InnoDB without a valid ib_logfile0 will be
refused. Previously, the redo log used to be created automatically
if it was missing. Only with with innodb_force_recovery=6, it is
possible to start InnoDB in read-only mode even if the log file
does not exist. This allows the contents of a possibly corrupted
database to be dumped.
Because a prepared backup from an earlier version of mariadb-backup
will create a 0-sized log file, we will allow an upgrade from such
log files, provided that the FIL_PAGE_FILE_FLUSH_LSN in the system
tablespace looks valid.
The 512-byte log checkpoint blocks at 0x200 and 0x600 will be replaced
with 64-byte log checkpoint blocks at 0x1000 and 0x2000.
The start of log records will move from 0x800 to 0x3000. This allows us
to use 4096-byte aligned blocks for all I/O in a future revision.
We extend the MDEV-12353 redo log record format as follows.
(1) Empty mini-transactions or extra NUL bytes will not be allowed.
(2) The end-of-minitransaction marker (a NUL byte) will be replaced
with a 1-bit sequence number, which will be toggled each time when the
circular log file wraps back to the beginning.
(3) After the sequence bit, a CRC-32C checksum of all data
(excluding the sequence bit) will written.
(4) If the log is encrypted, 8 bytes will be written before
the checksum and included in it. This is part of the
initialization vector (IV) of encrypted log data.
(5) File names, page numbers, and checkpoint information will not be
encrypted. Only the payload bytes of page-level log will be encrypted.
The tablespace ID and page number will form part of the IV.
(6) For padding, arbitrary-length FILE_CHECKPOINT records may be written,
with all-zero payload, and with the normal end marker and checksum.
The minimum size is 7 bytes, or 7+8 with innodb_encrypt_log=ON.
In mariadb-backup and in Galera snapshot transfer (SST) scripts, we will
no longer remove ib_logfile0 or create an empty ib_logfile0. Server startup
will require a valid log file. When resizing the log, we will create
a logically empty ib_logfile101 at the current LSN and use an atomic rename
to replace ib_logfile0 with it. See the test innodb.log_file_size.
Because there is no mandatory padding in the log file, we are able
to create a dummy log file as of an arbitrary log sequence number.
See the test mariabackup.huge_lsn.
The parameter innodb_log_write_ahead_size and the
INFORMATION_SCHEMA.INNODB_METRICS counter log_padded will be removed.
The minimum value of innodb_log_buffer_size will be increased to 2MiB
(because log_sys.buf will replace recv_sys.buf) and the increment
adjusted to 4096 bytes (the maximum log block size).
The following INFORMATION_SCHEMA.INNODB_METRICS counters will be removed:
os_log_fsyncs
os_log_pending_fsyncs
log_pending_log_flushes
log_pending_checkpoint_writes
The following status variables will be removed:
Innodb_os_log_fsyncs (this is included in Innodb_data_fsyncs)
Innodb_os_log_pending_fsyncs (this was limited to at most 1 by design)
log_sys.get_block_size(): Return the physical block size of the log file.
This is only implemented on Linux and Microsoft Windows for now, and for
the power-of-2 block sizes between 64 and 4096 bytes (the minimum and
maximum size of a checkpoint block). If the block size is anything else,
the traditional 512-byte size will be used via normal file system
buffering.
If the file system buffers can be bypassed, a message like the following
will be issued:
InnoDB: File system buffers for log disabled (block size=512 bytes)
InnoDB: File system buffers for log disabled (block size=4096 bytes)
This has been tested on Linux and Microsoft Windows with both sizes.
On Linux, only enable O_DIRECT on the log for innodb_flush_method=O_DSYNC.
Tests in 3 different environments where the log is stored in a device
with a physical block size of 512 bytes are yielding better throughput
without O_DIRECT. This could be due to the fact that in the event the
last log block is being overwritten (if multiple transactions would
become durable at the same time, and each of will write a small
number of bytes to the last log block), it should be faster to re-copy
data from log_sys.buf or log_sys.flush_buf to the kernel buffer,
to be finally written at fdatasync() time.
The parameter innodb_flush_method=O_DSYNC will imply O_DIRECT for
data files. This option will enable O_DIRECT on the log file on Linux.
It may be unsafe to use when the storage device does not support
FUA (Force Unit Access) mode.
When the server is compiled WITH_PMEM=ON, we will use memory-mapped
I/O for the log file if the log resides on a "mount -o dax" device.
We will identify PMEM in a start-up message:
InnoDB: log sequence number 0 (memory-mapped); transaction id 3
On Linux, we will also invoke mmap() on any ib_logfile0 that resides
in /dev/shm, effectively treating the log file as persistent memory.
This should speed up "./mtr --mem" and increase the test coverage of
PMEM on non-PMEM hardware. It also allows users to estimate how much
the performance would be improved by installing persistent memory.
On other tmpfs file systems such as /run, we will not use mmap().
mariadb-backup: Eliminated several variables. We will refer
directly to recv_sys and log_sys.
backup_wait_for_lsn(): Detect non-progress of
xtrabackup_copy_logfile(). In this new log format with
arbitrary-sized blocks, we can only detect log file overrun
indirectly, by observing that the scanned log sequence number
is not advancing.
xtrabackup_copy_logfile(): On PMEM, do not modify the sequence bit,
because we are not allowed to modify the server's log file, and our
memory mapping is read-only.
trx_flush_log_if_needed_low(): Do not use the callback on pmem.
Using neither flush_lock nor write_lock around PMEM writes seems
to yield the best performance. The pmem_persist() calls may
still be somewhat slower than the pwrite() and fdatasync() based
interface (PMEM mounted without -o dax).
recv_sys_t::buf: Remove. We will use log_sys.buf for parsing.
recv_sys_t::MTR_SIZE_MAX: Replaces RECV_SCAN_SIZE.
recv_sys_t::file_checkpoint: Renamed from mlog_checkpoint_lsn.
recv_sys_t, log_sys_t: Removed many data members.
recv_sys.lsn: Renamed from recv_sys.recovered_lsn.
recv_sys.offset: Renamed from recv_sys.recovered_offset.
log_sys.buf_size: Replaces srv_log_buffer_size.
recv_buf: A smart pointer that wraps log_sys.buf[recv_sys.offset]
when the buffer is being allocated from the memory heap.
recv_ring: A smart pointer that wraps a circular log_sys.buf[] that is
backed by ib_logfile0. The pointer will wrap from recv_sys.len
(log_sys.file_size) to log_sys.START_OFFSET. For the record that
wraps around, we may copy file name or record payload data to
the auxiliary buffer decrypt_buf in order to have a contiguous
block of memory. The maximum size of a record is less than
innodb_page_size bytes.
recv_sys_t::parse(): Take the smart pointer as a template parameter.
Do not temporarily add a trailing NUL byte to FILE_ records, because
we are not supposed to modify the memory-mapped log file. (It is
attached in read-write mode already during recovery.)
recv_sys_t::parse_mtr(): Wrapper for recv_sys_t::parse().
recv_sys_t::parse_pmem(): Like parse_mtr(), but if PREMATURE_EOF would be
returned on PMEM, use recv_ring to wrap around the buffer to the start.
mtr_t::finish_write(), log_close(): Do not enforce log_sys.max_buf_free
on PMEM, because it has no meaning on the mmap-based log.
log_sys.write_to_buf: Count writes to log_sys.buf. Replaces
srv_stats.log_write_requests and export_vars.innodb_log_write_requests.
Protected by log_sys.mutex. Updated consistently in log_close().
Previously, mtr_t::commit() conditionally updated the count,
which was inconsistent.
log_sys.write_to_log: Count swaps of log_sys.buf and log_sys.flush_buf,
for writing to log_sys.log (the ib_logfile0). Replaces
srv_stats.log_writes and export_vars.innodb_log_writes.
Protected by log_sys.mutex.
log_sys.waits: Count waits in append_prepare(). Replaces
srv_stats.log_waits and export_vars.innodb_log_waits.
recv_recover_page(): Do not unnecessarily acquire
log_sys.flush_order_mutex. We are inserting the blocks in arbitary
order anyway, to be adjusted in recv_sys.apply(true).
We will change the definition of flush_lock and write_lock to
avoid potential false sharing. Depending on sizeof(log_sys) and
CPU_LEVEL1_DCACHE_LINESIZE, the flush_lock and write_lock could
share a cache line with each other or with the last data members
of log_sys.
Thanks to Matthias Leich for providing https://rr-project.org traces
for various failures during the development, and to
Thirunarayanan Balathandayuthapani for his help in debugging
some of the recovery code. And thanks to the developers of the
rr debugger for a tool without which extensive changes to InnoDB
would be very challenging to get right.
Thanks to Vladislav Vaintroub for useful feedback and
to him, Axel Schwenke and Krunal Bauskar for testing the performance.
4 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-23399: Performance regression with write workloads
The buffer pool refactoring in MDEV-15053 and MDEV-22871 shifted
the performance bottleneck to the page flushing.
The configuration parameters will be changed as follows:
innodb_lru_flush_size=32 (new: how many pages to flush on LRU eviction)
innodb_lru_scan_depth=1536 (old: 1024)
innodb_max_dirty_pages_pct=90 (old: 75)
innodb_max_dirty_pages_pct_lwm=75 (old: 0)
Note: The parameter innodb_lru_scan_depth will only affect LRU
eviction of buffer pool pages when a new page is being allocated. The
page cleaner thread will no longer evict any pages. It used to
guarantee that some pages will remain free in the buffer pool. Now, we
perform that eviction 'on demand' in buf_LRU_get_free_block().
The parameter innodb_lru_scan_depth(srv_LRU_scan_depth) is used as follows:
* When the buffer pool is being shrunk in buf_pool_t::withdraw_blocks()
* As a buf_pool.free limit in buf_LRU_list_batch() for terminating
the flushing that is initiated e.g., by buf_LRU_get_free_block()
The parameter also used to serve as an initial limit for unzip_LRU
eviction (evicting uncompressed page frames while retaining
ROW_FORMAT=COMPRESSED pages), but now we will use a hard-coded limit
of 100 or unlimited for invoking buf_LRU_scan_and_free_block().
The status variables will be changed as follows:
innodb_buffer_pool_pages_flushed: This includes also the count of
innodb_buffer_pool_pages_LRU_flushed and should work reliably,
updated one by one in buf_flush_page() to give more real-time
statistics. The function buf_flush_stats(), which we are removing,
was not called in every code path. For both counters, we will use
regular variables that are incremented in a critical section of
buf_pool.mutex. Note that show_innodb_vars() directly links to the
variables, and reads of the counters will *not* be protected by
buf_pool.mutex, so you cannot get a consistent snapshot of both variables.
The following INFORMATION_SCHEMA.INNODB_METRICS counters will be
removed, because the page cleaner no longer deals with writing or
evicting least recently used pages, and because the single-page writes
have been removed:
* buffer_LRU_batch_flush_avg_time_slot
* buffer_LRU_batch_flush_avg_time_thread
* buffer_LRU_batch_flush_avg_time_est
* buffer_LRU_batch_flush_avg_pass
* buffer_LRU_single_flush_scanned
* buffer_LRU_single_flush_num_scan
* buffer_LRU_single_flush_scanned_per_call
When moving to a single buffer pool instance in MDEV-15058, we missed
some opportunity to simplify the buf_flush_page_cleaner thread. It was
unnecessarily using a mutex and some complex data structures, even
though we always have a single page cleaner thread.
Furthermore, the buf_flush_page_cleaner thread had separate 'recovery'
and 'shutdown' modes where it was waiting to be triggered by some
other thread, adding unnecessary latency and potential for hangs in
relatively rarely executed startup or shutdown code.
The page cleaner was also running two kinds of batches in an
interleaved fashion: "LRU flush" (writing out some least recently used
pages and evicting them on write completion) and the normal batches
that aim to increase the MIN(oldest_modification) in the buffer pool,
to help the log checkpoint advance.
The buf_pool.flush_list flushing was being blocked by
buf_block_t::lock for no good reason. Furthermore, if the FIL_PAGE_LSN
of a page is ahead of log_sys.get_flushed_lsn(), that is, what has
been persistently written to the redo log, we would trigger a log
flush and then resume the page flushing. This would unnecessarily
limit the performance of the page cleaner thread and trigger the
infamous messages "InnoDB: page_cleaner: 1000ms intended loop took 4450ms.
The settings might not be optimal" that were suppressed in
commit d1ab89037a518fcffbc50c24e4bd94e4ec33aed0 unless log_warnings>2.
Our revised algorithm will make log_sys.get_flushed_lsn() advance at
the start of buf_flush_lists(), and then execute a 'best effort' to
write out all pages. The flush batches will skip pages that were modified
since the log was written, or are are currently exclusively locked.
The MDEV-13670 message "page_cleaner: 1000ms intended loop took" message
will be removed, because by design, the buf_flush_page_cleaner() should
not be blocked during a batch for extended periods of time.
We will remove the single-page flushing altogether. Related to this,
the debug parameter innodb_doublewrite_batch_size will be removed,
because all of the doublewrite buffer will be used for flushing
batches. If a page needs to be evicted from the buffer pool and all
100 least recently used pages in the buffer pool have unflushed
changes, buf_LRU_get_free_block() will execute buf_flush_lists() to
write out and evict innodb_lru_flush_size pages. At most one thread
will execute buf_flush_lists() in buf_LRU_get_free_block(); other
threads will wait for that LRU flushing batch to finish.
To improve concurrency, we will replace the InnoDB ib_mutex_t and
os_event_t native mutexes and condition variables in this area of code.
Most notably, this means that the buffer pool mutex (buf_pool.mutex)
is no longer instrumented via any InnoDB interfaces. It will continue
to be instrumented via PERFORMANCE_SCHEMA.
For now, both buf_pool.flush_list_mutex and buf_pool.mutex will be
declared with MY_MUTEX_INIT_FAST (PTHREAD_MUTEX_ADAPTIVE_NP). The critical
sections of buf_pool.flush_list_mutex should be shorter than those for
buf_pool.mutex, because in the worst case, they cover a linear scan of
buf_pool.flush_list, while the worst case of a critical section of
buf_pool.mutex covers a linear scan of the potentially much longer
buf_pool.LRU list.
mysql_mutex_is_owner(), safe_mutex_is_owner(): New predicate, usable
with SAFE_MUTEX. Some InnoDB debug assertions need this predicate
instead of mysql_mutex_assert_owner() or mysql_mutex_assert_not_owner().
buf_pool_t::n_flush_LRU, buf_pool_t::n_flush_list:
Replaces buf_pool_t::init_flush[] and buf_pool_t::n_flush[].
The number of active flush operations.
buf_pool_t::mutex, buf_pool_t::flush_list_mutex: Use mysql_mutex_t
instead of ib_mutex_t, to have native mutexes with PERFORMANCE_SCHEMA
and SAFE_MUTEX instrumentation.
buf_pool_t::done_flush_LRU: Condition variable for !n_flush_LRU.
buf_pool_t::done_flush_list: Condition variable for !n_flush_list.
buf_pool_t::do_flush_list: Condition variable to wake up the
buf_flush_page_cleaner when a log checkpoint needs to be written
or the server is being shut down. Replaces buf_flush_event.
We will keep using timed waits (the page cleaner thread will wake
_at least_ once per second), because the calculations for
innodb_adaptive_flushing depend on fixed time intervals.
buf_dblwr: Allocate statically, and move all code to member functions.
Use a native mutex and condition variable. Remove code to deal with
single-page flushing.
buf_dblwr_check_block(): Make the check debug-only. We were spending
a significant amount of execution time in page_simple_validate_new().
flush_counters_t::unzip_LRU_evicted: Remove.
IORequest: Make more members const. FIXME: m_fil_node should be removed.
buf_flush_sync_lsn: Protect by std::atomic, not page_cleaner.mutex
(which we are removing).
page_cleaner_slot_t, page_cleaner_t: Remove many redundant members.
pc_request_flush_slot(): Replaces pc_request() and pc_flush_slot().
recv_writer_thread: Remove. Recovery works just fine without it, if we
simply invoke buf_flush_sync() at the end of each batch in
recv_sys_t::apply().
recv_recovery_from_checkpoint_finish(): Remove. We can simply call
recv_sys.debug_free() directly.
srv_started_redo: Replaces srv_start_state.
SRV_SHUTDOWN_FLUSH_PHASE: Remove. logs_empty_and_mark_files_at_shutdown()
can communicate with the normal page cleaner loop via the new function
flush_buffer_pool().
buf_flush_remove(): Assert that the calling thread is holding
buf_pool.flush_list_mutex. This removes unnecessary mutex operations
from buf_flush_remove_pages() and buf_flush_dirty_pages(),
which replace buf_LRU_flush_or_remove_pages().
buf_flush_lists(): Renamed from buf_flush_batch(), with simplified
interface. Return the number of flushed pages. Clarified comments and
renamed min_n to max_n. Identify LRU batch by lsn=0. Merge all the functions
buf_flush_start(), buf_flush_batch(), buf_flush_end() directly to this
function, which was their only caller, and remove 2 unnecessary
buf_pool.mutex release/re-acquisition that we used to perform around
the buf_flush_batch() call. At the start, if not all log has been
durably written, wait for a background task to do it, or start a new
task to do it. This allows the log write to run concurrently with our
page flushing batch. Any pages that were skipped due to too recent
FIL_PAGE_LSN or due to them being latched by a writer should be flushed
during the next batch, unless there are further modifications to those
pages. It is possible that a page that we must flush due to small
oldest_modification also carries a recent FIL_PAGE_LSN or is being
constantly modified. In the worst case, all writers would then end up
waiting in log_free_check() to allow the flushing and the checkpoint
to complete.
buf_do_flush_list_batch(): Clarify comments, and rename min_n to max_n.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_flush_space(): Auxiliary function to look up a tablespace for
page flushing.
buf_flush_page(): Defer the computation of space->full_crc32(). Never
call log_write_up_to(), but instead skip persistent pages whose latest
modification (FIL_PAGE_LSN) is newer than the redo log. Also skip
pages on which we cannot acquire a shared latch without waiting.
buf_flush_try_neighbors(): Do not bother checking buf_fix_count
because buf_flush_page() will no longer wait for the page latch.
Take the tablespace as a parameter, and only execute this function
when innodb_flush_neighbors>0. Avoid repeated calls of page_id_t::fold().
buf_flush_relocate_on_flush_list(): Declare as cold, and push down
a condition from the callers.
buf_flush_check_neighbor(): Take id.fold() as a parameter.
buf_flush_sync(): Ensure that the buf_pool.flush_list is empty,
because the flushing batch will skip pages whose modifications have
not yet been written to the log or were latched for modification.
buf_free_from_unzip_LRU_list_batch(): Remove redundant local variables.
buf_flush_LRU_list_batch(): Let the caller buf_do_LRU_batch() initialize
the counters, and report n->evicted.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_do_LRU_batch(): Return the number of pages flushed.
buf_LRU_free_page(): Only release and re-acquire buf_pool.mutex if
adaptive hash index entries are pointing to the block.
buf_LRU_get_free_block(): Do not wake up the page cleaner, because it
will no longer perform any useful work for us, and we do not want it
to compete for I/O while buf_flush_lists(innodb_lru_flush_size, 0)
writes out and evicts at most innodb_lru_flush_size pages. (The
function buf_do_LRU_batch() may complete after writing fewer pages if
more than innodb_lru_scan_depth pages end up in buf_pool.free list.)
Eliminate some mutex release-acquire cycles, and wait for the LRU
flush batch to complete before rescanning.
buf_LRU_check_size_of_non_data_objects(): Simplify the code.
buf_page_write_complete(): Remove the parameter evict, and always
evict pages that were part of an LRU flush.
buf_page_create(): Take a pre-allocated page as a parameter.
buf_pool_t::free_block(): Free a pre-allocated block.
recv_sys_t::recover_low(), recv_sys_t::apply(): Preallocate the block
while not holding recv_sys.mutex. During page allocation, we may
initiate a page flush, which in turn may initiate a log flush, which
would require acquiring log_sys.mutex, which should always be acquired
before recv_sys.mutex in order to avoid deadlocks. Therefore, we must
not be holding recv_sys.mutex while allocating a buffer pool block.
BtrBulk::logFreeCheck(): Skip a redundant condition.
row_undo_step(): Do not invoke srv_inc_activity_count() for every row
that is being rolled back. It should suffice to invoke the function in
trx_flush_log_if_needed() during trx_t::commit_in_memory() when the
rollback completes.
sync_check_enable(): Remove. We will enable innodb_sync_debug from the
very beginning.
Reviewed by: Vladislav Vaintroub
5 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-23399: Performance regression with write workloads
The buffer pool refactoring in MDEV-15053 and MDEV-22871 shifted
the performance bottleneck to the page flushing.
The configuration parameters will be changed as follows:
innodb_lru_flush_size=32 (new: how many pages to flush on LRU eviction)
innodb_lru_scan_depth=1536 (old: 1024)
innodb_max_dirty_pages_pct=90 (old: 75)
innodb_max_dirty_pages_pct_lwm=75 (old: 0)
Note: The parameter innodb_lru_scan_depth will only affect LRU
eviction of buffer pool pages when a new page is being allocated. The
page cleaner thread will no longer evict any pages. It used to
guarantee that some pages will remain free in the buffer pool. Now, we
perform that eviction 'on demand' in buf_LRU_get_free_block().
The parameter innodb_lru_scan_depth(srv_LRU_scan_depth) is used as follows:
* When the buffer pool is being shrunk in buf_pool_t::withdraw_blocks()
* As a buf_pool.free limit in buf_LRU_list_batch() for terminating
the flushing that is initiated e.g., by buf_LRU_get_free_block()
The parameter also used to serve as an initial limit for unzip_LRU
eviction (evicting uncompressed page frames while retaining
ROW_FORMAT=COMPRESSED pages), but now we will use a hard-coded limit
of 100 or unlimited for invoking buf_LRU_scan_and_free_block().
The status variables will be changed as follows:
innodb_buffer_pool_pages_flushed: This includes also the count of
innodb_buffer_pool_pages_LRU_flushed and should work reliably,
updated one by one in buf_flush_page() to give more real-time
statistics. The function buf_flush_stats(), which we are removing,
was not called in every code path. For both counters, we will use
regular variables that are incremented in a critical section of
buf_pool.mutex. Note that show_innodb_vars() directly links to the
variables, and reads of the counters will *not* be protected by
buf_pool.mutex, so you cannot get a consistent snapshot of both variables.
The following INFORMATION_SCHEMA.INNODB_METRICS counters will be
removed, because the page cleaner no longer deals with writing or
evicting least recently used pages, and because the single-page writes
have been removed:
* buffer_LRU_batch_flush_avg_time_slot
* buffer_LRU_batch_flush_avg_time_thread
* buffer_LRU_batch_flush_avg_time_est
* buffer_LRU_batch_flush_avg_pass
* buffer_LRU_single_flush_scanned
* buffer_LRU_single_flush_num_scan
* buffer_LRU_single_flush_scanned_per_call
When moving to a single buffer pool instance in MDEV-15058, we missed
some opportunity to simplify the buf_flush_page_cleaner thread. It was
unnecessarily using a mutex and some complex data structures, even
though we always have a single page cleaner thread.
Furthermore, the buf_flush_page_cleaner thread had separate 'recovery'
and 'shutdown' modes where it was waiting to be triggered by some
other thread, adding unnecessary latency and potential for hangs in
relatively rarely executed startup or shutdown code.
The page cleaner was also running two kinds of batches in an
interleaved fashion: "LRU flush" (writing out some least recently used
pages and evicting them on write completion) and the normal batches
that aim to increase the MIN(oldest_modification) in the buffer pool,
to help the log checkpoint advance.
The buf_pool.flush_list flushing was being blocked by
buf_block_t::lock for no good reason. Furthermore, if the FIL_PAGE_LSN
of a page is ahead of log_sys.get_flushed_lsn(), that is, what has
been persistently written to the redo log, we would trigger a log
flush and then resume the page flushing. This would unnecessarily
limit the performance of the page cleaner thread and trigger the
infamous messages "InnoDB: page_cleaner: 1000ms intended loop took 4450ms.
The settings might not be optimal" that were suppressed in
commit d1ab89037a518fcffbc50c24e4bd94e4ec33aed0 unless log_warnings>2.
Our revised algorithm will make log_sys.get_flushed_lsn() advance at
the start of buf_flush_lists(), and then execute a 'best effort' to
write out all pages. The flush batches will skip pages that were modified
since the log was written, or are are currently exclusively locked.
The MDEV-13670 message "page_cleaner: 1000ms intended loop took" message
will be removed, because by design, the buf_flush_page_cleaner() should
not be blocked during a batch for extended periods of time.
We will remove the single-page flushing altogether. Related to this,
the debug parameter innodb_doublewrite_batch_size will be removed,
because all of the doublewrite buffer will be used for flushing
batches. If a page needs to be evicted from the buffer pool and all
100 least recently used pages in the buffer pool have unflushed
changes, buf_LRU_get_free_block() will execute buf_flush_lists() to
write out and evict innodb_lru_flush_size pages. At most one thread
will execute buf_flush_lists() in buf_LRU_get_free_block(); other
threads will wait for that LRU flushing batch to finish.
To improve concurrency, we will replace the InnoDB ib_mutex_t and
os_event_t native mutexes and condition variables in this area of code.
Most notably, this means that the buffer pool mutex (buf_pool.mutex)
is no longer instrumented via any InnoDB interfaces. It will continue
to be instrumented via PERFORMANCE_SCHEMA.
For now, both buf_pool.flush_list_mutex and buf_pool.mutex will be
declared with MY_MUTEX_INIT_FAST (PTHREAD_MUTEX_ADAPTIVE_NP). The critical
sections of buf_pool.flush_list_mutex should be shorter than those for
buf_pool.mutex, because in the worst case, they cover a linear scan of
buf_pool.flush_list, while the worst case of a critical section of
buf_pool.mutex covers a linear scan of the potentially much longer
buf_pool.LRU list.
mysql_mutex_is_owner(), safe_mutex_is_owner(): New predicate, usable
with SAFE_MUTEX. Some InnoDB debug assertions need this predicate
instead of mysql_mutex_assert_owner() or mysql_mutex_assert_not_owner().
buf_pool_t::n_flush_LRU, buf_pool_t::n_flush_list:
Replaces buf_pool_t::init_flush[] and buf_pool_t::n_flush[].
The number of active flush operations.
buf_pool_t::mutex, buf_pool_t::flush_list_mutex: Use mysql_mutex_t
instead of ib_mutex_t, to have native mutexes with PERFORMANCE_SCHEMA
and SAFE_MUTEX instrumentation.
buf_pool_t::done_flush_LRU: Condition variable for !n_flush_LRU.
buf_pool_t::done_flush_list: Condition variable for !n_flush_list.
buf_pool_t::do_flush_list: Condition variable to wake up the
buf_flush_page_cleaner when a log checkpoint needs to be written
or the server is being shut down. Replaces buf_flush_event.
We will keep using timed waits (the page cleaner thread will wake
_at least_ once per second), because the calculations for
innodb_adaptive_flushing depend on fixed time intervals.
buf_dblwr: Allocate statically, and move all code to member functions.
Use a native mutex and condition variable. Remove code to deal with
single-page flushing.
buf_dblwr_check_block(): Make the check debug-only. We were spending
a significant amount of execution time in page_simple_validate_new().
flush_counters_t::unzip_LRU_evicted: Remove.
IORequest: Make more members const. FIXME: m_fil_node should be removed.
buf_flush_sync_lsn: Protect by std::atomic, not page_cleaner.mutex
(which we are removing).
page_cleaner_slot_t, page_cleaner_t: Remove many redundant members.
pc_request_flush_slot(): Replaces pc_request() and pc_flush_slot().
recv_writer_thread: Remove. Recovery works just fine without it, if we
simply invoke buf_flush_sync() at the end of each batch in
recv_sys_t::apply().
recv_recovery_from_checkpoint_finish(): Remove. We can simply call
recv_sys.debug_free() directly.
srv_started_redo: Replaces srv_start_state.
SRV_SHUTDOWN_FLUSH_PHASE: Remove. logs_empty_and_mark_files_at_shutdown()
can communicate with the normal page cleaner loop via the new function
flush_buffer_pool().
buf_flush_remove(): Assert that the calling thread is holding
buf_pool.flush_list_mutex. This removes unnecessary mutex operations
from buf_flush_remove_pages() and buf_flush_dirty_pages(),
which replace buf_LRU_flush_or_remove_pages().
buf_flush_lists(): Renamed from buf_flush_batch(), with simplified
interface. Return the number of flushed pages. Clarified comments and
renamed min_n to max_n. Identify LRU batch by lsn=0. Merge all the functions
buf_flush_start(), buf_flush_batch(), buf_flush_end() directly to this
function, which was their only caller, and remove 2 unnecessary
buf_pool.mutex release/re-acquisition that we used to perform around
the buf_flush_batch() call. At the start, if not all log has been
durably written, wait for a background task to do it, or start a new
task to do it. This allows the log write to run concurrently with our
page flushing batch. Any pages that were skipped due to too recent
FIL_PAGE_LSN or due to them being latched by a writer should be flushed
during the next batch, unless there are further modifications to those
pages. It is possible that a page that we must flush due to small
oldest_modification also carries a recent FIL_PAGE_LSN or is being
constantly modified. In the worst case, all writers would then end up
waiting in log_free_check() to allow the flushing and the checkpoint
to complete.
buf_do_flush_list_batch(): Clarify comments, and rename min_n to max_n.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_flush_space(): Auxiliary function to look up a tablespace for
page flushing.
buf_flush_page(): Defer the computation of space->full_crc32(). Never
call log_write_up_to(), but instead skip persistent pages whose latest
modification (FIL_PAGE_LSN) is newer than the redo log. Also skip
pages on which we cannot acquire a shared latch without waiting.
buf_flush_try_neighbors(): Do not bother checking buf_fix_count
because buf_flush_page() will no longer wait for the page latch.
Take the tablespace as a parameter, and only execute this function
when innodb_flush_neighbors>0. Avoid repeated calls of page_id_t::fold().
buf_flush_relocate_on_flush_list(): Declare as cold, and push down
a condition from the callers.
buf_flush_check_neighbor(): Take id.fold() as a parameter.
buf_flush_sync(): Ensure that the buf_pool.flush_list is empty,
because the flushing batch will skip pages whose modifications have
not yet been written to the log or were latched for modification.
buf_free_from_unzip_LRU_list_batch(): Remove redundant local variables.
buf_flush_LRU_list_batch(): Let the caller buf_do_LRU_batch() initialize
the counters, and report n->evicted.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_do_LRU_batch(): Return the number of pages flushed.
buf_LRU_free_page(): Only release and re-acquire buf_pool.mutex if
adaptive hash index entries are pointing to the block.
buf_LRU_get_free_block(): Do not wake up the page cleaner, because it
will no longer perform any useful work for us, and we do not want it
to compete for I/O while buf_flush_lists(innodb_lru_flush_size, 0)
writes out and evicts at most innodb_lru_flush_size pages. (The
function buf_do_LRU_batch() may complete after writing fewer pages if
more than innodb_lru_scan_depth pages end up in buf_pool.free list.)
Eliminate some mutex release-acquire cycles, and wait for the LRU
flush batch to complete before rescanning.
buf_LRU_check_size_of_non_data_objects(): Simplify the code.
buf_page_write_complete(): Remove the parameter evict, and always
evict pages that were part of an LRU flush.
buf_page_create(): Take a pre-allocated page as a parameter.
buf_pool_t::free_block(): Free a pre-allocated block.
recv_sys_t::recover_low(), recv_sys_t::apply(): Preallocate the block
while not holding recv_sys.mutex. During page allocation, we may
initiate a page flush, which in turn may initiate a log flush, which
would require acquiring log_sys.mutex, which should always be acquired
before recv_sys.mutex in order to avoid deadlocks. Therefore, we must
not be holding recv_sys.mutex while allocating a buffer pool block.
BtrBulk::logFreeCheck(): Skip a redundant condition.
row_undo_step(): Do not invoke srv_inc_activity_count() for every row
that is being rolled back. It should suffice to invoke the function in
trx_flush_log_if_needed() during trx_t::commit_in_memory() when the
rollback completes.
sync_check_enable(): Remove. We will enable innodb_sync_debug from the
very beginning.
Reviewed by: Vladislav Vaintroub
5 years ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-23399: Performance regression with write workloads
The buffer pool refactoring in MDEV-15053 and MDEV-22871 shifted
the performance bottleneck to the page flushing.
The configuration parameters will be changed as follows:
innodb_lru_flush_size=32 (new: how many pages to flush on LRU eviction)
innodb_lru_scan_depth=1536 (old: 1024)
innodb_max_dirty_pages_pct=90 (old: 75)
innodb_max_dirty_pages_pct_lwm=75 (old: 0)
Note: The parameter innodb_lru_scan_depth will only affect LRU
eviction of buffer pool pages when a new page is being allocated. The
page cleaner thread will no longer evict any pages. It used to
guarantee that some pages will remain free in the buffer pool. Now, we
perform that eviction 'on demand' in buf_LRU_get_free_block().
The parameter innodb_lru_scan_depth(srv_LRU_scan_depth) is used as follows:
* When the buffer pool is being shrunk in buf_pool_t::withdraw_blocks()
* As a buf_pool.free limit in buf_LRU_list_batch() for terminating
the flushing that is initiated e.g., by buf_LRU_get_free_block()
The parameter also used to serve as an initial limit for unzip_LRU
eviction (evicting uncompressed page frames while retaining
ROW_FORMAT=COMPRESSED pages), but now we will use a hard-coded limit
of 100 or unlimited for invoking buf_LRU_scan_and_free_block().
The status variables will be changed as follows:
innodb_buffer_pool_pages_flushed: This includes also the count of
innodb_buffer_pool_pages_LRU_flushed and should work reliably,
updated one by one in buf_flush_page() to give more real-time
statistics. The function buf_flush_stats(), which we are removing,
was not called in every code path. For both counters, we will use
regular variables that are incremented in a critical section of
buf_pool.mutex. Note that show_innodb_vars() directly links to the
variables, and reads of the counters will *not* be protected by
buf_pool.mutex, so you cannot get a consistent snapshot of both variables.
The following INFORMATION_SCHEMA.INNODB_METRICS counters will be
removed, because the page cleaner no longer deals with writing or
evicting least recently used pages, and because the single-page writes
have been removed:
* buffer_LRU_batch_flush_avg_time_slot
* buffer_LRU_batch_flush_avg_time_thread
* buffer_LRU_batch_flush_avg_time_est
* buffer_LRU_batch_flush_avg_pass
* buffer_LRU_single_flush_scanned
* buffer_LRU_single_flush_num_scan
* buffer_LRU_single_flush_scanned_per_call
When moving to a single buffer pool instance in MDEV-15058, we missed
some opportunity to simplify the buf_flush_page_cleaner thread. It was
unnecessarily using a mutex and some complex data structures, even
though we always have a single page cleaner thread.
Furthermore, the buf_flush_page_cleaner thread had separate 'recovery'
and 'shutdown' modes where it was waiting to be triggered by some
other thread, adding unnecessary latency and potential for hangs in
relatively rarely executed startup or shutdown code.
The page cleaner was also running two kinds of batches in an
interleaved fashion: "LRU flush" (writing out some least recently used
pages and evicting them on write completion) and the normal batches
that aim to increase the MIN(oldest_modification) in the buffer pool,
to help the log checkpoint advance.
The buf_pool.flush_list flushing was being blocked by
buf_block_t::lock for no good reason. Furthermore, if the FIL_PAGE_LSN
of a page is ahead of log_sys.get_flushed_lsn(), that is, what has
been persistently written to the redo log, we would trigger a log
flush and then resume the page flushing. This would unnecessarily
limit the performance of the page cleaner thread and trigger the
infamous messages "InnoDB: page_cleaner: 1000ms intended loop took 4450ms.
The settings might not be optimal" that were suppressed in
commit d1ab89037a518fcffbc50c24e4bd94e4ec33aed0 unless log_warnings>2.
Our revised algorithm will make log_sys.get_flushed_lsn() advance at
the start of buf_flush_lists(), and then execute a 'best effort' to
write out all pages. The flush batches will skip pages that were modified
since the log was written, or are are currently exclusively locked.
The MDEV-13670 message "page_cleaner: 1000ms intended loop took" message
will be removed, because by design, the buf_flush_page_cleaner() should
not be blocked during a batch for extended periods of time.
We will remove the single-page flushing altogether. Related to this,
the debug parameter innodb_doublewrite_batch_size will be removed,
because all of the doublewrite buffer will be used for flushing
batches. If a page needs to be evicted from the buffer pool and all
100 least recently used pages in the buffer pool have unflushed
changes, buf_LRU_get_free_block() will execute buf_flush_lists() to
write out and evict innodb_lru_flush_size pages. At most one thread
will execute buf_flush_lists() in buf_LRU_get_free_block(); other
threads will wait for that LRU flushing batch to finish.
To improve concurrency, we will replace the InnoDB ib_mutex_t and
os_event_t native mutexes and condition variables in this area of code.
Most notably, this means that the buffer pool mutex (buf_pool.mutex)
is no longer instrumented via any InnoDB interfaces. It will continue
to be instrumented via PERFORMANCE_SCHEMA.
For now, both buf_pool.flush_list_mutex and buf_pool.mutex will be
declared with MY_MUTEX_INIT_FAST (PTHREAD_MUTEX_ADAPTIVE_NP). The critical
sections of buf_pool.flush_list_mutex should be shorter than those for
buf_pool.mutex, because in the worst case, they cover a linear scan of
buf_pool.flush_list, while the worst case of a critical section of
buf_pool.mutex covers a linear scan of the potentially much longer
buf_pool.LRU list.
mysql_mutex_is_owner(), safe_mutex_is_owner(): New predicate, usable
with SAFE_MUTEX. Some InnoDB debug assertions need this predicate
instead of mysql_mutex_assert_owner() or mysql_mutex_assert_not_owner().
buf_pool_t::n_flush_LRU, buf_pool_t::n_flush_list:
Replaces buf_pool_t::init_flush[] and buf_pool_t::n_flush[].
The number of active flush operations.
buf_pool_t::mutex, buf_pool_t::flush_list_mutex: Use mysql_mutex_t
instead of ib_mutex_t, to have native mutexes with PERFORMANCE_SCHEMA
and SAFE_MUTEX instrumentation.
buf_pool_t::done_flush_LRU: Condition variable for !n_flush_LRU.
buf_pool_t::done_flush_list: Condition variable for !n_flush_list.
buf_pool_t::do_flush_list: Condition variable to wake up the
buf_flush_page_cleaner when a log checkpoint needs to be written
or the server is being shut down. Replaces buf_flush_event.
We will keep using timed waits (the page cleaner thread will wake
_at least_ once per second), because the calculations for
innodb_adaptive_flushing depend on fixed time intervals.
buf_dblwr: Allocate statically, and move all code to member functions.
Use a native mutex and condition variable. Remove code to deal with
single-page flushing.
buf_dblwr_check_block(): Make the check debug-only. We were spending
a significant amount of execution time in page_simple_validate_new().
flush_counters_t::unzip_LRU_evicted: Remove.
IORequest: Make more members const. FIXME: m_fil_node should be removed.
buf_flush_sync_lsn: Protect by std::atomic, not page_cleaner.mutex
(which we are removing).
page_cleaner_slot_t, page_cleaner_t: Remove many redundant members.
pc_request_flush_slot(): Replaces pc_request() and pc_flush_slot().
recv_writer_thread: Remove. Recovery works just fine without it, if we
simply invoke buf_flush_sync() at the end of each batch in
recv_sys_t::apply().
recv_recovery_from_checkpoint_finish(): Remove. We can simply call
recv_sys.debug_free() directly.
srv_started_redo: Replaces srv_start_state.
SRV_SHUTDOWN_FLUSH_PHASE: Remove. logs_empty_and_mark_files_at_shutdown()
can communicate with the normal page cleaner loop via the new function
flush_buffer_pool().
buf_flush_remove(): Assert that the calling thread is holding
buf_pool.flush_list_mutex. This removes unnecessary mutex operations
from buf_flush_remove_pages() and buf_flush_dirty_pages(),
which replace buf_LRU_flush_or_remove_pages().
buf_flush_lists(): Renamed from buf_flush_batch(), with simplified
interface. Return the number of flushed pages. Clarified comments and
renamed min_n to max_n. Identify LRU batch by lsn=0. Merge all the functions
buf_flush_start(), buf_flush_batch(), buf_flush_end() directly to this
function, which was their only caller, and remove 2 unnecessary
buf_pool.mutex release/re-acquisition that we used to perform around
the buf_flush_batch() call. At the start, if not all log has been
durably written, wait for a background task to do it, or start a new
task to do it. This allows the log write to run concurrently with our
page flushing batch. Any pages that were skipped due to too recent
FIL_PAGE_LSN or due to them being latched by a writer should be flushed
during the next batch, unless there are further modifications to those
pages. It is possible that a page that we must flush due to small
oldest_modification also carries a recent FIL_PAGE_LSN or is being
constantly modified. In the worst case, all writers would then end up
waiting in log_free_check() to allow the flushing and the checkpoint
to complete.
buf_do_flush_list_batch(): Clarify comments, and rename min_n to max_n.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_flush_space(): Auxiliary function to look up a tablespace for
page flushing.
buf_flush_page(): Defer the computation of space->full_crc32(). Never
call log_write_up_to(), but instead skip persistent pages whose latest
modification (FIL_PAGE_LSN) is newer than the redo log. Also skip
pages on which we cannot acquire a shared latch without waiting.
buf_flush_try_neighbors(): Do not bother checking buf_fix_count
because buf_flush_page() will no longer wait for the page latch.
Take the tablespace as a parameter, and only execute this function
when innodb_flush_neighbors>0. Avoid repeated calls of page_id_t::fold().
buf_flush_relocate_on_flush_list(): Declare as cold, and push down
a condition from the callers.
buf_flush_check_neighbor(): Take id.fold() as a parameter.
buf_flush_sync(): Ensure that the buf_pool.flush_list is empty,
because the flushing batch will skip pages whose modifications have
not yet been written to the log or were latched for modification.
buf_free_from_unzip_LRU_list_batch(): Remove redundant local variables.
buf_flush_LRU_list_batch(): Let the caller buf_do_LRU_batch() initialize
the counters, and report n->evicted.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_do_LRU_batch(): Return the number of pages flushed.
buf_LRU_free_page(): Only release and re-acquire buf_pool.mutex if
adaptive hash index entries are pointing to the block.
buf_LRU_get_free_block(): Do not wake up the page cleaner, because it
will no longer perform any useful work for us, and we do not want it
to compete for I/O while buf_flush_lists(innodb_lru_flush_size, 0)
writes out and evicts at most innodb_lru_flush_size pages. (The
function buf_do_LRU_batch() may complete after writing fewer pages if
more than innodb_lru_scan_depth pages end up in buf_pool.free list.)
Eliminate some mutex release-acquire cycles, and wait for the LRU
flush batch to complete before rescanning.
buf_LRU_check_size_of_non_data_objects(): Simplify the code.
buf_page_write_complete(): Remove the parameter evict, and always
evict pages that were part of an LRU flush.
buf_page_create(): Take a pre-allocated page as a parameter.
buf_pool_t::free_block(): Free a pre-allocated block.
recv_sys_t::recover_low(), recv_sys_t::apply(): Preallocate the block
while not holding recv_sys.mutex. During page allocation, we may
initiate a page flush, which in turn may initiate a log flush, which
would require acquiring log_sys.mutex, which should always be acquired
before recv_sys.mutex in order to avoid deadlocks. Therefore, we must
not be holding recv_sys.mutex while allocating a buffer pool block.
BtrBulk::logFreeCheck(): Skip a redundant condition.
row_undo_step(): Do not invoke srv_inc_activity_count() for every row
that is being rolled back. It should suffice to invoke the function in
trx_flush_log_if_needed() during trx_t::commit_in_memory() when the
rollback completes.
sync_check_enable(): Remove. We will enable innodb_sync_debug from the
very beginning.
Reviewed by: Vladislav Vaintroub
5 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-24671: Replace lock_wait_timeout_task with mysql_cond_timedwait()
lock_wait(): Replaces lock_wait_suspend_thread(). Wait for the lock to
be granted or the transaction to be killed using mysql_cond_timedwait()
or mysql_cond_wait().
lock_wait_end(): Replaces que_thr_end_lock_wait() and
lock_wait_release_thread_if_suspended().
lock_wait_timeout_task: Remove. The operating system kernel will
resume the mysql_cond_timedwait() in lock_wait(). An added benefit
is that innodb_lock_wait_timeout no longer has a 'jitter' of 1 second,
which was caused by this wake-up task waking up only once per second,
and then waking up any threads for which the timeout (which was only
measured in seconds) was exceeded.
innobase_kill_query(): Set trx->error_state=DB_INTERRUPTED,
so that a call trx_is_interrupted(trx) in lock_wait() can be avoided.
We will protect things more consistently with lock_sys.wait_mutex,
which will be moved below lock_sys.mutex in the latching order.
trx_lock_t::cond: Condition variable for !wait_lock, used with
lock_sys.wait_mutex.
srv_slot_t: Remove. Replaced by trx_lock_t::cond,
lock_grant_after_reset(): Merged to to lock_grant().
lock_rec_get_index_name(): Remove.
lock_sys_t: Introduce wait_pending, wait_count, wait_time, wait_time_max
that are protected by wait_mutex.
trx_lock_t::que_state: Remove.
que_thr_state_t: Remove QUE_THR_COMMAND_WAIT, QUE_THR_LOCK_WAIT.
que_thr_t: Remove is_active, start_running(), stop_no_error().
que_fork_t::n_active_thrs, trx_lock_t::n_active_thrs: Remove.
5 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-22871: Clean up hash_table_t
HASH_TABLE_SYNC_MUTEX was kind-of used for the adaptive hash index,
even though that hash table is already protected by btr_search_latches[].
HASH_TABLE_SYNC_RWLOCK was only being used for buf_pool.page_hash.
It is cleaner to decouple that synchronization from hash_table_t,
and move it to the actual user.
buf_pool_t::page_hash_latches[]: Synchronization for buf_pool.page_hash.
LATCH_ID_HASH_TABLE_MUTEX: Remove.
hash_table_t::sync_obj, hash_table_t::n_sync_obj: Remove.
hash_table_t::type, hash_table_sync_t: Remove.
HASH_ASSERT_OWN(), hash_get_mutex(), hash_get_nth_mutex(): Remove.
ib_recreate(): Merge to the only caller, buf_pool_resize_hash().
ib_create(): Merge to the callers.
ha_clear(): Merge to the only caller buf_pool_t::close().
buf_pool_t::create(): Merge the ib_create() and
hash_create_sync_obj() invocations.
ha_insert_for_fold_func(): Clarify an assertion.
buf_pool_t::page_hash_lock(): Simplify the logic.
hash_assert_can_search(), hash_assert_can_modify(): Remove.
These predicates were only being invoked for the adaptive hash index,
while they only are effective for buf_pool.page_hash.
HASH_DELETE_AND_COMPACT(): Merge to ha_delete_hash_node().
hash_get_sync_obj_index(): Remove.
hash_table_t::heaps[], hash_get_nth_heap(): Remove. It was actually unused!
hash_get_heap(): Remove. It was only used in ha_delete_hash_node(),
where we always use hash_table_t::heap.
hash_table_t::calc_hash(): Replaces hash_calc_hash().
5 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-31953 madvise(..., MADV_FREE) is causing a performance regression
buf_page_t::set_os_unused(): Remove the system call that had been added in
commit 16c9718758cb3bbff76672405d4ce1bce6da6c6f and revised in
commit c1fd082e9c7369f4511eb5a52e58cb15489caa74 for Microsoft Windows.
buf_pool_t::garbage_collect(): A new function to collect any garbage
from the InnoDB buffer pool that can be removed without writing any
log or data files. This will also invoke madvise() for all of buf_pool.free.
To trigger this the following MDEV is implemented:
MDEV-24670 avoid OOM by linux kernel co-operative memory management
To avoid frequent triggers that caused the MDEV-31953 regression, while
still preserving the 10.11 functionality of non-greedy kernel memory
usage, memory triggers are used.
On the triggering of memory pressure, if supported in the Linux kernel,
trigger the garbage collection of the innodb buffer pool.
The hard coded triggers occur where there is:
* some memory pressure in 5 of the last 10 seconds
* a full stall on memory pressure for 10ms in the last 2 seconds
The kernel will trigger only one in each of these time windows. To avoid
mariadb being in a constant state of memory garbage collection, this has
been limited to once per minute.
For a small set of kernels in 2023 (6.5, 6.6), there was a limit requiring
CAP_SYS_RESOURCE that was lifted[1] to support the use case of user
memory pressure. It not currently possible to set CAP_SYS_RESOURCES in
a systemd service as its setting a capability inside a usernamespace.
Running under systemd v254+ requires the default MemoryPressureWatch=auto
(or alternately "on").
Functionality was tested in a 6.4 kernel Fedora successfully under a
systemd service.
Running in a container requires that (unmask=)/sys/fs/cgroup be writable
by the mariadbd process.
To aid testing, the buf_pool_resize was a convient trigger point on
which to trigger garbage collection.
ref [1]: https://lore.kernel.org/all/CAMw=ZnQ56cm4Txgy5EhGYvR+Jt4s-KVgoA9_65HKWVMOXp7a9A@mail.gmail.com/T/#m3bd2a73c5ee49965cb73a830b1ccaa37ccf4e427
Co-Author: Daniel Black (on memory pressure trigger)
Reviewed by: Marko Mäkelä, Vladislav Vaintroub, Vladislav Lesin,
Thirunarayanan Balathandayuthapani
Tested by: Matthias Leich
2 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-23399: Performance regression with write workloads
The buffer pool refactoring in MDEV-15053 and MDEV-22871 shifted
the performance bottleneck to the page flushing.
The configuration parameters will be changed as follows:
innodb_lru_flush_size=32 (new: how many pages to flush on LRU eviction)
innodb_lru_scan_depth=1536 (old: 1024)
innodb_max_dirty_pages_pct=90 (old: 75)
innodb_max_dirty_pages_pct_lwm=75 (old: 0)
Note: The parameter innodb_lru_scan_depth will only affect LRU
eviction of buffer pool pages when a new page is being allocated. The
page cleaner thread will no longer evict any pages. It used to
guarantee that some pages will remain free in the buffer pool. Now, we
perform that eviction 'on demand' in buf_LRU_get_free_block().
The parameter innodb_lru_scan_depth(srv_LRU_scan_depth) is used as follows:
* When the buffer pool is being shrunk in buf_pool_t::withdraw_blocks()
* As a buf_pool.free limit in buf_LRU_list_batch() for terminating
the flushing that is initiated e.g., by buf_LRU_get_free_block()
The parameter also used to serve as an initial limit for unzip_LRU
eviction (evicting uncompressed page frames while retaining
ROW_FORMAT=COMPRESSED pages), but now we will use a hard-coded limit
of 100 or unlimited for invoking buf_LRU_scan_and_free_block().
The status variables will be changed as follows:
innodb_buffer_pool_pages_flushed: This includes also the count of
innodb_buffer_pool_pages_LRU_flushed and should work reliably,
updated one by one in buf_flush_page() to give more real-time
statistics. The function buf_flush_stats(), which we are removing,
was not called in every code path. For both counters, we will use
regular variables that are incremented in a critical section of
buf_pool.mutex. Note that show_innodb_vars() directly links to the
variables, and reads of the counters will *not* be protected by
buf_pool.mutex, so you cannot get a consistent snapshot of both variables.
The following INFORMATION_SCHEMA.INNODB_METRICS counters will be
removed, because the page cleaner no longer deals with writing or
evicting least recently used pages, and because the single-page writes
have been removed:
* buffer_LRU_batch_flush_avg_time_slot
* buffer_LRU_batch_flush_avg_time_thread
* buffer_LRU_batch_flush_avg_time_est
* buffer_LRU_batch_flush_avg_pass
* buffer_LRU_single_flush_scanned
* buffer_LRU_single_flush_num_scan
* buffer_LRU_single_flush_scanned_per_call
When moving to a single buffer pool instance in MDEV-15058, we missed
some opportunity to simplify the buf_flush_page_cleaner thread. It was
unnecessarily using a mutex and some complex data structures, even
though we always have a single page cleaner thread.
Furthermore, the buf_flush_page_cleaner thread had separate 'recovery'
and 'shutdown' modes where it was waiting to be triggered by some
other thread, adding unnecessary latency and potential for hangs in
relatively rarely executed startup or shutdown code.
The page cleaner was also running two kinds of batches in an
interleaved fashion: "LRU flush" (writing out some least recently used
pages and evicting them on write completion) and the normal batches
that aim to increase the MIN(oldest_modification) in the buffer pool,
to help the log checkpoint advance.
The buf_pool.flush_list flushing was being blocked by
buf_block_t::lock for no good reason. Furthermore, if the FIL_PAGE_LSN
of a page is ahead of log_sys.get_flushed_lsn(), that is, what has
been persistently written to the redo log, we would trigger a log
flush and then resume the page flushing. This would unnecessarily
limit the performance of the page cleaner thread and trigger the
infamous messages "InnoDB: page_cleaner: 1000ms intended loop took 4450ms.
The settings might not be optimal" that were suppressed in
commit d1ab89037a518fcffbc50c24e4bd94e4ec33aed0 unless log_warnings>2.
Our revised algorithm will make log_sys.get_flushed_lsn() advance at
the start of buf_flush_lists(), and then execute a 'best effort' to
write out all pages. The flush batches will skip pages that were modified
since the log was written, or are are currently exclusively locked.
The MDEV-13670 message "page_cleaner: 1000ms intended loop took" message
will be removed, because by design, the buf_flush_page_cleaner() should
not be blocked during a batch for extended periods of time.
We will remove the single-page flushing altogether. Related to this,
the debug parameter innodb_doublewrite_batch_size will be removed,
because all of the doublewrite buffer will be used for flushing
batches. If a page needs to be evicted from the buffer pool and all
100 least recently used pages in the buffer pool have unflushed
changes, buf_LRU_get_free_block() will execute buf_flush_lists() to
write out and evict innodb_lru_flush_size pages. At most one thread
will execute buf_flush_lists() in buf_LRU_get_free_block(); other
threads will wait for that LRU flushing batch to finish.
To improve concurrency, we will replace the InnoDB ib_mutex_t and
os_event_t native mutexes and condition variables in this area of code.
Most notably, this means that the buffer pool mutex (buf_pool.mutex)
is no longer instrumented via any InnoDB interfaces. It will continue
to be instrumented via PERFORMANCE_SCHEMA.
For now, both buf_pool.flush_list_mutex and buf_pool.mutex will be
declared with MY_MUTEX_INIT_FAST (PTHREAD_MUTEX_ADAPTIVE_NP). The critical
sections of buf_pool.flush_list_mutex should be shorter than those for
buf_pool.mutex, because in the worst case, they cover a linear scan of
buf_pool.flush_list, while the worst case of a critical section of
buf_pool.mutex covers a linear scan of the potentially much longer
buf_pool.LRU list.
mysql_mutex_is_owner(), safe_mutex_is_owner(): New predicate, usable
with SAFE_MUTEX. Some InnoDB debug assertions need this predicate
instead of mysql_mutex_assert_owner() or mysql_mutex_assert_not_owner().
buf_pool_t::n_flush_LRU, buf_pool_t::n_flush_list:
Replaces buf_pool_t::init_flush[] and buf_pool_t::n_flush[].
The number of active flush operations.
buf_pool_t::mutex, buf_pool_t::flush_list_mutex: Use mysql_mutex_t
instead of ib_mutex_t, to have native mutexes with PERFORMANCE_SCHEMA
and SAFE_MUTEX instrumentation.
buf_pool_t::done_flush_LRU: Condition variable for !n_flush_LRU.
buf_pool_t::done_flush_list: Condition variable for !n_flush_list.
buf_pool_t::do_flush_list: Condition variable to wake up the
buf_flush_page_cleaner when a log checkpoint needs to be written
or the server is being shut down. Replaces buf_flush_event.
We will keep using timed waits (the page cleaner thread will wake
_at least_ once per second), because the calculations for
innodb_adaptive_flushing depend on fixed time intervals.
buf_dblwr: Allocate statically, and move all code to member functions.
Use a native mutex and condition variable. Remove code to deal with
single-page flushing.
buf_dblwr_check_block(): Make the check debug-only. We were spending
a significant amount of execution time in page_simple_validate_new().
flush_counters_t::unzip_LRU_evicted: Remove.
IORequest: Make more members const. FIXME: m_fil_node should be removed.
buf_flush_sync_lsn: Protect by std::atomic, not page_cleaner.mutex
(which we are removing).
page_cleaner_slot_t, page_cleaner_t: Remove many redundant members.
pc_request_flush_slot(): Replaces pc_request() and pc_flush_slot().
recv_writer_thread: Remove. Recovery works just fine without it, if we
simply invoke buf_flush_sync() at the end of each batch in
recv_sys_t::apply().
recv_recovery_from_checkpoint_finish(): Remove. We can simply call
recv_sys.debug_free() directly.
srv_started_redo: Replaces srv_start_state.
SRV_SHUTDOWN_FLUSH_PHASE: Remove. logs_empty_and_mark_files_at_shutdown()
can communicate with the normal page cleaner loop via the new function
flush_buffer_pool().
buf_flush_remove(): Assert that the calling thread is holding
buf_pool.flush_list_mutex. This removes unnecessary mutex operations
from buf_flush_remove_pages() and buf_flush_dirty_pages(),
which replace buf_LRU_flush_or_remove_pages().
buf_flush_lists(): Renamed from buf_flush_batch(), with simplified
interface. Return the number of flushed pages. Clarified comments and
renamed min_n to max_n. Identify LRU batch by lsn=0. Merge all the functions
buf_flush_start(), buf_flush_batch(), buf_flush_end() directly to this
function, which was their only caller, and remove 2 unnecessary
buf_pool.mutex release/re-acquisition that we used to perform around
the buf_flush_batch() call. At the start, if not all log has been
durably written, wait for a background task to do it, or start a new
task to do it. This allows the log write to run concurrently with our
page flushing batch. Any pages that were skipped due to too recent
FIL_PAGE_LSN or due to them being latched by a writer should be flushed
during the next batch, unless there are further modifications to those
pages. It is possible that a page that we must flush due to small
oldest_modification also carries a recent FIL_PAGE_LSN or is being
constantly modified. In the worst case, all writers would then end up
waiting in log_free_check() to allow the flushing and the checkpoint
to complete.
buf_do_flush_list_batch(): Clarify comments, and rename min_n to max_n.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_flush_space(): Auxiliary function to look up a tablespace for
page flushing.
buf_flush_page(): Defer the computation of space->full_crc32(). Never
call log_write_up_to(), but instead skip persistent pages whose latest
modification (FIL_PAGE_LSN) is newer than the redo log. Also skip
pages on which we cannot acquire a shared latch without waiting.
buf_flush_try_neighbors(): Do not bother checking buf_fix_count
because buf_flush_page() will no longer wait for the page latch.
Take the tablespace as a parameter, and only execute this function
when innodb_flush_neighbors>0. Avoid repeated calls of page_id_t::fold().
buf_flush_relocate_on_flush_list(): Declare as cold, and push down
a condition from the callers.
buf_flush_check_neighbor(): Take id.fold() as a parameter.
buf_flush_sync(): Ensure that the buf_pool.flush_list is empty,
because the flushing batch will skip pages whose modifications have
not yet been written to the log or were latched for modification.
buf_free_from_unzip_LRU_list_batch(): Remove redundant local variables.
buf_flush_LRU_list_batch(): Let the caller buf_do_LRU_batch() initialize
the counters, and report n->evicted.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_do_LRU_batch(): Return the number of pages flushed.
buf_LRU_free_page(): Only release and re-acquire buf_pool.mutex if
adaptive hash index entries are pointing to the block.
buf_LRU_get_free_block(): Do not wake up the page cleaner, because it
will no longer perform any useful work for us, and we do not want it
to compete for I/O while buf_flush_lists(innodb_lru_flush_size, 0)
writes out and evicts at most innodb_lru_flush_size pages. (The
function buf_do_LRU_batch() may complete after writing fewer pages if
more than innodb_lru_scan_depth pages end up in buf_pool.free list.)
Eliminate some mutex release-acquire cycles, and wait for the LRU
flush batch to complete before rescanning.
buf_LRU_check_size_of_non_data_objects(): Simplify the code.
buf_page_write_complete(): Remove the parameter evict, and always
evict pages that were part of an LRU flush.
buf_page_create(): Take a pre-allocated page as a parameter.
buf_pool_t::free_block(): Free a pre-allocated block.
recv_sys_t::recover_low(), recv_sys_t::apply(): Preallocate the block
while not holding recv_sys.mutex. During page allocation, we may
initiate a page flush, which in turn may initiate a log flush, which
would require acquiring log_sys.mutex, which should always be acquired
before recv_sys.mutex in order to avoid deadlocks. Therefore, we must
not be holding recv_sys.mutex while allocating a buffer pool block.
BtrBulk::logFreeCheck(): Skip a redundant condition.
row_undo_step(): Do not invoke srv_inc_activity_count() for every row
that is being rolled back. It should suffice to invoke the function in
trx_flush_log_if_needed() during trx_t::commit_in_memory() when the
rollback completes.
sync_check_enable(): Remove. We will enable innodb_sync_debug from the
very beginning.
Reviewed by: Vladislav Vaintroub
5 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-23399: Performance regression with write workloads
The buffer pool refactoring in MDEV-15053 and MDEV-22871 shifted
the performance bottleneck to the page flushing.
The configuration parameters will be changed as follows:
innodb_lru_flush_size=32 (new: how many pages to flush on LRU eviction)
innodb_lru_scan_depth=1536 (old: 1024)
innodb_max_dirty_pages_pct=90 (old: 75)
innodb_max_dirty_pages_pct_lwm=75 (old: 0)
Note: The parameter innodb_lru_scan_depth will only affect LRU
eviction of buffer pool pages when a new page is being allocated. The
page cleaner thread will no longer evict any pages. It used to
guarantee that some pages will remain free in the buffer pool. Now, we
perform that eviction 'on demand' in buf_LRU_get_free_block().
The parameter innodb_lru_scan_depth(srv_LRU_scan_depth) is used as follows:
* When the buffer pool is being shrunk in buf_pool_t::withdraw_blocks()
* As a buf_pool.free limit in buf_LRU_list_batch() for terminating
the flushing that is initiated e.g., by buf_LRU_get_free_block()
The parameter also used to serve as an initial limit for unzip_LRU
eviction (evicting uncompressed page frames while retaining
ROW_FORMAT=COMPRESSED pages), but now we will use a hard-coded limit
of 100 or unlimited for invoking buf_LRU_scan_and_free_block().
The status variables will be changed as follows:
innodb_buffer_pool_pages_flushed: This includes also the count of
innodb_buffer_pool_pages_LRU_flushed and should work reliably,
updated one by one in buf_flush_page() to give more real-time
statistics. The function buf_flush_stats(), which we are removing,
was not called in every code path. For both counters, we will use
regular variables that are incremented in a critical section of
buf_pool.mutex. Note that show_innodb_vars() directly links to the
variables, and reads of the counters will *not* be protected by
buf_pool.mutex, so you cannot get a consistent snapshot of both variables.
The following INFORMATION_SCHEMA.INNODB_METRICS counters will be
removed, because the page cleaner no longer deals with writing or
evicting least recently used pages, and because the single-page writes
have been removed:
* buffer_LRU_batch_flush_avg_time_slot
* buffer_LRU_batch_flush_avg_time_thread
* buffer_LRU_batch_flush_avg_time_est
* buffer_LRU_batch_flush_avg_pass
* buffer_LRU_single_flush_scanned
* buffer_LRU_single_flush_num_scan
* buffer_LRU_single_flush_scanned_per_call
When moving to a single buffer pool instance in MDEV-15058, we missed
some opportunity to simplify the buf_flush_page_cleaner thread. It was
unnecessarily using a mutex and some complex data structures, even
though we always have a single page cleaner thread.
Furthermore, the buf_flush_page_cleaner thread had separate 'recovery'
and 'shutdown' modes where it was waiting to be triggered by some
other thread, adding unnecessary latency and potential for hangs in
relatively rarely executed startup or shutdown code.
The page cleaner was also running two kinds of batches in an
interleaved fashion: "LRU flush" (writing out some least recently used
pages and evicting them on write completion) and the normal batches
that aim to increase the MIN(oldest_modification) in the buffer pool,
to help the log checkpoint advance.
The buf_pool.flush_list flushing was being blocked by
buf_block_t::lock for no good reason. Furthermore, if the FIL_PAGE_LSN
of a page is ahead of log_sys.get_flushed_lsn(), that is, what has
been persistently written to the redo log, we would trigger a log
flush and then resume the page flushing. This would unnecessarily
limit the performance of the page cleaner thread and trigger the
infamous messages "InnoDB: page_cleaner: 1000ms intended loop took 4450ms.
The settings might not be optimal" that were suppressed in
commit d1ab89037a518fcffbc50c24e4bd94e4ec33aed0 unless log_warnings>2.
Our revised algorithm will make log_sys.get_flushed_lsn() advance at
the start of buf_flush_lists(), and then execute a 'best effort' to
write out all pages. The flush batches will skip pages that were modified
since the log was written, or are are currently exclusively locked.
The MDEV-13670 message "page_cleaner: 1000ms intended loop took" message
will be removed, because by design, the buf_flush_page_cleaner() should
not be blocked during a batch for extended periods of time.
We will remove the single-page flushing altogether. Related to this,
the debug parameter innodb_doublewrite_batch_size will be removed,
because all of the doublewrite buffer will be used for flushing
batches. If a page needs to be evicted from the buffer pool and all
100 least recently used pages in the buffer pool have unflushed
changes, buf_LRU_get_free_block() will execute buf_flush_lists() to
write out and evict innodb_lru_flush_size pages. At most one thread
will execute buf_flush_lists() in buf_LRU_get_free_block(); other
threads will wait for that LRU flushing batch to finish.
To improve concurrency, we will replace the InnoDB ib_mutex_t and
os_event_t native mutexes and condition variables in this area of code.
Most notably, this means that the buffer pool mutex (buf_pool.mutex)
is no longer instrumented via any InnoDB interfaces. It will continue
to be instrumented via PERFORMANCE_SCHEMA.
For now, both buf_pool.flush_list_mutex and buf_pool.mutex will be
declared with MY_MUTEX_INIT_FAST (PTHREAD_MUTEX_ADAPTIVE_NP). The critical
sections of buf_pool.flush_list_mutex should be shorter than those for
buf_pool.mutex, because in the worst case, they cover a linear scan of
buf_pool.flush_list, while the worst case of a critical section of
buf_pool.mutex covers a linear scan of the potentially much longer
buf_pool.LRU list.
mysql_mutex_is_owner(), safe_mutex_is_owner(): New predicate, usable
with SAFE_MUTEX. Some InnoDB debug assertions need this predicate
instead of mysql_mutex_assert_owner() or mysql_mutex_assert_not_owner().
buf_pool_t::n_flush_LRU, buf_pool_t::n_flush_list:
Replaces buf_pool_t::init_flush[] and buf_pool_t::n_flush[].
The number of active flush operations.
buf_pool_t::mutex, buf_pool_t::flush_list_mutex: Use mysql_mutex_t
instead of ib_mutex_t, to have native mutexes with PERFORMANCE_SCHEMA
and SAFE_MUTEX instrumentation.
buf_pool_t::done_flush_LRU: Condition variable for !n_flush_LRU.
buf_pool_t::done_flush_list: Condition variable for !n_flush_list.
buf_pool_t::do_flush_list: Condition variable to wake up the
buf_flush_page_cleaner when a log checkpoint needs to be written
or the server is being shut down. Replaces buf_flush_event.
We will keep using timed waits (the page cleaner thread will wake
_at least_ once per second), because the calculations for
innodb_adaptive_flushing depend on fixed time intervals.
buf_dblwr: Allocate statically, and move all code to member functions.
Use a native mutex and condition variable. Remove code to deal with
single-page flushing.
buf_dblwr_check_block(): Make the check debug-only. We were spending
a significant amount of execution time in page_simple_validate_new().
flush_counters_t::unzip_LRU_evicted: Remove.
IORequest: Make more members const. FIXME: m_fil_node should be removed.
buf_flush_sync_lsn: Protect by std::atomic, not page_cleaner.mutex
(which we are removing).
page_cleaner_slot_t, page_cleaner_t: Remove many redundant members.
pc_request_flush_slot(): Replaces pc_request() and pc_flush_slot().
recv_writer_thread: Remove. Recovery works just fine without it, if we
simply invoke buf_flush_sync() at the end of each batch in
recv_sys_t::apply().
recv_recovery_from_checkpoint_finish(): Remove. We can simply call
recv_sys.debug_free() directly.
srv_started_redo: Replaces srv_start_state.
SRV_SHUTDOWN_FLUSH_PHASE: Remove. logs_empty_and_mark_files_at_shutdown()
can communicate with the normal page cleaner loop via the new function
flush_buffer_pool().
buf_flush_remove(): Assert that the calling thread is holding
buf_pool.flush_list_mutex. This removes unnecessary mutex operations
from buf_flush_remove_pages() and buf_flush_dirty_pages(),
which replace buf_LRU_flush_or_remove_pages().
buf_flush_lists(): Renamed from buf_flush_batch(), with simplified
interface. Return the number of flushed pages. Clarified comments and
renamed min_n to max_n. Identify LRU batch by lsn=0. Merge all the functions
buf_flush_start(), buf_flush_batch(), buf_flush_end() directly to this
function, which was their only caller, and remove 2 unnecessary
buf_pool.mutex release/re-acquisition that we used to perform around
the buf_flush_batch() call. At the start, if not all log has been
durably written, wait for a background task to do it, or start a new
task to do it. This allows the log write to run concurrently with our
page flushing batch. Any pages that were skipped due to too recent
FIL_PAGE_LSN or due to them being latched by a writer should be flushed
during the next batch, unless there are further modifications to those
pages. It is possible that a page that we must flush due to small
oldest_modification also carries a recent FIL_PAGE_LSN or is being
constantly modified. In the worst case, all writers would then end up
waiting in log_free_check() to allow the flushing and the checkpoint
to complete.
buf_do_flush_list_batch(): Clarify comments, and rename min_n to max_n.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_flush_space(): Auxiliary function to look up a tablespace for
page flushing.
buf_flush_page(): Defer the computation of space->full_crc32(). Never
call log_write_up_to(), but instead skip persistent pages whose latest
modification (FIL_PAGE_LSN) is newer than the redo log. Also skip
pages on which we cannot acquire a shared latch without waiting.
buf_flush_try_neighbors(): Do not bother checking buf_fix_count
because buf_flush_page() will no longer wait for the page latch.
Take the tablespace as a parameter, and only execute this function
when innodb_flush_neighbors>0. Avoid repeated calls of page_id_t::fold().
buf_flush_relocate_on_flush_list(): Declare as cold, and push down
a condition from the callers.
buf_flush_check_neighbor(): Take id.fold() as a parameter.
buf_flush_sync(): Ensure that the buf_pool.flush_list is empty,
because the flushing batch will skip pages whose modifications have
not yet been written to the log or were latched for modification.
buf_free_from_unzip_LRU_list_batch(): Remove redundant local variables.
buf_flush_LRU_list_batch(): Let the caller buf_do_LRU_batch() initialize
the counters, and report n->evicted.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_do_LRU_batch(): Return the number of pages flushed.
buf_LRU_free_page(): Only release and re-acquire buf_pool.mutex if
adaptive hash index entries are pointing to the block.
buf_LRU_get_free_block(): Do not wake up the page cleaner, because it
will no longer perform any useful work for us, and we do not want it
to compete for I/O while buf_flush_lists(innodb_lru_flush_size, 0)
writes out and evicts at most innodb_lru_flush_size pages. (The
function buf_do_LRU_batch() may complete after writing fewer pages if
more than innodb_lru_scan_depth pages end up in buf_pool.free list.)
Eliminate some mutex release-acquire cycles, and wait for the LRU
flush batch to complete before rescanning.
buf_LRU_check_size_of_non_data_objects(): Simplify the code.
buf_page_write_complete(): Remove the parameter evict, and always
evict pages that were part of an LRU flush.
buf_page_create(): Take a pre-allocated page as a parameter.
buf_pool_t::free_block(): Free a pre-allocated block.
recv_sys_t::recover_low(), recv_sys_t::apply(): Preallocate the block
while not holding recv_sys.mutex. During page allocation, we may
initiate a page flush, which in turn may initiate a log flush, which
would require acquiring log_sys.mutex, which should always be acquired
before recv_sys.mutex in order to avoid deadlocks. Therefore, we must
not be holding recv_sys.mutex while allocating a buffer pool block.
BtrBulk::logFreeCheck(): Skip a redundant condition.
row_undo_step(): Do not invoke srv_inc_activity_count() for every row
that is being rolled back. It should suffice to invoke the function in
trx_flush_log_if_needed() during trx_t::commit_in_memory() when the
rollback completes.
sync_check_enable(): Remove. We will enable innodb_sync_debug from the
very beginning.
Reviewed by: Vladislav Vaintroub
5 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-23399: Performance regression with write workloads
The buffer pool refactoring in MDEV-15053 and MDEV-22871 shifted
the performance bottleneck to the page flushing.
The configuration parameters will be changed as follows:
innodb_lru_flush_size=32 (new: how many pages to flush on LRU eviction)
innodb_lru_scan_depth=1536 (old: 1024)
innodb_max_dirty_pages_pct=90 (old: 75)
innodb_max_dirty_pages_pct_lwm=75 (old: 0)
Note: The parameter innodb_lru_scan_depth will only affect LRU
eviction of buffer pool pages when a new page is being allocated. The
page cleaner thread will no longer evict any pages. It used to
guarantee that some pages will remain free in the buffer pool. Now, we
perform that eviction 'on demand' in buf_LRU_get_free_block().
The parameter innodb_lru_scan_depth(srv_LRU_scan_depth) is used as follows:
* When the buffer pool is being shrunk in buf_pool_t::withdraw_blocks()
* As a buf_pool.free limit in buf_LRU_list_batch() for terminating
the flushing that is initiated e.g., by buf_LRU_get_free_block()
The parameter also used to serve as an initial limit for unzip_LRU
eviction (evicting uncompressed page frames while retaining
ROW_FORMAT=COMPRESSED pages), but now we will use a hard-coded limit
of 100 or unlimited for invoking buf_LRU_scan_and_free_block().
The status variables will be changed as follows:
innodb_buffer_pool_pages_flushed: This includes also the count of
innodb_buffer_pool_pages_LRU_flushed and should work reliably,
updated one by one in buf_flush_page() to give more real-time
statistics. The function buf_flush_stats(), which we are removing,
was not called in every code path. For both counters, we will use
regular variables that are incremented in a critical section of
buf_pool.mutex. Note that show_innodb_vars() directly links to the
variables, and reads of the counters will *not* be protected by
buf_pool.mutex, so you cannot get a consistent snapshot of both variables.
The following INFORMATION_SCHEMA.INNODB_METRICS counters will be
removed, because the page cleaner no longer deals with writing or
evicting least recently used pages, and because the single-page writes
have been removed:
* buffer_LRU_batch_flush_avg_time_slot
* buffer_LRU_batch_flush_avg_time_thread
* buffer_LRU_batch_flush_avg_time_est
* buffer_LRU_batch_flush_avg_pass
* buffer_LRU_single_flush_scanned
* buffer_LRU_single_flush_num_scan
* buffer_LRU_single_flush_scanned_per_call
When moving to a single buffer pool instance in MDEV-15058, we missed
some opportunity to simplify the buf_flush_page_cleaner thread. It was
unnecessarily using a mutex and some complex data structures, even
though we always have a single page cleaner thread.
Furthermore, the buf_flush_page_cleaner thread had separate 'recovery'
and 'shutdown' modes where it was waiting to be triggered by some
other thread, adding unnecessary latency and potential for hangs in
relatively rarely executed startup or shutdown code.
The page cleaner was also running two kinds of batches in an
interleaved fashion: "LRU flush" (writing out some least recently used
pages and evicting them on write completion) and the normal batches
that aim to increase the MIN(oldest_modification) in the buffer pool,
to help the log checkpoint advance.
The buf_pool.flush_list flushing was being blocked by
buf_block_t::lock for no good reason. Furthermore, if the FIL_PAGE_LSN
of a page is ahead of log_sys.get_flushed_lsn(), that is, what has
been persistently written to the redo log, we would trigger a log
flush and then resume the page flushing. This would unnecessarily
limit the performance of the page cleaner thread and trigger the
infamous messages "InnoDB: page_cleaner: 1000ms intended loop took 4450ms.
The settings might not be optimal" that were suppressed in
commit d1ab89037a518fcffbc50c24e4bd94e4ec33aed0 unless log_warnings>2.
Our revised algorithm will make log_sys.get_flushed_lsn() advance at
the start of buf_flush_lists(), and then execute a 'best effort' to
write out all pages. The flush batches will skip pages that were modified
since the log was written, or are are currently exclusively locked.
The MDEV-13670 message "page_cleaner: 1000ms intended loop took" message
will be removed, because by design, the buf_flush_page_cleaner() should
not be blocked during a batch for extended periods of time.
We will remove the single-page flushing altogether. Related to this,
the debug parameter innodb_doublewrite_batch_size will be removed,
because all of the doublewrite buffer will be used for flushing
batches. If a page needs to be evicted from the buffer pool and all
100 least recently used pages in the buffer pool have unflushed
changes, buf_LRU_get_free_block() will execute buf_flush_lists() to
write out and evict innodb_lru_flush_size pages. At most one thread
will execute buf_flush_lists() in buf_LRU_get_free_block(); other
threads will wait for that LRU flushing batch to finish.
To improve concurrency, we will replace the InnoDB ib_mutex_t and
os_event_t native mutexes and condition variables in this area of code.
Most notably, this means that the buffer pool mutex (buf_pool.mutex)
is no longer instrumented via any InnoDB interfaces. It will continue
to be instrumented via PERFORMANCE_SCHEMA.
For now, both buf_pool.flush_list_mutex and buf_pool.mutex will be
declared with MY_MUTEX_INIT_FAST (PTHREAD_MUTEX_ADAPTIVE_NP). The critical
sections of buf_pool.flush_list_mutex should be shorter than those for
buf_pool.mutex, because in the worst case, they cover a linear scan of
buf_pool.flush_list, while the worst case of a critical section of
buf_pool.mutex covers a linear scan of the potentially much longer
buf_pool.LRU list.
mysql_mutex_is_owner(), safe_mutex_is_owner(): New predicate, usable
with SAFE_MUTEX. Some InnoDB debug assertions need this predicate
instead of mysql_mutex_assert_owner() or mysql_mutex_assert_not_owner().
buf_pool_t::n_flush_LRU, buf_pool_t::n_flush_list:
Replaces buf_pool_t::init_flush[] and buf_pool_t::n_flush[].
The number of active flush operations.
buf_pool_t::mutex, buf_pool_t::flush_list_mutex: Use mysql_mutex_t
instead of ib_mutex_t, to have native mutexes with PERFORMANCE_SCHEMA
and SAFE_MUTEX instrumentation.
buf_pool_t::done_flush_LRU: Condition variable for !n_flush_LRU.
buf_pool_t::done_flush_list: Condition variable for !n_flush_list.
buf_pool_t::do_flush_list: Condition variable to wake up the
buf_flush_page_cleaner when a log checkpoint needs to be written
or the server is being shut down. Replaces buf_flush_event.
We will keep using timed waits (the page cleaner thread will wake
_at least_ once per second), because the calculations for
innodb_adaptive_flushing depend on fixed time intervals.
buf_dblwr: Allocate statically, and move all code to member functions.
Use a native mutex and condition variable. Remove code to deal with
single-page flushing.
buf_dblwr_check_block(): Make the check debug-only. We were spending
a significant amount of execution time in page_simple_validate_new().
flush_counters_t::unzip_LRU_evicted: Remove.
IORequest: Make more members const. FIXME: m_fil_node should be removed.
buf_flush_sync_lsn: Protect by std::atomic, not page_cleaner.mutex
(which we are removing).
page_cleaner_slot_t, page_cleaner_t: Remove many redundant members.
pc_request_flush_slot(): Replaces pc_request() and pc_flush_slot().
recv_writer_thread: Remove. Recovery works just fine without it, if we
simply invoke buf_flush_sync() at the end of each batch in
recv_sys_t::apply().
recv_recovery_from_checkpoint_finish(): Remove. We can simply call
recv_sys.debug_free() directly.
srv_started_redo: Replaces srv_start_state.
SRV_SHUTDOWN_FLUSH_PHASE: Remove. logs_empty_and_mark_files_at_shutdown()
can communicate with the normal page cleaner loop via the new function
flush_buffer_pool().
buf_flush_remove(): Assert that the calling thread is holding
buf_pool.flush_list_mutex. This removes unnecessary mutex operations
from buf_flush_remove_pages() and buf_flush_dirty_pages(),
which replace buf_LRU_flush_or_remove_pages().
buf_flush_lists(): Renamed from buf_flush_batch(), with simplified
interface. Return the number of flushed pages. Clarified comments and
renamed min_n to max_n. Identify LRU batch by lsn=0. Merge all the functions
buf_flush_start(), buf_flush_batch(), buf_flush_end() directly to this
function, which was their only caller, and remove 2 unnecessary
buf_pool.mutex release/re-acquisition that we used to perform around
the buf_flush_batch() call. At the start, if not all log has been
durably written, wait for a background task to do it, or start a new
task to do it. This allows the log write to run concurrently with our
page flushing batch. Any pages that were skipped due to too recent
FIL_PAGE_LSN or due to them being latched by a writer should be flushed
during the next batch, unless there are further modifications to those
pages. It is possible that a page that we must flush due to small
oldest_modification also carries a recent FIL_PAGE_LSN or is being
constantly modified. In the worst case, all writers would then end up
waiting in log_free_check() to allow the flushing and the checkpoint
to complete.
buf_do_flush_list_batch(): Clarify comments, and rename min_n to max_n.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_flush_space(): Auxiliary function to look up a tablespace for
page flushing.
buf_flush_page(): Defer the computation of space->full_crc32(). Never
call log_write_up_to(), but instead skip persistent pages whose latest
modification (FIL_PAGE_LSN) is newer than the redo log. Also skip
pages on which we cannot acquire a shared latch without waiting.
buf_flush_try_neighbors(): Do not bother checking buf_fix_count
because buf_flush_page() will no longer wait for the page latch.
Take the tablespace as a parameter, and only execute this function
when innodb_flush_neighbors>0. Avoid repeated calls of page_id_t::fold().
buf_flush_relocate_on_flush_list(): Declare as cold, and push down
a condition from the callers.
buf_flush_check_neighbor(): Take id.fold() as a parameter.
buf_flush_sync(): Ensure that the buf_pool.flush_list is empty,
because the flushing batch will skip pages whose modifications have
not yet been written to the log or were latched for modification.
buf_free_from_unzip_LRU_list_batch(): Remove redundant local variables.
buf_flush_LRU_list_batch(): Let the caller buf_do_LRU_batch() initialize
the counters, and report n->evicted.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_do_LRU_batch(): Return the number of pages flushed.
buf_LRU_free_page(): Only release and re-acquire buf_pool.mutex if
adaptive hash index entries are pointing to the block.
buf_LRU_get_free_block(): Do not wake up the page cleaner, because it
will no longer perform any useful work for us, and we do not want it
to compete for I/O while buf_flush_lists(innodb_lru_flush_size, 0)
writes out and evicts at most innodb_lru_flush_size pages. (The
function buf_do_LRU_batch() may complete after writing fewer pages if
more than innodb_lru_scan_depth pages end up in buf_pool.free list.)
Eliminate some mutex release-acquire cycles, and wait for the LRU
flush batch to complete before rescanning.
buf_LRU_check_size_of_non_data_objects(): Simplify the code.
buf_page_write_complete(): Remove the parameter evict, and always
evict pages that were part of an LRU flush.
buf_page_create(): Take a pre-allocated page as a parameter.
buf_pool_t::free_block(): Free a pre-allocated block.
recv_sys_t::recover_low(), recv_sys_t::apply(): Preallocate the block
while not holding recv_sys.mutex. During page allocation, we may
initiate a page flush, which in turn may initiate a log flush, which
would require acquiring log_sys.mutex, which should always be acquired
before recv_sys.mutex in order to avoid deadlocks. Therefore, we must
not be holding recv_sys.mutex while allocating a buffer pool block.
BtrBulk::logFreeCheck(): Skip a redundant condition.
row_undo_step(): Do not invoke srv_inc_activity_count() for every row
that is being rolled back. It should suffice to invoke the function in
trx_flush_log_if_needed() during trx_t::commit_in_memory() when the
rollback completes.
sync_check_enable(): Remove. We will enable innodb_sync_debug from the
very beginning.
Reviewed by: Vladislav Vaintroub
5 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-24142: Replace InnoDB rw_lock_t with sux_lock
InnoDB buffer pool block and index tree latches depend on a
special kind of read-update-write lock that allows reentrant
(recursive) acquisition of the 'update' and 'write' locks
as well as an upgrade from 'update' lock to 'write' lock.
The 'update' lock allows any number of reader locks from
other threads, but no concurrent 'update' or 'write' lock.
If there were no requirement to support an upgrade from 'update'
to 'write', we could compose the lock out of two srw_lock
(implemented as any type of native rw-lock, such as SRWLOCK on
Microsoft Windows). Removing this requirement is very difficult,
so in commit f7e7f487d4b06695f91f6fbeb0396b9d87fc7bbf we
implemented an 'update' mode to our srw_lock.
Re-entrant or recursive locking is mostly needed when writing or
freeing BLOB pages, but also in crash recovery or when merging
buffered changes to an index page. The re-entrancy allows us to
attach a previously acquired page to a sub-mini-transaction that
will be committed before whatever else is holding the page latch.
The SUX lock supports Shared ('read'), Update, and eXclusive ('write')
locking modes. The S latches are not re-entrant, but a single S latch
may be acquired even if the thread already holds an U latch.
The idea of the U latch is to allow a write of something that concurrent
readers do not care about (such as the contents of BTR_SEG_LEAF,
BTR_SEG_TOP and other page allocation metadata structures, or
the MDEV-6076 PAGE_ROOT_AUTO_INC). (The PAGE_ROOT_AUTO_INC field
is only updated when a dict_table_t for the table exists, and only
read when a dict_table_t for the table is being added to dict_sys.)
block_lock::u_lock_try(bool for_io=true) is used in buf_flush_page()
to allow concurrent readers but no concurrent modifications while the
page is being written to the data file. That latch will be released
by buf_page_write_complete() in a different thread. Hence, we use
the special lock owner value FOR_IO.
The index_lock::u_lock() improves concurrency on operations that
involve non-leaf index pages.
The interface has been cleaned up a little. We will use
x_lock_recursive() instead of x_lock() when we know that a
lock is already held by the current thread. Similarly,
a lock upgrade from U to X is only allowed via u_x_upgrade()
or x_lock_upgraded() but not via x_lock().
We will disable the LatchDebug and sync_array interfaces to
InnoDB rw-locks.
The SEMAPHORES section of SHOW ENGINE INNODB STATUS output
will no longer include any information about InnoDB rw-locks,
only TTASEventMutex (cmake -DMUTEXTYPE=event) waits.
This will make a part of the 'innotop' script dead code.
The block_lock buf_block_t::lock will not be covered by any
PERFORMANCE_SCHEMA instrumentation.
SHOW ENGINE INNODB MUTEX and INFORMATION_SCHEMA.INNODB_MUTEXES
will no longer output source code file names or line numbers.
The dict_index_t::lock will be identified by index and table names,
which should be much more useful. PERFORMANCE_SCHEMA is lumping
information about all dict_index_t::lock together as
event_name='wait/synch/sxlock/innodb/index_tree_rw_lock'.
buf_page_free(): Remove the file,line parameters. The sux_lock will
not store such diagnostic information.
buf_block_dbg_add_level(): Define as empty macro, to be removed
in a subsequent commit.
Unless the build was configured with cmake -DPLUGIN_PERFSCHEMA=NO
the index_lock dict_index_t::lock will be instrumented via
PERFORMANCE_SCHEMA. Similar to
commit 1669c8890ca2e9092213626e5b047e58ca8b1e77
we will distinguish lock waits by registering shared_lock,exclusive_lock
events instead of try_shared_lock,try_exclusive_lock.
Actual 'try' operations will not be instrumented at all.
rw_lock_list: Remove. After MDEV-24167, this only covered
buf_block_t::lock and dict_index_t::lock. We will output their
information by traversing buf_pool or dict_sys.
5 years ago MDEV-15528 Punch holes when pages are freed
When a InnoDB data file page is freed, its contents becomes garbage,
and any storage allocated in the data file is wasted. During flushing,
InnoDB initializes the page with zeros if scrubbing is enabled. If the
tablespace is compressed then InnoDB should punch a hole else ignore the
flushing of the freed page.
buf_page_t:
- Replaced the variable file_page_was_freed, init_on_flush in buf_page_t
with status enum variable.
- Changed all debug assert of file_page_was_freed to DBUG_ASSERT
of buf_page_t::status
Removed buf_page_set_file_page_was_freed(),
buf_page_reset_file_page_was_freed().
buf_page_free(): Newly added function which takes X-lock on the page
before marking the status as FREED. So that InnoDB flush handler can
avoid concurrent flush of the freed page. Also while flushing the page,
InnoDB make sure that redo log which does freeing of the page also written
to the disk. Currently, this function only marks the page as FREED if
it is in buffer pool
buf_flush_freed_page(): Newly added function which initializes zeros
asynchorously if innodb_immediate_scrub_data_uncompressed is enabled.
Punch a hole to the file synchorously if page_compressed is enabled.
Reset the io_fix to NORMAL. Release the block from flush list and
associated mutex before writing zeros or punch a hole to the file.
buf_flush_page(): Removed the unnecessary usage of temporary
variable "flush"
fil_io(): Introduce new parameter called punch_hole. It allows fil_io()
to punch the hole to the file for the given offset.
buf_page_create(): Let the callers assign buf_page_t::status.
Every caller should eventually invoke mtr_t::init().
fsp_page_create(): Remove the unused mtr_t parameter.
In all other callers of buf_page_create() except fsp_page_create(),
before invoking mtr_t::init(), invoke
mtr_t::sx_latch_at_savepoint() or mtr_t::x_latch_at_savepoint().
mtr_t::init(): Initialize buf_page_t::status also for the temporary
tablespace (when redo logging is disabled), to avoid assertion failures.
6 years ago MDEV-13542: Crashing on corrupted page is unhelpful
The approach to handling corruption that was chosen by Oracle in
commit 177d8b0c125b841c0650d27d735e3b87509dc286
is not really useful. Not only did it actually fail to prevent InnoDB
from crashing, but it is making things worse by blocking attempts to
rescue data from or rebuild a partially readable table.
We will try to prevent crashes in a different way: by propagating
errors up the call stack. We will never mark the clustered index
persistently corrupted, so that data recovery may be attempted by
reading from the table, or by rebuilding the table.
This should also fix MDEV-13680 (crash on btr_page_alloc() failure);
it was extensively tested with innodb_file_per_table=0 and a
non-autoextend system tablespace.
We should now avoid crashes in many cases, such as when a page
cannot be read or allocated, or an inconsistency is detected when
attempting to update multiple pages. We will not crash on double-free,
such as on the recovery of DDL in system tablespace in case something
was corrupted.
Crashes on corrupted data are still possible. The fault injection mechanism
that is introduced in the subsequent commit may help catch more of them.
buf_page_import_corrupt_failure: Remove the fault injection, and instead
corrupt some pages using Perl code in the tests.
btr_cur_pessimistic_insert(): Always reserve extents (except for the
change buffer), in order to prevent a subsequent allocation failure.
btr_pcur_open_at_rnd_pos(): Merged to the only caller ibuf_merge_pages().
btr_assert_not_corrupted(), btr_corruption_report(): Remove.
Similar checks are already part of btr_block_get().
FSEG_MAGIC_N_BYTES: Replaces FSEG_MAGIC_N_VALUE.
dict_hdr_get(), trx_rsegf_get_new(), trx_undo_page_get(),
trx_undo_page_get_s_latched(): Replaced with error-checking calls.
trx_rseg_t::get(mtr_t*): Replaces trx_rsegf_get().
trx_rseg_header_create(): Let the caller update the TRX_SYS page if needed.
trx_sys_create_sys_pages(): Merged with trx_sysf_create().
dict_check_tablespaces_and_store_max_id(): Do not access
DICT_HDR_MAX_SPACE_ID, because it was already recovered in dict_boot().
Merge dict_check_sys_tables() with this function.
dir_pathname(): Replaces os_file_make_new_pathname().
row_undo_ins_remove_sec(): Do not modify the undo page by adding
a terminating NUL byte to the record.
btr_decryption_failed(): Report decryption failures
dict_set_corrupted_by_space(), dict_set_encrypted_by_space(),
dict_set_corrupted_index_cache_only(): Remove.
dict_set_corrupted(): Remove the constant parameter dict_locked=false.
Never flag the clustered index corrupted in SYS_INDEXES, because
that would deny further access to the table. It might be possible to
repair the table by executing ALTER TABLE or OPTIMIZE TABLE, in case
no B-tree leaf page is corrupted.
dict_table_skip_corrupt_index(), dict_table_next_uncorrupted_index(),
row_purge_skip_uncommitted_virtual_index(): Remove, and refactor
the callers to read dict_index_t::type only once.
dict_table_is_corrupted(): Remove.
dict_index_t::is_btree(): Determine if the index is a valid B-tree.
BUF_GET_NO_LATCH, BUF_EVICT_IF_IN_POOL: Remove.
UNIV_BTR_DEBUG: Remove. Any inconsistency will no longer trigger
assertion failures, but error codes being returned.
buf_corrupt_page_release(): Replaced with a direct call to
buf_pool.corrupted_evict().
fil_invalid_page_access_msg(): Never crash on an invalid read;
let the caller of buf_page_get_gen() decide.
btr_pcur_t::restore_position(): Propagate failure status to the caller
by returning CORRUPTED.
opt_search_plan_for_table(): Simplify the code.
row_purge_del_mark(), row_purge_upd_exist_or_extern_func(),
row_undo_ins_remove_sec_rec(), row_undo_mod_upd_del_sec(),
row_undo_mod_del_mark_sec(): Avoid mem_heap_create()/mem_heap_free()
when no secondary indexes exist.
row_undo_mod_upd_exist_sec(): Simplify the code.
row_upd_clust_step(), dict_load_table_one(): Return DB_TABLE_CORRUPT
if the clustered index (and therefore the table) is corrupted, similar
to what we do in row_insert_for_mysql().
fut_get_ptr(): Replace with buf_page_get_gen() calls.
buf_page_get_gen(): Return nullptr and *err=DB_CORRUPTION
if the page is marked as freed. For other modes than
BUF_GET_POSSIBLY_FREED or BUF_PEEK_IF_IN_POOL this will
trigger a debug assertion failure. For BUF_GET_POSSIBLY_FREED,
we will return nullptr for freed pages, so that the callers
can be simplified. The purge of transaction history will be
a new user of BUF_GET_POSSIBLY_FREED, to avoid crashes on
corrupted data.
buf_page_get_low(): Never crash on a corrupted page, but simply
return nullptr.
fseg_page_is_allocated(): Replaces fseg_page_is_free().
fts_drop_common_tables(): Return an error if the transaction
was rolled back.
fil_space_t::set_corrupted(): Report a tablespace as corrupted if
it was not reported already.
fil_space_t::io(): Invoke fil_space_t::set_corrupted() to report
out-of-bounds page access or other errors.
Clean up mtr_t::page_lock()
buf_page_get_low(): Validate the page identifier (to check for
recently read corrupted pages) after acquiring the page latch.
buf_page_t::read_complete(): Flag uninitialized (all-zero) pages
with DB_FAIL. Return DB_PAGE_CORRUPTED on page number mismatch.
mtr_t::defer_drop_ahi(): Renamed from mtr_defer_drop_ahi().
recv_sys_t::free_corrupted_page(): Only set_corrupt_fs()
if any log records exist for the page. We do not mind if read-ahead
produces corrupted (or all-zero) pages that were not actually needed
during recovery.
recv_recover_page(): Return whether the operation succeeded.
recv_sys_t::recover_low(): Simplify the logic. Check for recovery error.
Thanks to Matthias Leich for testing this extensively and to the
authors of https://rr-project.org for making it easy to diagnose
and fix any failures that were found during the testing.
3 years ago MDEV-15528 Punch holes when pages are freed
When a InnoDB data file page is freed, its contents becomes garbage,
and any storage allocated in the data file is wasted. During flushing,
InnoDB initializes the page with zeros if scrubbing is enabled. If the
tablespace is compressed then InnoDB should punch a hole else ignore the
flushing of the freed page.
buf_page_t:
- Replaced the variable file_page_was_freed, init_on_flush in buf_page_t
with status enum variable.
- Changed all debug assert of file_page_was_freed to DBUG_ASSERT
of buf_page_t::status
Removed buf_page_set_file_page_was_freed(),
buf_page_reset_file_page_was_freed().
buf_page_free(): Newly added function which takes X-lock on the page
before marking the status as FREED. So that InnoDB flush handler can
avoid concurrent flush of the freed page. Also while flushing the page,
InnoDB make sure that redo log which does freeing of the page also written
to the disk. Currently, this function only marks the page as FREED if
it is in buffer pool
buf_flush_freed_page(): Newly added function which initializes zeros
asynchorously if innodb_immediate_scrub_data_uncompressed is enabled.
Punch a hole to the file synchorously if page_compressed is enabled.
Reset the io_fix to NORMAL. Release the block from flush list and
associated mutex before writing zeros or punch a hole to the file.
buf_flush_page(): Removed the unnecessary usage of temporary
variable "flush"
fil_io(): Introduce new parameter called punch_hole. It allows fil_io()
to punch the hole to the file for the given offset.
buf_page_create(): Let the callers assign buf_page_t::status.
Every caller should eventually invoke mtr_t::init().
fsp_page_create(): Remove the unused mtr_t parameter.
In all other callers of buf_page_create() except fsp_page_create(),
before invoking mtr_t::init(), invoke
mtr_t::sx_latch_at_savepoint() or mtr_t::x_latch_at_savepoint().
mtr_t::init(): Initialize buf_page_t::status also for the temporary
tablespace (when redo logging is disabled), to avoid assertion failures.
6 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-13542: Crashing on corrupted page is unhelpful
The approach to handling corruption that was chosen by Oracle in
commit 177d8b0c125b841c0650d27d735e3b87509dc286
is not really useful. Not only did it actually fail to prevent InnoDB
from crashing, but it is making things worse by blocking attempts to
rescue data from or rebuild a partially readable table.
We will try to prevent crashes in a different way: by propagating
errors up the call stack. We will never mark the clustered index
persistently corrupted, so that data recovery may be attempted by
reading from the table, or by rebuilding the table.
This should also fix MDEV-13680 (crash on btr_page_alloc() failure);
it was extensively tested with innodb_file_per_table=0 and a
non-autoextend system tablespace.
We should now avoid crashes in many cases, such as when a page
cannot be read or allocated, or an inconsistency is detected when
attempting to update multiple pages. We will not crash on double-free,
such as on the recovery of DDL in system tablespace in case something
was corrupted.
Crashes on corrupted data are still possible. The fault injection mechanism
that is introduced in the subsequent commit may help catch more of them.
buf_page_import_corrupt_failure: Remove the fault injection, and instead
corrupt some pages using Perl code in the tests.
btr_cur_pessimistic_insert(): Always reserve extents (except for the
change buffer), in order to prevent a subsequent allocation failure.
btr_pcur_open_at_rnd_pos(): Merged to the only caller ibuf_merge_pages().
btr_assert_not_corrupted(), btr_corruption_report(): Remove.
Similar checks are already part of btr_block_get().
FSEG_MAGIC_N_BYTES: Replaces FSEG_MAGIC_N_VALUE.
dict_hdr_get(), trx_rsegf_get_new(), trx_undo_page_get(),
trx_undo_page_get_s_latched(): Replaced with error-checking calls.
trx_rseg_t::get(mtr_t*): Replaces trx_rsegf_get().
trx_rseg_header_create(): Let the caller update the TRX_SYS page if needed.
trx_sys_create_sys_pages(): Merged with trx_sysf_create().
dict_check_tablespaces_and_store_max_id(): Do not access
DICT_HDR_MAX_SPACE_ID, because it was already recovered in dict_boot().
Merge dict_check_sys_tables() with this function.
dir_pathname(): Replaces os_file_make_new_pathname().
row_undo_ins_remove_sec(): Do not modify the undo page by adding
a terminating NUL byte to the record.
btr_decryption_failed(): Report decryption failures
dict_set_corrupted_by_space(), dict_set_encrypted_by_space(),
dict_set_corrupted_index_cache_only(): Remove.
dict_set_corrupted(): Remove the constant parameter dict_locked=false.
Never flag the clustered index corrupted in SYS_INDEXES, because
that would deny further access to the table. It might be possible to
repair the table by executing ALTER TABLE or OPTIMIZE TABLE, in case
no B-tree leaf page is corrupted.
dict_table_skip_corrupt_index(), dict_table_next_uncorrupted_index(),
row_purge_skip_uncommitted_virtual_index(): Remove, and refactor
the callers to read dict_index_t::type only once.
dict_table_is_corrupted(): Remove.
dict_index_t::is_btree(): Determine if the index is a valid B-tree.
BUF_GET_NO_LATCH, BUF_EVICT_IF_IN_POOL: Remove.
UNIV_BTR_DEBUG: Remove. Any inconsistency will no longer trigger
assertion failures, but error codes being returned.
buf_corrupt_page_release(): Replaced with a direct call to
buf_pool.corrupted_evict().
fil_invalid_page_access_msg(): Never crash on an invalid read;
let the caller of buf_page_get_gen() decide.
btr_pcur_t::restore_position(): Propagate failure status to the caller
by returning CORRUPTED.
opt_search_plan_for_table(): Simplify the code.
row_purge_del_mark(), row_purge_upd_exist_or_extern_func(),
row_undo_ins_remove_sec_rec(), row_undo_mod_upd_del_sec(),
row_undo_mod_del_mark_sec(): Avoid mem_heap_create()/mem_heap_free()
when no secondary indexes exist.
row_undo_mod_upd_exist_sec(): Simplify the code.
row_upd_clust_step(), dict_load_table_one(): Return DB_TABLE_CORRUPT
if the clustered index (and therefore the table) is corrupted, similar
to what we do in row_insert_for_mysql().
fut_get_ptr(): Replace with buf_page_get_gen() calls.
buf_page_get_gen(): Return nullptr and *err=DB_CORRUPTION
if the page is marked as freed. For other modes than
BUF_GET_POSSIBLY_FREED or BUF_PEEK_IF_IN_POOL this will
trigger a debug assertion failure. For BUF_GET_POSSIBLY_FREED,
we will return nullptr for freed pages, so that the callers
can be simplified. The purge of transaction history will be
a new user of BUF_GET_POSSIBLY_FREED, to avoid crashes on
corrupted data.
buf_page_get_low(): Never crash on a corrupted page, but simply
return nullptr.
fseg_page_is_allocated(): Replaces fseg_page_is_free().
fts_drop_common_tables(): Return an error if the transaction
was rolled back.
fil_space_t::set_corrupted(): Report a tablespace as corrupted if
it was not reported already.
fil_space_t::io(): Invoke fil_space_t::set_corrupted() to report
out-of-bounds page access or other errors.
Clean up mtr_t::page_lock()
buf_page_get_low(): Validate the page identifier (to check for
recently read corrupted pages) after acquiring the page latch.
buf_page_t::read_complete(): Flag uninitialized (all-zero) pages
with DB_FAIL. Return DB_PAGE_CORRUPTED on page number mismatch.
mtr_t::defer_drop_ahi(): Renamed from mtr_defer_drop_ahi().
recv_sys_t::free_corrupted_page(): Only set_corrupt_fs()
if any log records exist for the page. We do not mind if read-ahead
produces corrupted (or all-zero) pages that were not actually needed
during recovery.
recv_recover_page(): Return whether the operation succeeded.
recv_sys_t::recover_low(): Simplify the logic. Check for recovery error.
Thanks to Matthias Leich for testing this extensively and to the
authors of https://rr-project.org for making it easy to diagnose
and fix any failures that were found during the testing.
3 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-25105 Remove innodb_checksum_algorithm values none,innodb,...
Historically, InnoDB supported a buggy page checksum algorithm that did not
compute a checksum over the full page. Later, well before MySQL 4.1
introduced .ibd files and the innodb_file_per_table option, the algorithm
was corrected and the first 4 bytes of each page were redefined to be
a checksum.
The original checksum was so slow that an option to disable page checksum
was introduced for benchmarketing purposes.
The Intel Nehalem microarchitecture introduced the SSE4.2 instruction set
extension, which includes instructions for faster computation of CRC-32C.
In MySQL 5.6 (and MariaDB 10.0), innodb_checksum_algorithm=crc32 was
implemented to make of that. As that option was changed to be the default
in MySQL 5.7, a bug was found on big-endian platforms and some work-around
code was added to weaken that checksum further. MariaDB disables that
work-around by default since MDEV-17958.
Later, SIMD-accelerated CRC-32C has been implemented in MariaDB for POWER
and ARM and also for IA-32/AMD64, making use of carry-less multiplication
where available.
Long story short, innodb_checksum_algorithm=crc32 is faster and more secure
than the pre-MySQL 5.6 checksum, called innodb_checksum_algorithm=innodb.
It should have removed any need to use innodb_checksum_algorithm=none.
The setting innodb_checksum_algorithm=crc32 is the default in
MySQL 5.7 and MariaDB Server 10.2, 10.3, 10.4. In MariaDB 10.5,
MDEV-19534 made innodb_checksum_algorithm=full_crc32 the default.
It is even faster and more secure.
The default settings in MariaDB do allow old data files to be read,
no matter if a worse checksum algorithm had been used.
(Unfortunately, before innodb_checksum_algorithm=full_crc32,
the data files did not identify which checksum algorithm is being used.)
The non-default settings innodb_checksum_algorithm=strict_crc32 or
innodb_checksum_algorithm=strict_full_crc32 would only allow CRC-32C
checksums. The incompatibility with old data files is why they are
not the default.
The newest server not to support innodb_checksum_algorithm=crc32
were MySQL 5.5 and MariaDB 5.5. Both have reached their end of life.
A valid reason for using innodb_checksum_algorithm=innodb could have
been the ability to downgrade. If it is really needed, data files
can be converted with an older version of the innochecksum utility.
Because there is no good reason to allow data files to be written
with insecure checksums, we will reject those option values:
innodb_checksum_algorithm=none
innodb_checksum_algorithm=innodb
innodb_checksum_algorithm=strict_none
innodb_checksum_algorithm=strict_innodb
Furthermore, the following innochecksum options will be removed,
because only strict crc32 will be supported:
innochecksum --strict-check=crc32
innochecksum -C crc32
innochecksum --write=crc32
innochecksum -w crc32
If a user wishes to convert a data file to use a different checksum
(so that it might be used with the no-longer-supported
MySQL 5.5 or MariaDB 5.5, which do not support IMPORT TABLESPACE
nor system tablespace format changes that were made in MariaDB 10.3),
then the innochecksum tool from MariaDB 10.2, 10.3, 10.4, 10.5 or
MySQL 5.7 can be used.
Reviewed by: Thirunarayanan Balathandayuthapani
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-13542: Crashing on corrupted page is unhelpful
The approach to handling corruption that was chosen by Oracle in
commit 177d8b0c125b841c0650d27d735e3b87509dc286
is not really useful. Not only did it actually fail to prevent InnoDB
from crashing, but it is making things worse by blocking attempts to
rescue data from or rebuild a partially readable table.
We will try to prevent crashes in a different way: by propagating
errors up the call stack. We will never mark the clustered index
persistently corrupted, so that data recovery may be attempted by
reading from the table, or by rebuilding the table.
This should also fix MDEV-13680 (crash on btr_page_alloc() failure);
it was extensively tested with innodb_file_per_table=0 and a
non-autoextend system tablespace.
We should now avoid crashes in many cases, such as when a page
cannot be read or allocated, or an inconsistency is detected when
attempting to update multiple pages. We will not crash on double-free,
such as on the recovery of DDL in system tablespace in case something
was corrupted.
Crashes on corrupted data are still possible. The fault injection mechanism
that is introduced in the subsequent commit may help catch more of them.
buf_page_import_corrupt_failure: Remove the fault injection, and instead
corrupt some pages using Perl code in the tests.
btr_cur_pessimistic_insert(): Always reserve extents (except for the
change buffer), in order to prevent a subsequent allocation failure.
btr_pcur_open_at_rnd_pos(): Merged to the only caller ibuf_merge_pages().
btr_assert_not_corrupted(), btr_corruption_report(): Remove.
Similar checks are already part of btr_block_get().
FSEG_MAGIC_N_BYTES: Replaces FSEG_MAGIC_N_VALUE.
dict_hdr_get(), trx_rsegf_get_new(), trx_undo_page_get(),
trx_undo_page_get_s_latched(): Replaced with error-checking calls.
trx_rseg_t::get(mtr_t*): Replaces trx_rsegf_get().
trx_rseg_header_create(): Let the caller update the TRX_SYS page if needed.
trx_sys_create_sys_pages(): Merged with trx_sysf_create().
dict_check_tablespaces_and_store_max_id(): Do not access
DICT_HDR_MAX_SPACE_ID, because it was already recovered in dict_boot().
Merge dict_check_sys_tables() with this function.
dir_pathname(): Replaces os_file_make_new_pathname().
row_undo_ins_remove_sec(): Do not modify the undo page by adding
a terminating NUL byte to the record.
btr_decryption_failed(): Report decryption failures
dict_set_corrupted_by_space(), dict_set_encrypted_by_space(),
dict_set_corrupted_index_cache_only(): Remove.
dict_set_corrupted(): Remove the constant parameter dict_locked=false.
Never flag the clustered index corrupted in SYS_INDEXES, because
that would deny further access to the table. It might be possible to
repair the table by executing ALTER TABLE or OPTIMIZE TABLE, in case
no B-tree leaf page is corrupted.
dict_table_skip_corrupt_index(), dict_table_next_uncorrupted_index(),
row_purge_skip_uncommitted_virtual_index(): Remove, and refactor
the callers to read dict_index_t::type only once.
dict_table_is_corrupted(): Remove.
dict_index_t::is_btree(): Determine if the index is a valid B-tree.
BUF_GET_NO_LATCH, BUF_EVICT_IF_IN_POOL: Remove.
UNIV_BTR_DEBUG: Remove. Any inconsistency will no longer trigger
assertion failures, but error codes being returned.
buf_corrupt_page_release(): Replaced with a direct call to
buf_pool.corrupted_evict().
fil_invalid_page_access_msg(): Never crash on an invalid read;
let the caller of buf_page_get_gen() decide.
btr_pcur_t::restore_position(): Propagate failure status to the caller
by returning CORRUPTED.
opt_search_plan_for_table(): Simplify the code.
row_purge_del_mark(), row_purge_upd_exist_or_extern_func(),
row_undo_ins_remove_sec_rec(), row_undo_mod_upd_del_sec(),
row_undo_mod_del_mark_sec(): Avoid mem_heap_create()/mem_heap_free()
when no secondary indexes exist.
row_undo_mod_upd_exist_sec(): Simplify the code.
row_upd_clust_step(), dict_load_table_one(): Return DB_TABLE_CORRUPT
if the clustered index (and therefore the table) is corrupted, similar
to what we do in row_insert_for_mysql().
fut_get_ptr(): Replace with buf_page_get_gen() calls.
buf_page_get_gen(): Return nullptr and *err=DB_CORRUPTION
if the page is marked as freed. For other modes than
BUF_GET_POSSIBLY_FREED or BUF_PEEK_IF_IN_POOL this will
trigger a debug assertion failure. For BUF_GET_POSSIBLY_FREED,
we will return nullptr for freed pages, so that the callers
can be simplified. The purge of transaction history will be
a new user of BUF_GET_POSSIBLY_FREED, to avoid crashes on
corrupted data.
buf_page_get_low(): Never crash on a corrupted page, but simply
return nullptr.
fseg_page_is_allocated(): Replaces fseg_page_is_free().
fts_drop_common_tables(): Return an error if the transaction
was rolled back.
fil_space_t::set_corrupted(): Report a tablespace as corrupted if
it was not reported already.
fil_space_t::io(): Invoke fil_space_t::set_corrupted() to report
out-of-bounds page access or other errors.
Clean up mtr_t::page_lock()
buf_page_get_low(): Validate the page identifier (to check for
recently read corrupted pages) after acquiring the page latch.
buf_page_t::read_complete(): Flag uninitialized (all-zero) pages
with DB_FAIL. Return DB_PAGE_CORRUPTED on page number mismatch.
mtr_t::defer_drop_ahi(): Renamed from mtr_defer_drop_ahi().
recv_sys_t::free_corrupted_page(): Only set_corrupt_fs()
if any log records exist for the page. We do not mind if read-ahead
produces corrupted (or all-zero) pages that were not actually needed
during recovery.
recv_recover_page(): Return whether the operation succeeded.
recv_sys_t::recover_low(): Simplify the logic. Check for recovery error.
Thanks to Matthias Leich for testing this extensively and to the
authors of https://rr-project.org for making it easy to diagnose
and fix any failures that were found during the testing.
3 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-13542: Crashing on corrupted page is unhelpful
The approach to handling corruption that was chosen by Oracle in
commit 177d8b0c125b841c0650d27d735e3b87509dc286
is not really useful. Not only did it actually fail to prevent InnoDB
from crashing, but it is making things worse by blocking attempts to
rescue data from or rebuild a partially readable table.
We will try to prevent crashes in a different way: by propagating
errors up the call stack. We will never mark the clustered index
persistently corrupted, so that data recovery may be attempted by
reading from the table, or by rebuilding the table.
This should also fix MDEV-13680 (crash on btr_page_alloc() failure);
it was extensively tested with innodb_file_per_table=0 and a
non-autoextend system tablespace.
We should now avoid crashes in many cases, such as when a page
cannot be read or allocated, or an inconsistency is detected when
attempting to update multiple pages. We will not crash on double-free,
such as on the recovery of DDL in system tablespace in case something
was corrupted.
Crashes on corrupted data are still possible. The fault injection mechanism
that is introduced in the subsequent commit may help catch more of them.
buf_page_import_corrupt_failure: Remove the fault injection, and instead
corrupt some pages using Perl code in the tests.
btr_cur_pessimistic_insert(): Always reserve extents (except for the
change buffer), in order to prevent a subsequent allocation failure.
btr_pcur_open_at_rnd_pos(): Merged to the only caller ibuf_merge_pages().
btr_assert_not_corrupted(), btr_corruption_report(): Remove.
Similar checks are already part of btr_block_get().
FSEG_MAGIC_N_BYTES: Replaces FSEG_MAGIC_N_VALUE.
dict_hdr_get(), trx_rsegf_get_new(), trx_undo_page_get(),
trx_undo_page_get_s_latched(): Replaced with error-checking calls.
trx_rseg_t::get(mtr_t*): Replaces trx_rsegf_get().
trx_rseg_header_create(): Let the caller update the TRX_SYS page if needed.
trx_sys_create_sys_pages(): Merged with trx_sysf_create().
dict_check_tablespaces_and_store_max_id(): Do not access
DICT_HDR_MAX_SPACE_ID, because it was already recovered in dict_boot().
Merge dict_check_sys_tables() with this function.
dir_pathname(): Replaces os_file_make_new_pathname().
row_undo_ins_remove_sec(): Do not modify the undo page by adding
a terminating NUL byte to the record.
btr_decryption_failed(): Report decryption failures
dict_set_corrupted_by_space(), dict_set_encrypted_by_space(),
dict_set_corrupted_index_cache_only(): Remove.
dict_set_corrupted(): Remove the constant parameter dict_locked=false.
Never flag the clustered index corrupted in SYS_INDEXES, because
that would deny further access to the table. It might be possible to
repair the table by executing ALTER TABLE or OPTIMIZE TABLE, in case
no B-tree leaf page is corrupted.
dict_table_skip_corrupt_index(), dict_table_next_uncorrupted_index(),
row_purge_skip_uncommitted_virtual_index(): Remove, and refactor
the callers to read dict_index_t::type only once.
dict_table_is_corrupted(): Remove.
dict_index_t::is_btree(): Determine if the index is a valid B-tree.
BUF_GET_NO_LATCH, BUF_EVICT_IF_IN_POOL: Remove.
UNIV_BTR_DEBUG: Remove. Any inconsistency will no longer trigger
assertion failures, but error codes being returned.
buf_corrupt_page_release(): Replaced with a direct call to
buf_pool.corrupted_evict().
fil_invalid_page_access_msg(): Never crash on an invalid read;
let the caller of buf_page_get_gen() decide.
btr_pcur_t::restore_position(): Propagate failure status to the caller
by returning CORRUPTED.
opt_search_plan_for_table(): Simplify the code.
row_purge_del_mark(), row_purge_upd_exist_or_extern_func(),
row_undo_ins_remove_sec_rec(), row_undo_mod_upd_del_sec(),
row_undo_mod_del_mark_sec(): Avoid mem_heap_create()/mem_heap_free()
when no secondary indexes exist.
row_undo_mod_upd_exist_sec(): Simplify the code.
row_upd_clust_step(), dict_load_table_one(): Return DB_TABLE_CORRUPT
if the clustered index (and therefore the table) is corrupted, similar
to what we do in row_insert_for_mysql().
fut_get_ptr(): Replace with buf_page_get_gen() calls.
buf_page_get_gen(): Return nullptr and *err=DB_CORRUPTION
if the page is marked as freed. For other modes than
BUF_GET_POSSIBLY_FREED or BUF_PEEK_IF_IN_POOL this will
trigger a debug assertion failure. For BUF_GET_POSSIBLY_FREED,
we will return nullptr for freed pages, so that the callers
can be simplified. The purge of transaction history will be
a new user of BUF_GET_POSSIBLY_FREED, to avoid crashes on
corrupted data.
buf_page_get_low(): Never crash on a corrupted page, but simply
return nullptr.
fseg_page_is_allocated(): Replaces fseg_page_is_free().
fts_drop_common_tables(): Return an error if the transaction
was rolled back.
fil_space_t::set_corrupted(): Report a tablespace as corrupted if
it was not reported already.
fil_space_t::io(): Invoke fil_space_t::set_corrupted() to report
out-of-bounds page access or other errors.
Clean up mtr_t::page_lock()
buf_page_get_low(): Validate the page identifier (to check for
recently read corrupted pages) after acquiring the page latch.
buf_page_t::read_complete(): Flag uninitialized (all-zero) pages
with DB_FAIL. Return DB_PAGE_CORRUPTED on page number mismatch.
mtr_t::defer_drop_ahi(): Renamed from mtr_defer_drop_ahi().
recv_sys_t::free_corrupted_page(): Only set_corrupt_fs()
if any log records exist for the page. We do not mind if read-ahead
produces corrupted (or all-zero) pages that were not actually needed
during recovery.
recv_recover_page(): Return whether the operation succeeded.
recv_sys_t::recover_low(): Simplify the logic. Check for recovery error.
Thanks to Matthias Leich for testing this extensively and to the
authors of https://rr-project.org for making it easy to diagnose
and fix any failures that were found during the testing.
3 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-34515: Contention between purge and workload
In a Sysbench oltp_update_index workload that involves 1 table,
a serious contention between the workload and the purge of history
was observed. This was the worst when the table contained only 1 record.
This turned out to be fixed by setting innodb_purge_batch_size=128,
which corresponds to the number of usable persistent rollback segments.
When we go above that, there would be contention between row_purge_poss_sec()
and the workload, typically on the clustered index page latch, sometimes
also on a secondary index page latch. It might be that with smaller
batches, trx_sys.history_size() will end up pausing all concurrent
transaction start/commit frequently enough so that purge will be able
to make some progress, so that there would be less contention on the
index page latches between purge and SQL execution.
In commit aa719b5010c929132b4460b78113fbd07497d9c8 (part of MDEV-32050)
the interpretation of the parameter innodb_purge_batch_size was slightly
changed. It would correspond to the maximum desired size of the
purge_sys.pages cache. Before that change, the parameter was referring to
a number of undo log pages, but the accounting might have been inaccurate.
To avoid a regression, we will reduce the default value to
innodb_purge_batch_size=127, which will also be compatible with
innodb_undo_tablespaces>1 (which will disable rollback segment 0).
Additionally, some logic in the purge and MVCC checks is simplified.
The purge tasks will make use of purge_sys.pages when accessing undo
log pages to find out if a secondary index record can be removed.
If an undo page needs to be looked up in buf_pool.page_hash, we will
merely buffer-fix it. This is correct, because the undo pages are
append-only in nature. Holding purge_sys.latch or purge_sys.end_latch
or the fact that the current thread is executing as a part of an
in-progress purge batch will prevent the contents of the undo page from
being freed and subsequently reused. The buffer-fix will prevent the
page from being evicted form the buffer pool. Thanks to this logic,
we can refer to the undo log record directly in the buffer pool page
and avoid copying the record.
buf_pool_t::page_fix(): Look up and buffer-fix a page. This is useful
for accessing undo log pages, which are append-only by nature.
There will be no need to deal with change buffer or ROW_FORMAT=COMPRESSED
in that case.
purge_sys_t::view_guard::view_guard(): Allow the type of guard to be
acquired: end_latch, latch, or no latch (in case we are a purge thread).
purge_sys_t::view_guard::get(): Read-only accessor to purge_sys.pages.
purge_sys_t::get_page(): Invoke buf_pool_t::page_fix().
row_vers_old_has_index_entry(): Replaced with row_purge_is_unsafe()
and row_undo_mod_sec_unsafe().
trx_undo_get_undo_rec(): Merged to trx_undo_prev_version_build().
row_purge_poss_sec(): Add the parameter mtr and remove redundant
or unused parameters sec_pcur, sec_mtr, is_tree. We will use the
caller's mtr object but release any acquired page latches before
returning.
btr_cur_get_page(), page_cur_get_page(): Do not invoke page_align().
row_purge_remove_sec_if_poss_leaf(): Return the value of PAGE_MAX_TRX_ID
to be checked against the page in row_purge_remove_sec_if_poss_tree().
If the secondary index page was not changed meanwhile, it will be
unnecessary to invoke row_purge_poss_sec() again.
trx_undo_prev_version_build(): Access any undo log pages using
the caller's mini-transaction object.
row_purge_vc_matches_cluster(): Moved to the only compilation unit that
needs it.
Reviewed by: Debarun Banerjee
1 year ago MDEV-34515: Contention between purge and workload
In a Sysbench oltp_update_index workload that involves 1 table,
a serious contention between the workload and the purge of history
was observed. This was the worst when the table contained only 1 record.
This turned out to be fixed by setting innodb_purge_batch_size=128,
which corresponds to the number of usable persistent rollback segments.
When we go above that, there would be contention between row_purge_poss_sec()
and the workload, typically on the clustered index page latch, sometimes
also on a secondary index page latch. It might be that with smaller
batches, trx_sys.history_size() will end up pausing all concurrent
transaction start/commit frequently enough so that purge will be able
to make some progress, so that there would be less contention on the
index page latches between purge and SQL execution.
In commit aa719b5010c929132b4460b78113fbd07497d9c8 (part of MDEV-32050)
the interpretation of the parameter innodb_purge_batch_size was slightly
changed. It would correspond to the maximum desired size of the
purge_sys.pages cache. Before that change, the parameter was referring to
a number of undo log pages, but the accounting might have been inaccurate.
To avoid a regression, we will reduce the default value to
innodb_purge_batch_size=127, which will also be compatible with
innodb_undo_tablespaces>1 (which will disable rollback segment 0).
Additionally, some logic in the purge and MVCC checks is simplified.
The purge tasks will make use of purge_sys.pages when accessing undo
log pages to find out if a secondary index record can be removed.
If an undo page needs to be looked up in buf_pool.page_hash, we will
merely buffer-fix it. This is correct, because the undo pages are
append-only in nature. Holding purge_sys.latch or purge_sys.end_latch
or the fact that the current thread is executing as a part of an
in-progress purge batch will prevent the contents of the undo page from
being freed and subsequently reused. The buffer-fix will prevent the
page from being evicted form the buffer pool. Thanks to this logic,
we can refer to the undo log record directly in the buffer pool page
and avoid copying the record.
buf_pool_t::page_fix(): Look up and buffer-fix a page. This is useful
for accessing undo log pages, which are append-only by nature.
There will be no need to deal with change buffer or ROW_FORMAT=COMPRESSED
in that case.
purge_sys_t::view_guard::view_guard(): Allow the type of guard to be
acquired: end_latch, latch, or no latch (in case we are a purge thread).
purge_sys_t::view_guard::get(): Read-only accessor to purge_sys.pages.
purge_sys_t::get_page(): Invoke buf_pool_t::page_fix().
row_vers_old_has_index_entry(): Replaced with row_purge_is_unsafe()
and row_undo_mod_sec_unsafe().
trx_undo_get_undo_rec(): Merged to trx_undo_prev_version_build().
row_purge_poss_sec(): Add the parameter mtr and remove redundant
or unused parameters sec_pcur, sec_mtr, is_tree. We will use the
caller's mtr object but release any acquired page latches before
returning.
btr_cur_get_page(), page_cur_get_page(): Do not invoke page_align().
row_purge_remove_sec_if_poss_leaf(): Return the value of PAGE_MAX_TRX_ID
to be checked against the page in row_purge_remove_sec_if_poss_tree().
If the secondary index page was not changed meanwhile, it will be
unnecessary to invoke row_purge_poss_sec() again.
trx_undo_prev_version_build(): Access any undo log pages using
the caller's mini-transaction object.
row_purge_vc_matches_cluster(): Moved to the only compilation unit that
needs it.
Reviewed by: Debarun Banerjee
1 year ago MDEV-34515: Contention between purge and workload
In a Sysbench oltp_update_index workload that involves 1 table,
a serious contention between the workload and the purge of history
was observed. This was the worst when the table contained only 1 record.
This turned out to be fixed by setting innodb_purge_batch_size=128,
which corresponds to the number of usable persistent rollback segments.
When we go above that, there would be contention between row_purge_poss_sec()
and the workload, typically on the clustered index page latch, sometimes
also on a secondary index page latch. It might be that with smaller
batches, trx_sys.history_size() will end up pausing all concurrent
transaction start/commit frequently enough so that purge will be able
to make some progress, so that there would be less contention on the
index page latches between purge and SQL execution.
In commit aa719b5010c929132b4460b78113fbd07497d9c8 (part of MDEV-32050)
the interpretation of the parameter innodb_purge_batch_size was slightly
changed. It would correspond to the maximum desired size of the
purge_sys.pages cache. Before that change, the parameter was referring to
a number of undo log pages, but the accounting might have been inaccurate.
To avoid a regression, we will reduce the default value to
innodb_purge_batch_size=127, which will also be compatible with
innodb_undo_tablespaces>1 (which will disable rollback segment 0).
Additionally, some logic in the purge and MVCC checks is simplified.
The purge tasks will make use of purge_sys.pages when accessing undo
log pages to find out if a secondary index record can be removed.
If an undo page needs to be looked up in buf_pool.page_hash, we will
merely buffer-fix it. This is correct, because the undo pages are
append-only in nature. Holding purge_sys.latch or purge_sys.end_latch
or the fact that the current thread is executing as a part of an
in-progress purge batch will prevent the contents of the undo page from
being freed and subsequently reused. The buffer-fix will prevent the
page from being evicted form the buffer pool. Thanks to this logic,
we can refer to the undo log record directly in the buffer pool page
and avoid copying the record.
buf_pool_t::page_fix(): Look up and buffer-fix a page. This is useful
for accessing undo log pages, which are append-only by nature.
There will be no need to deal with change buffer or ROW_FORMAT=COMPRESSED
in that case.
purge_sys_t::view_guard::view_guard(): Allow the type of guard to be
acquired: end_latch, latch, or no latch (in case we are a purge thread).
purge_sys_t::view_guard::get(): Read-only accessor to purge_sys.pages.
purge_sys_t::get_page(): Invoke buf_pool_t::page_fix().
row_vers_old_has_index_entry(): Replaced with row_purge_is_unsafe()
and row_undo_mod_sec_unsafe().
trx_undo_get_undo_rec(): Merged to trx_undo_prev_version_build().
row_purge_poss_sec(): Add the parameter mtr and remove redundant
or unused parameters sec_pcur, sec_mtr, is_tree. We will use the
caller's mtr object but release any acquired page latches before
returning.
btr_cur_get_page(), page_cur_get_page(): Do not invoke page_align().
row_purge_remove_sec_if_poss_leaf(): Return the value of PAGE_MAX_TRX_ID
to be checked against the page in row_purge_remove_sec_if_poss_tree().
If the secondary index page was not changed meanwhile, it will be
unnecessary to invoke row_purge_poss_sec() again.
trx_undo_prev_version_build(): Access any undo log pages using
the caller's mini-transaction object.
row_purge_vc_matches_cluster(): Moved to the only compilation unit that
needs it.
Reviewed by: Debarun Banerjee
1 year ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-34515: Contention between purge and workload
In a Sysbench oltp_update_index workload that involves 1 table,
a serious contention between the workload and the purge of history
was observed. This was the worst when the table contained only 1 record.
This turned out to be fixed by setting innodb_purge_batch_size=128,
which corresponds to the number of usable persistent rollback segments.
When we go above that, there would be contention between row_purge_poss_sec()
and the workload, typically on the clustered index page latch, sometimes
also on a secondary index page latch. It might be that with smaller
batches, trx_sys.history_size() will end up pausing all concurrent
transaction start/commit frequently enough so that purge will be able
to make some progress, so that there would be less contention on the
index page latches between purge and SQL execution.
In commit aa719b5010c929132b4460b78113fbd07497d9c8 (part of MDEV-32050)
the interpretation of the parameter innodb_purge_batch_size was slightly
changed. It would correspond to the maximum desired size of the
purge_sys.pages cache. Before that change, the parameter was referring to
a number of undo log pages, but the accounting might have been inaccurate.
To avoid a regression, we will reduce the default value to
innodb_purge_batch_size=127, which will also be compatible with
innodb_undo_tablespaces>1 (which will disable rollback segment 0).
Additionally, some logic in the purge and MVCC checks is simplified.
The purge tasks will make use of purge_sys.pages when accessing undo
log pages to find out if a secondary index record can be removed.
If an undo page needs to be looked up in buf_pool.page_hash, we will
merely buffer-fix it. This is correct, because the undo pages are
append-only in nature. Holding purge_sys.latch or purge_sys.end_latch
or the fact that the current thread is executing as a part of an
in-progress purge batch will prevent the contents of the undo page from
being freed and subsequently reused. The buffer-fix will prevent the
page from being evicted form the buffer pool. Thanks to this logic,
we can refer to the undo log record directly in the buffer pool page
and avoid copying the record.
buf_pool_t::page_fix(): Look up and buffer-fix a page. This is useful
for accessing undo log pages, which are append-only by nature.
There will be no need to deal with change buffer or ROW_FORMAT=COMPRESSED
in that case.
purge_sys_t::view_guard::view_guard(): Allow the type of guard to be
acquired: end_latch, latch, or no latch (in case we are a purge thread).
purge_sys_t::view_guard::get(): Read-only accessor to purge_sys.pages.
purge_sys_t::get_page(): Invoke buf_pool_t::page_fix().
row_vers_old_has_index_entry(): Replaced with row_purge_is_unsafe()
and row_undo_mod_sec_unsafe().
trx_undo_get_undo_rec(): Merged to trx_undo_prev_version_build().
row_purge_poss_sec(): Add the parameter mtr and remove redundant
or unused parameters sec_pcur, sec_mtr, is_tree. We will use the
caller's mtr object but release any acquired page latches before
returning.
btr_cur_get_page(), page_cur_get_page(): Do not invoke page_align().
row_purge_remove_sec_if_poss_leaf(): Return the value of PAGE_MAX_TRX_ID
to be checked against the page in row_purge_remove_sec_if_poss_tree().
If the secondary index page was not changed meanwhile, it will be
unnecessary to invoke row_purge_poss_sec() again.
trx_undo_prev_version_build(): Access any undo log pages using
the caller's mini-transaction object.
row_purge_vc_matches_cluster(): Moved to the only compilation unit that
needs it.
Reviewed by: Debarun Banerjee
1 year ago MDEV-34515: Contention between purge and workload
In a Sysbench oltp_update_index workload that involves 1 table,
a serious contention between the workload and the purge of history
was observed. This was the worst when the table contained only 1 record.
This turned out to be fixed by setting innodb_purge_batch_size=128,
which corresponds to the number of usable persistent rollback segments.
When we go above that, there would be contention between row_purge_poss_sec()
and the workload, typically on the clustered index page latch, sometimes
also on a secondary index page latch. It might be that with smaller
batches, trx_sys.history_size() will end up pausing all concurrent
transaction start/commit frequently enough so that purge will be able
to make some progress, so that there would be less contention on the
index page latches between purge and SQL execution.
In commit aa719b5010c929132b4460b78113fbd07497d9c8 (part of MDEV-32050)
the interpretation of the parameter innodb_purge_batch_size was slightly
changed. It would correspond to the maximum desired size of the
purge_sys.pages cache. Before that change, the parameter was referring to
a number of undo log pages, but the accounting might have been inaccurate.
To avoid a regression, we will reduce the default value to
innodb_purge_batch_size=127, which will also be compatible with
innodb_undo_tablespaces>1 (which will disable rollback segment 0).
Additionally, some logic in the purge and MVCC checks is simplified.
The purge tasks will make use of purge_sys.pages when accessing undo
log pages to find out if a secondary index record can be removed.
If an undo page needs to be looked up in buf_pool.page_hash, we will
merely buffer-fix it. This is correct, because the undo pages are
append-only in nature. Holding purge_sys.latch or purge_sys.end_latch
or the fact that the current thread is executing as a part of an
in-progress purge batch will prevent the contents of the undo page from
being freed and subsequently reused. The buffer-fix will prevent the
page from being evicted form the buffer pool. Thanks to this logic,
we can refer to the undo log record directly in the buffer pool page
and avoid copying the record.
buf_pool_t::page_fix(): Look up and buffer-fix a page. This is useful
for accessing undo log pages, which are append-only by nature.
There will be no need to deal with change buffer or ROW_FORMAT=COMPRESSED
in that case.
purge_sys_t::view_guard::view_guard(): Allow the type of guard to be
acquired: end_latch, latch, or no latch (in case we are a purge thread).
purge_sys_t::view_guard::get(): Read-only accessor to purge_sys.pages.
purge_sys_t::get_page(): Invoke buf_pool_t::page_fix().
row_vers_old_has_index_entry(): Replaced with row_purge_is_unsafe()
and row_undo_mod_sec_unsafe().
trx_undo_get_undo_rec(): Merged to trx_undo_prev_version_build().
row_purge_poss_sec(): Add the parameter mtr and remove redundant
or unused parameters sec_pcur, sec_mtr, is_tree. We will use the
caller's mtr object but release any acquired page latches before
returning.
btr_cur_get_page(), page_cur_get_page(): Do not invoke page_align().
row_purge_remove_sec_if_poss_leaf(): Return the value of PAGE_MAX_TRX_ID
to be checked against the page in row_purge_remove_sec_if_poss_tree().
If the secondary index page was not changed meanwhile, it will be
unnecessary to invoke row_purge_poss_sec() again.
trx_undo_prev_version_build(): Access any undo log pages using
the caller's mini-transaction object.
row_purge_vc_matches_cluster(): Moved to the only compilation unit that
needs it.
Reviewed by: Debarun Banerjee
1 year ago MDEV-34515: Contention between purge and workload
In a Sysbench oltp_update_index workload that involves 1 table,
a serious contention between the workload and the purge of history
was observed. This was the worst when the table contained only 1 record.
This turned out to be fixed by setting innodb_purge_batch_size=128,
which corresponds to the number of usable persistent rollback segments.
When we go above that, there would be contention between row_purge_poss_sec()
and the workload, typically on the clustered index page latch, sometimes
also on a secondary index page latch. It might be that with smaller
batches, trx_sys.history_size() will end up pausing all concurrent
transaction start/commit frequently enough so that purge will be able
to make some progress, so that there would be less contention on the
index page latches between purge and SQL execution.
In commit aa719b5010c929132b4460b78113fbd07497d9c8 (part of MDEV-32050)
the interpretation of the parameter innodb_purge_batch_size was slightly
changed. It would correspond to the maximum desired size of the
purge_sys.pages cache. Before that change, the parameter was referring to
a number of undo log pages, but the accounting might have been inaccurate.
To avoid a regression, we will reduce the default value to
innodb_purge_batch_size=127, which will also be compatible with
innodb_undo_tablespaces>1 (which will disable rollback segment 0).
Additionally, some logic in the purge and MVCC checks is simplified.
The purge tasks will make use of purge_sys.pages when accessing undo
log pages to find out if a secondary index record can be removed.
If an undo page needs to be looked up in buf_pool.page_hash, we will
merely buffer-fix it. This is correct, because the undo pages are
append-only in nature. Holding purge_sys.latch or purge_sys.end_latch
or the fact that the current thread is executing as a part of an
in-progress purge batch will prevent the contents of the undo page from
being freed and subsequently reused. The buffer-fix will prevent the
page from being evicted form the buffer pool. Thanks to this logic,
we can refer to the undo log record directly in the buffer pool page
and avoid copying the record.
buf_pool_t::page_fix(): Look up and buffer-fix a page. This is useful
for accessing undo log pages, which are append-only by nature.
There will be no need to deal with change buffer or ROW_FORMAT=COMPRESSED
in that case.
purge_sys_t::view_guard::view_guard(): Allow the type of guard to be
acquired: end_latch, latch, or no latch (in case we are a purge thread).
purge_sys_t::view_guard::get(): Read-only accessor to purge_sys.pages.
purge_sys_t::get_page(): Invoke buf_pool_t::page_fix().
row_vers_old_has_index_entry(): Replaced with row_purge_is_unsafe()
and row_undo_mod_sec_unsafe().
trx_undo_get_undo_rec(): Merged to trx_undo_prev_version_build().
row_purge_poss_sec(): Add the parameter mtr and remove redundant
or unused parameters sec_pcur, sec_mtr, is_tree. We will use the
caller's mtr object but release any acquired page latches before
returning.
btr_cur_get_page(), page_cur_get_page(): Do not invoke page_align().
row_purge_remove_sec_if_poss_leaf(): Return the value of PAGE_MAX_TRX_ID
to be checked against the page in row_purge_remove_sec_if_poss_tree().
If the secondary index page was not changed meanwhile, it will be
unnecessary to invoke row_purge_poss_sec() again.
trx_undo_prev_version_build(): Access any undo log pages using
the caller's mini-transaction object.
row_purge_vc_matches_cluster(): Moved to the only compilation unit that
needs it.
Reviewed by: Debarun Banerjee
1 year ago MDEV-34515: Contention between purge and workload
In a Sysbench oltp_update_index workload that involves 1 table,
a serious contention between the workload and the purge of history
was observed. This was the worst when the table contained only 1 record.
This turned out to be fixed by setting innodb_purge_batch_size=128,
which corresponds to the number of usable persistent rollback segments.
When we go above that, there would be contention between row_purge_poss_sec()
and the workload, typically on the clustered index page latch, sometimes
also on a secondary index page latch. It might be that with smaller
batches, trx_sys.history_size() will end up pausing all concurrent
transaction start/commit frequently enough so that purge will be able
to make some progress, so that there would be less contention on the
index page latches between purge and SQL execution.
In commit aa719b5010c929132b4460b78113fbd07497d9c8 (part of MDEV-32050)
the interpretation of the parameter innodb_purge_batch_size was slightly
changed. It would correspond to the maximum desired size of the
purge_sys.pages cache. Before that change, the parameter was referring to
a number of undo log pages, but the accounting might have been inaccurate.
To avoid a regression, we will reduce the default value to
innodb_purge_batch_size=127, which will also be compatible with
innodb_undo_tablespaces>1 (which will disable rollback segment 0).
Additionally, some logic in the purge and MVCC checks is simplified.
The purge tasks will make use of purge_sys.pages when accessing undo
log pages to find out if a secondary index record can be removed.
If an undo page needs to be looked up in buf_pool.page_hash, we will
merely buffer-fix it. This is correct, because the undo pages are
append-only in nature. Holding purge_sys.latch or purge_sys.end_latch
or the fact that the current thread is executing as a part of an
in-progress purge batch will prevent the contents of the undo page from
being freed and subsequently reused. The buffer-fix will prevent the
page from being evicted form the buffer pool. Thanks to this logic,
we can refer to the undo log record directly in the buffer pool page
and avoid copying the record.
buf_pool_t::page_fix(): Look up and buffer-fix a page. This is useful
for accessing undo log pages, which are append-only by nature.
There will be no need to deal with change buffer or ROW_FORMAT=COMPRESSED
in that case.
purge_sys_t::view_guard::view_guard(): Allow the type of guard to be
acquired: end_latch, latch, or no latch (in case we are a purge thread).
purge_sys_t::view_guard::get(): Read-only accessor to purge_sys.pages.
purge_sys_t::get_page(): Invoke buf_pool_t::page_fix().
row_vers_old_has_index_entry(): Replaced with row_purge_is_unsafe()
and row_undo_mod_sec_unsafe().
trx_undo_get_undo_rec(): Merged to trx_undo_prev_version_build().
row_purge_poss_sec(): Add the parameter mtr and remove redundant
or unused parameters sec_pcur, sec_mtr, is_tree. We will use the
caller's mtr object but release any acquired page latches before
returning.
btr_cur_get_page(), page_cur_get_page(): Do not invoke page_align().
row_purge_remove_sec_if_poss_leaf(): Return the value of PAGE_MAX_TRX_ID
to be checked against the page in row_purge_remove_sec_if_poss_tree().
If the secondary index page was not changed meanwhile, it will be
unnecessary to invoke row_purge_poss_sec() again.
trx_undo_prev_version_build(): Access any undo log pages using
the caller's mini-transaction object.
row_purge_vc_matches_cluster(): Moved to the only compilation unit that
needs it.
Reviewed by: Debarun Banerjee
1 year ago MDEV-34515: Contention between purge and workload
In a Sysbench oltp_update_index workload that involves 1 table,
a serious contention between the workload and the purge of history
was observed. This was the worst when the table contained only 1 record.
This turned out to be fixed by setting innodb_purge_batch_size=128,
which corresponds to the number of usable persistent rollback segments.
When we go above that, there would be contention between row_purge_poss_sec()
and the workload, typically on the clustered index page latch, sometimes
also on a secondary index page latch. It might be that with smaller
batches, trx_sys.history_size() will end up pausing all concurrent
transaction start/commit frequently enough so that purge will be able
to make some progress, so that there would be less contention on the
index page latches between purge and SQL execution.
In commit aa719b5010c929132b4460b78113fbd07497d9c8 (part of MDEV-32050)
the interpretation of the parameter innodb_purge_batch_size was slightly
changed. It would correspond to the maximum desired size of the
purge_sys.pages cache. Before that change, the parameter was referring to
a number of undo log pages, but the accounting might have been inaccurate.
To avoid a regression, we will reduce the default value to
innodb_purge_batch_size=127, which will also be compatible with
innodb_undo_tablespaces>1 (which will disable rollback segment 0).
Additionally, some logic in the purge and MVCC checks is simplified.
The purge tasks will make use of purge_sys.pages when accessing undo
log pages to find out if a secondary index record can be removed.
If an undo page needs to be looked up in buf_pool.page_hash, we will
merely buffer-fix it. This is correct, because the undo pages are
append-only in nature. Holding purge_sys.latch or purge_sys.end_latch
or the fact that the current thread is executing as a part of an
in-progress purge batch will prevent the contents of the undo page from
being freed and subsequently reused. The buffer-fix will prevent the
page from being evicted form the buffer pool. Thanks to this logic,
we can refer to the undo log record directly in the buffer pool page
and avoid copying the record.
buf_pool_t::page_fix(): Look up and buffer-fix a page. This is useful
for accessing undo log pages, which are append-only by nature.
There will be no need to deal with change buffer or ROW_FORMAT=COMPRESSED
in that case.
purge_sys_t::view_guard::view_guard(): Allow the type of guard to be
acquired: end_latch, latch, or no latch (in case we are a purge thread).
purge_sys_t::view_guard::get(): Read-only accessor to purge_sys.pages.
purge_sys_t::get_page(): Invoke buf_pool_t::page_fix().
row_vers_old_has_index_entry(): Replaced with row_purge_is_unsafe()
and row_undo_mod_sec_unsafe().
trx_undo_get_undo_rec(): Merged to trx_undo_prev_version_build().
row_purge_poss_sec(): Add the parameter mtr and remove redundant
or unused parameters sec_pcur, sec_mtr, is_tree. We will use the
caller's mtr object but release any acquired page latches before
returning.
btr_cur_get_page(), page_cur_get_page(): Do not invoke page_align().
row_purge_remove_sec_if_poss_leaf(): Return the value of PAGE_MAX_TRX_ID
to be checked against the page in row_purge_remove_sec_if_poss_tree().
If the secondary index page was not changed meanwhile, it will be
unnecessary to invoke row_purge_poss_sec() again.
trx_undo_prev_version_build(): Access any undo log pages using
the caller's mini-transaction object.
row_purge_vc_matches_cluster(): Moved to the only compilation unit that
needs it.
Reviewed by: Debarun Banerjee
1 year ago MDEV-34515: Contention between purge and workload
In a Sysbench oltp_update_index workload that involves 1 table,
a serious contention between the workload and the purge of history
was observed. This was the worst when the table contained only 1 record.
This turned out to be fixed by setting innodb_purge_batch_size=128,
which corresponds to the number of usable persistent rollback segments.
When we go above that, there would be contention between row_purge_poss_sec()
and the workload, typically on the clustered index page latch, sometimes
also on a secondary index page latch. It might be that with smaller
batches, trx_sys.history_size() will end up pausing all concurrent
transaction start/commit frequently enough so that purge will be able
to make some progress, so that there would be less contention on the
index page latches between purge and SQL execution.
In commit aa719b5010c929132b4460b78113fbd07497d9c8 (part of MDEV-32050)
the interpretation of the parameter innodb_purge_batch_size was slightly
changed. It would correspond to the maximum desired size of the
purge_sys.pages cache. Before that change, the parameter was referring to
a number of undo log pages, but the accounting might have been inaccurate.
To avoid a regression, we will reduce the default value to
innodb_purge_batch_size=127, which will also be compatible with
innodb_undo_tablespaces>1 (which will disable rollback segment 0).
Additionally, some logic in the purge and MVCC checks is simplified.
The purge tasks will make use of purge_sys.pages when accessing undo
log pages to find out if a secondary index record can be removed.
If an undo page needs to be looked up in buf_pool.page_hash, we will
merely buffer-fix it. This is correct, because the undo pages are
append-only in nature. Holding purge_sys.latch or purge_sys.end_latch
or the fact that the current thread is executing as a part of an
in-progress purge batch will prevent the contents of the undo page from
being freed and subsequently reused. The buffer-fix will prevent the
page from being evicted form the buffer pool. Thanks to this logic,
we can refer to the undo log record directly in the buffer pool page
and avoid copying the record.
buf_pool_t::page_fix(): Look up and buffer-fix a page. This is useful
for accessing undo log pages, which are append-only by nature.
There will be no need to deal with change buffer or ROW_FORMAT=COMPRESSED
in that case.
purge_sys_t::view_guard::view_guard(): Allow the type of guard to be
acquired: end_latch, latch, or no latch (in case we are a purge thread).
purge_sys_t::view_guard::get(): Read-only accessor to purge_sys.pages.
purge_sys_t::get_page(): Invoke buf_pool_t::page_fix().
row_vers_old_has_index_entry(): Replaced with row_purge_is_unsafe()
and row_undo_mod_sec_unsafe().
trx_undo_get_undo_rec(): Merged to trx_undo_prev_version_build().
row_purge_poss_sec(): Add the parameter mtr and remove redundant
or unused parameters sec_pcur, sec_mtr, is_tree. We will use the
caller's mtr object but release any acquired page latches before
returning.
btr_cur_get_page(), page_cur_get_page(): Do not invoke page_align().
row_purge_remove_sec_if_poss_leaf(): Return the value of PAGE_MAX_TRX_ID
to be checked against the page in row_purge_remove_sec_if_poss_tree().
If the secondary index page was not changed meanwhile, it will be
unnecessary to invoke row_purge_poss_sec() again.
trx_undo_prev_version_build(): Access any undo log pages using
the caller's mini-transaction object.
row_purge_vc_matches_cluster(): Moved to the only compilation unit that
needs it.
Reviewed by: Debarun Banerjee
1 year ago MDEV-34515: Contention between purge and workload
In a Sysbench oltp_update_index workload that involves 1 table,
a serious contention between the workload and the purge of history
was observed. This was the worst when the table contained only 1 record.
This turned out to be fixed by setting innodb_purge_batch_size=128,
which corresponds to the number of usable persistent rollback segments.
When we go above that, there would be contention between row_purge_poss_sec()
and the workload, typically on the clustered index page latch, sometimes
also on a secondary index page latch. It might be that with smaller
batches, trx_sys.history_size() will end up pausing all concurrent
transaction start/commit frequently enough so that purge will be able
to make some progress, so that there would be less contention on the
index page latches between purge and SQL execution.
In commit aa719b5010c929132b4460b78113fbd07497d9c8 (part of MDEV-32050)
the interpretation of the parameter innodb_purge_batch_size was slightly
changed. It would correspond to the maximum desired size of the
purge_sys.pages cache. Before that change, the parameter was referring to
a number of undo log pages, but the accounting might have been inaccurate.
To avoid a regression, we will reduce the default value to
innodb_purge_batch_size=127, which will also be compatible with
innodb_undo_tablespaces>1 (which will disable rollback segment 0).
Additionally, some logic in the purge and MVCC checks is simplified.
The purge tasks will make use of purge_sys.pages when accessing undo
log pages to find out if a secondary index record can be removed.
If an undo page needs to be looked up in buf_pool.page_hash, we will
merely buffer-fix it. This is correct, because the undo pages are
append-only in nature. Holding purge_sys.latch or purge_sys.end_latch
or the fact that the current thread is executing as a part of an
in-progress purge batch will prevent the contents of the undo page from
being freed and subsequently reused. The buffer-fix will prevent the
page from being evicted form the buffer pool. Thanks to this logic,
we can refer to the undo log record directly in the buffer pool page
and avoid copying the record.
buf_pool_t::page_fix(): Look up and buffer-fix a page. This is useful
for accessing undo log pages, which are append-only by nature.
There will be no need to deal with change buffer or ROW_FORMAT=COMPRESSED
in that case.
purge_sys_t::view_guard::view_guard(): Allow the type of guard to be
acquired: end_latch, latch, or no latch (in case we are a purge thread).
purge_sys_t::view_guard::get(): Read-only accessor to purge_sys.pages.
purge_sys_t::get_page(): Invoke buf_pool_t::page_fix().
row_vers_old_has_index_entry(): Replaced with row_purge_is_unsafe()
and row_undo_mod_sec_unsafe().
trx_undo_get_undo_rec(): Merged to trx_undo_prev_version_build().
row_purge_poss_sec(): Add the parameter mtr and remove redundant
or unused parameters sec_pcur, sec_mtr, is_tree. We will use the
caller's mtr object but release any acquired page latches before
returning.
btr_cur_get_page(), page_cur_get_page(): Do not invoke page_align().
row_purge_remove_sec_if_poss_leaf(): Return the value of PAGE_MAX_TRX_ID
to be checked against the page in row_purge_remove_sec_if_poss_tree().
If the secondary index page was not changed meanwhile, it will be
unnecessary to invoke row_purge_poss_sec() again.
trx_undo_prev_version_build(): Access any undo log pages using
the caller's mini-transaction object.
row_purge_vc_matches_cluster(): Moved to the only compilation unit that
needs it.
Reviewed by: Debarun Banerjee
1 year ago MDEV-34515: Contention between purge and workload
In a Sysbench oltp_update_index workload that involves 1 table,
a serious contention between the workload and the purge of history
was observed. This was the worst when the table contained only 1 record.
This turned out to be fixed by setting innodb_purge_batch_size=128,
which corresponds to the number of usable persistent rollback segments.
When we go above that, there would be contention between row_purge_poss_sec()
and the workload, typically on the clustered index page latch, sometimes
also on a secondary index page latch. It might be that with smaller
batches, trx_sys.history_size() will end up pausing all concurrent
transaction start/commit frequently enough so that purge will be able
to make some progress, so that there would be less contention on the
index page latches between purge and SQL execution.
In commit aa719b5010c929132b4460b78113fbd07497d9c8 (part of MDEV-32050)
the interpretation of the parameter innodb_purge_batch_size was slightly
changed. It would correspond to the maximum desired size of the
purge_sys.pages cache. Before that change, the parameter was referring to
a number of undo log pages, but the accounting might have been inaccurate.
To avoid a regression, we will reduce the default value to
innodb_purge_batch_size=127, which will also be compatible with
innodb_undo_tablespaces>1 (which will disable rollback segment 0).
Additionally, some logic in the purge and MVCC checks is simplified.
The purge tasks will make use of purge_sys.pages when accessing undo
log pages to find out if a secondary index record can be removed.
If an undo page needs to be looked up in buf_pool.page_hash, we will
merely buffer-fix it. This is correct, because the undo pages are
append-only in nature. Holding purge_sys.latch or purge_sys.end_latch
or the fact that the current thread is executing as a part of an
in-progress purge batch will prevent the contents of the undo page from
being freed and subsequently reused. The buffer-fix will prevent the
page from being evicted form the buffer pool. Thanks to this logic,
we can refer to the undo log record directly in the buffer pool page
and avoid copying the record.
buf_pool_t::page_fix(): Look up and buffer-fix a page. This is useful
for accessing undo log pages, which are append-only by nature.
There will be no need to deal with change buffer or ROW_FORMAT=COMPRESSED
in that case.
purge_sys_t::view_guard::view_guard(): Allow the type of guard to be
acquired: end_latch, latch, or no latch (in case we are a purge thread).
purge_sys_t::view_guard::get(): Read-only accessor to purge_sys.pages.
purge_sys_t::get_page(): Invoke buf_pool_t::page_fix().
row_vers_old_has_index_entry(): Replaced with row_purge_is_unsafe()
and row_undo_mod_sec_unsafe().
trx_undo_get_undo_rec(): Merged to trx_undo_prev_version_build().
row_purge_poss_sec(): Add the parameter mtr and remove redundant
or unused parameters sec_pcur, sec_mtr, is_tree. We will use the
caller's mtr object but release any acquired page latches before
returning.
btr_cur_get_page(), page_cur_get_page(): Do not invoke page_align().
row_purge_remove_sec_if_poss_leaf(): Return the value of PAGE_MAX_TRX_ID
to be checked against the page in row_purge_remove_sec_if_poss_tree().
If the secondary index page was not changed meanwhile, it will be
unnecessary to invoke row_purge_poss_sec() again.
trx_undo_prev_version_build(): Access any undo log pages using
the caller's mini-transaction object.
row_purge_vc_matches_cluster(): Moved to the only compilation unit that
needs it.
Reviewed by: Debarun Banerjee
1 year ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-19514 Defer change buffer merge until pages are requested
We will remove the InnoDB background operation of merging buffered
changes to secondary index leaf pages. Changes will only be merged as a
result of an operation that accesses a secondary index leaf page,
such as a SQL statement that performs a lookup via that index,
or is modifying the index. Also ROLLBACK and some background operations,
such as purging the history of committed transactions, or computing
index cardinality statistics, can cause change buffer merge.
Encryption key rotation will not perform change buffer merge.
The motivation of this change is to simplify the I/O logic and to
allow crash recovery to happen in the background (MDEV-14481).
We also hope that this will reduce the number of "mystery" crashes
due to corrupted data. Because change buffer merge will typically
take place as a result of executing SQL statements, there should be
a clearer connection between the crash and the SQL statements that
were executed when the server crashed.
In many cases, a slight performance improvement was observed.
This is joint work with Thirunarayanan Balathandayuthapani
and was tested by Axel Schwenke and Matthias Leich.
The InnoDB monitor counter innodb_ibuf_merge_usec will be removed.
On slow shutdown (innodb_fast_shutdown=0), we will continue to
merge all buffered changes (and purge all undo log history).
Two InnoDB configuration parameters will be changed as follows:
innodb_disable_background_merge: Removed.
This parameter existed only in debug builds.
All change buffer merges will use synchronous reads.
innodb_force_recovery will be changed as follows:
* innodb_force_recovery=4 will be the same as innodb_force_recovery=3
(the change buffer merge cannot be disabled; it can only happen as
a result of an operation that accesses a secondary index leaf page).
The option used to be capable of corrupting secondary index leaf pages.
Now that capability is removed, and innodb_force_recovery=4 becomes 'safe'.
* innodb_force_recovery=5 (which essentially hard-wires
SET GLOBAL TRANSACTION ISOLATION LEVEL READ UNCOMMITTED)
becomes safe to use. Bogus data can be returned to SQL, but
persistent InnoDB data files will not be corrupted further.
* innodb_force_recovery=6 (ignore the redo log files)
will be the only option that can potentially cause
persistent corruption of InnoDB data files.
Code changes:
buf_page_t::ibuf_exist: New flag, to indicate whether buffered
changes exist for a buffer pool page. Pages with pending changes
can be returned by buf_page_get_gen(). Previously, the changes
were always merged inside buf_page_get_gen() if needed.
ibuf_page_exists(const buf_page_t&): Check if a buffered changes
exist for an X-latched or read-fixed page.
buf_page_get_gen(): Add the parameter allow_ibuf_merge=false.
All callers that know that they may be accessing a secondary index
leaf page must pass this parameter as allow_ibuf_merge=true,
unless it does not matter for that caller whether all buffered
changes have been applied. Assert that whenever allow_ibuf_merge
holds, the page actually is a leaf page. Attempt change buffer
merge only to secondary B-tree index leaf pages.
btr_block_get(): Add parameter 'bool merge'.
All callers of btr_block_get() should know whether the page could be
a secondary index leaf page. If it is not, we should avoid consulting
the change buffer bitmap to even consider a merge. This is the main
interface to requesting index pages from the buffer pool.
ibuf_merge_or_delete_for_page(), recv_recover_page(): Replace
buf_page_get_known_nowait() with much simpler logic, because
it is now guaranteed that that the block is x-latched or read-fixed.
mlog_init_t::mark_ibuf_exist(): Renamed from mlog_init_t::ibuf_merge().
On crash recovery, we will no longer merge any buffered changes
for the pages that we read into the buffer pool during the last batch
of applying log records.
buf_page_get_gen_known_nowait(), BUF_MAKE_YOUNG, BUF_KEEP_OLD: Remove.
btr_search_guess_on_hash(): Merge buf_page_get_gen_known_nowait()
to its only remaining caller.
buf_page_make_young_if_needed(): Define as an inline function.
Add the parameter buf_pool.
buf_page_peek_if_young(), buf_page_peek_if_too_old(): Add the
parameter buf_pool.
fil_space_validate_for_mtr_commit(): Remove a bogus comment
about background merge of the change buffer.
btr_cur_open_at_rnd_pos_func(), btr_cur_search_to_nth_level_func(),
btr_cur_open_at_index_side_func(): Use narrower data types and scopes.
ibuf_read_merge_pages(): Replaces buf_read_ibuf_merge_pages().
Merge the change buffer by invoking buf_page_get_gen().
6 years ago MDEV-19514 Defer change buffer merge until pages are requested
We will remove the InnoDB background operation of merging buffered
changes to secondary index leaf pages. Changes will only be merged as a
result of an operation that accesses a secondary index leaf page,
such as a SQL statement that performs a lookup via that index,
or is modifying the index. Also ROLLBACK and some background operations,
such as purging the history of committed transactions, or computing
index cardinality statistics, can cause change buffer merge.
Encryption key rotation will not perform change buffer merge.
The motivation of this change is to simplify the I/O logic and to
allow crash recovery to happen in the background (MDEV-14481).
We also hope that this will reduce the number of "mystery" crashes
due to corrupted data. Because change buffer merge will typically
take place as a result of executing SQL statements, there should be
a clearer connection between the crash and the SQL statements that
were executed when the server crashed.
In many cases, a slight performance improvement was observed.
This is joint work with Thirunarayanan Balathandayuthapani
and was tested by Axel Schwenke and Matthias Leich.
The InnoDB monitor counter innodb_ibuf_merge_usec will be removed.
On slow shutdown (innodb_fast_shutdown=0), we will continue to
merge all buffered changes (and purge all undo log history).
Two InnoDB configuration parameters will be changed as follows:
innodb_disable_background_merge: Removed.
This parameter existed only in debug builds.
All change buffer merges will use synchronous reads.
innodb_force_recovery will be changed as follows:
* innodb_force_recovery=4 will be the same as innodb_force_recovery=3
(the change buffer merge cannot be disabled; it can only happen as
a result of an operation that accesses a secondary index leaf page).
The option used to be capable of corrupting secondary index leaf pages.
Now that capability is removed, and innodb_force_recovery=4 becomes 'safe'.
* innodb_force_recovery=5 (which essentially hard-wires
SET GLOBAL TRANSACTION ISOLATION LEVEL READ UNCOMMITTED)
becomes safe to use. Bogus data can be returned to SQL, but
persistent InnoDB data files will not be corrupted further.
* innodb_force_recovery=6 (ignore the redo log files)
will be the only option that can potentially cause
persistent corruption of InnoDB data files.
Code changes:
buf_page_t::ibuf_exist: New flag, to indicate whether buffered
changes exist for a buffer pool page. Pages with pending changes
can be returned by buf_page_get_gen(). Previously, the changes
were always merged inside buf_page_get_gen() if needed.
ibuf_page_exists(const buf_page_t&): Check if a buffered changes
exist for an X-latched or read-fixed page.
buf_page_get_gen(): Add the parameter allow_ibuf_merge=false.
All callers that know that they may be accessing a secondary index
leaf page must pass this parameter as allow_ibuf_merge=true,
unless it does not matter for that caller whether all buffered
changes have been applied. Assert that whenever allow_ibuf_merge
holds, the page actually is a leaf page. Attempt change buffer
merge only to secondary B-tree index leaf pages.
btr_block_get(): Add parameter 'bool merge'.
All callers of btr_block_get() should know whether the page could be
a secondary index leaf page. If it is not, we should avoid consulting
the change buffer bitmap to even consider a merge. This is the main
interface to requesting index pages from the buffer pool.
ibuf_merge_or_delete_for_page(), recv_recover_page(): Replace
buf_page_get_known_nowait() with much simpler logic, because
it is now guaranteed that that the block is x-latched or read-fixed.
mlog_init_t::mark_ibuf_exist(): Renamed from mlog_init_t::ibuf_merge().
On crash recovery, we will no longer merge any buffered changes
for the pages that we read into the buffer pool during the last batch
of applying log records.
buf_page_get_gen_known_nowait(), BUF_MAKE_YOUNG, BUF_KEEP_OLD: Remove.
btr_search_guess_on_hash(): Merge buf_page_get_gen_known_nowait()
to its only remaining caller.
buf_page_make_young_if_needed(): Define as an inline function.
Add the parameter buf_pool.
buf_page_peek_if_young(), buf_page_peek_if_too_old(): Add the
parameter buf_pool.
fil_space_validate_for_mtr_commit(): Remove a bogus comment
about background merge of the change buffer.
btr_cur_open_at_rnd_pos_func(), btr_cur_search_to_nth_level_func(),
btr_cur_open_at_index_side_func(): Use narrower data types and scopes.
ibuf_read_merge_pages(): Replaces buf_read_ibuf_merge_pages().
Merge the change buffer by invoking buf_page_get_gen().
6 years ago MDEV-13542: Crashing on corrupted page is unhelpful
The approach to handling corruption that was chosen by Oracle in
commit 177d8b0c125b841c0650d27d735e3b87509dc286
is not really useful. Not only did it actually fail to prevent InnoDB
from crashing, but it is making things worse by blocking attempts to
rescue data from or rebuild a partially readable table.
We will try to prevent crashes in a different way: by propagating
errors up the call stack. We will never mark the clustered index
persistently corrupted, so that data recovery may be attempted by
reading from the table, or by rebuilding the table.
This should also fix MDEV-13680 (crash on btr_page_alloc() failure);
it was extensively tested with innodb_file_per_table=0 and a
non-autoextend system tablespace.
We should now avoid crashes in many cases, such as when a page
cannot be read or allocated, or an inconsistency is detected when
attempting to update multiple pages. We will not crash on double-free,
such as on the recovery of DDL in system tablespace in case something
was corrupted.
Crashes on corrupted data are still possible. The fault injection mechanism
that is introduced in the subsequent commit may help catch more of them.
buf_page_import_corrupt_failure: Remove the fault injection, and instead
corrupt some pages using Perl code in the tests.
btr_cur_pessimistic_insert(): Always reserve extents (except for the
change buffer), in order to prevent a subsequent allocation failure.
btr_pcur_open_at_rnd_pos(): Merged to the only caller ibuf_merge_pages().
btr_assert_not_corrupted(), btr_corruption_report(): Remove.
Similar checks are already part of btr_block_get().
FSEG_MAGIC_N_BYTES: Replaces FSEG_MAGIC_N_VALUE.
dict_hdr_get(), trx_rsegf_get_new(), trx_undo_page_get(),
trx_undo_page_get_s_latched(): Replaced with error-checking calls.
trx_rseg_t::get(mtr_t*): Replaces trx_rsegf_get().
trx_rseg_header_create(): Let the caller update the TRX_SYS page if needed.
trx_sys_create_sys_pages(): Merged with trx_sysf_create().
dict_check_tablespaces_and_store_max_id(): Do not access
DICT_HDR_MAX_SPACE_ID, because it was already recovered in dict_boot().
Merge dict_check_sys_tables() with this function.
dir_pathname(): Replaces os_file_make_new_pathname().
row_undo_ins_remove_sec(): Do not modify the undo page by adding
a terminating NUL byte to the record.
btr_decryption_failed(): Report decryption failures
dict_set_corrupted_by_space(), dict_set_encrypted_by_space(),
dict_set_corrupted_index_cache_only(): Remove.
dict_set_corrupted(): Remove the constant parameter dict_locked=false.
Never flag the clustered index corrupted in SYS_INDEXES, because
that would deny further access to the table. It might be possible to
repair the table by executing ALTER TABLE or OPTIMIZE TABLE, in case
no B-tree leaf page is corrupted.
dict_table_skip_corrupt_index(), dict_table_next_uncorrupted_index(),
row_purge_skip_uncommitted_virtual_index(): Remove, and refactor
the callers to read dict_index_t::type only once.
dict_table_is_corrupted(): Remove.
dict_index_t::is_btree(): Determine if the index is a valid B-tree.
BUF_GET_NO_LATCH, BUF_EVICT_IF_IN_POOL: Remove.
UNIV_BTR_DEBUG: Remove. Any inconsistency will no longer trigger
assertion failures, but error codes being returned.
buf_corrupt_page_release(): Replaced with a direct call to
buf_pool.corrupted_evict().
fil_invalid_page_access_msg(): Never crash on an invalid read;
let the caller of buf_page_get_gen() decide.
btr_pcur_t::restore_position(): Propagate failure status to the caller
by returning CORRUPTED.
opt_search_plan_for_table(): Simplify the code.
row_purge_del_mark(), row_purge_upd_exist_or_extern_func(),
row_undo_ins_remove_sec_rec(), row_undo_mod_upd_del_sec(),
row_undo_mod_del_mark_sec(): Avoid mem_heap_create()/mem_heap_free()
when no secondary indexes exist.
row_undo_mod_upd_exist_sec(): Simplify the code.
row_upd_clust_step(), dict_load_table_one(): Return DB_TABLE_CORRUPT
if the clustered index (and therefore the table) is corrupted, similar
to what we do in row_insert_for_mysql().
fut_get_ptr(): Replace with buf_page_get_gen() calls.
buf_page_get_gen(): Return nullptr and *err=DB_CORRUPTION
if the page is marked as freed. For other modes than
BUF_GET_POSSIBLY_FREED or BUF_PEEK_IF_IN_POOL this will
trigger a debug assertion failure. For BUF_GET_POSSIBLY_FREED,
we will return nullptr for freed pages, so that the callers
can be simplified. The purge of transaction history will be
a new user of BUF_GET_POSSIBLY_FREED, to avoid crashes on
corrupted data.
buf_page_get_low(): Never crash on a corrupted page, but simply
return nullptr.
fseg_page_is_allocated(): Replaces fseg_page_is_free().
fts_drop_common_tables(): Return an error if the transaction
was rolled back.
fil_space_t::set_corrupted(): Report a tablespace as corrupted if
it was not reported already.
fil_space_t::io(): Invoke fil_space_t::set_corrupted() to report
out-of-bounds page access or other errors.
Clean up mtr_t::page_lock()
buf_page_get_low(): Validate the page identifier (to check for
recently read corrupted pages) after acquiring the page latch.
buf_page_t::read_complete(): Flag uninitialized (all-zero) pages
with DB_FAIL. Return DB_PAGE_CORRUPTED on page number mismatch.
mtr_t::defer_drop_ahi(): Renamed from mtr_defer_drop_ahi().
recv_sys_t::free_corrupted_page(): Only set_corrupt_fs()
if any log records exist for the page. We do not mind if read-ahead
produces corrupted (or all-zero) pages that were not actually needed
during recovery.
recv_recover_page(): Return whether the operation succeeded.
recv_sys_t::recover_low(): Simplify the logic. Check for recovery error.
Thanks to Matthias Leich for testing this extensively and to the
authors of https://rr-project.org for making it easy to diagnose
and fix any failures that were found during the testing.
3 years ago MDEV-19514 Defer change buffer merge until pages are requested
We will remove the InnoDB background operation of merging buffered
changes to secondary index leaf pages. Changes will only be merged as a
result of an operation that accesses a secondary index leaf page,
such as a SQL statement that performs a lookup via that index,
or is modifying the index. Also ROLLBACK and some background operations,
such as purging the history of committed transactions, or computing
index cardinality statistics, can cause change buffer merge.
Encryption key rotation will not perform change buffer merge.
The motivation of this change is to simplify the I/O logic and to
allow crash recovery to happen in the background (MDEV-14481).
We also hope that this will reduce the number of "mystery" crashes
due to corrupted data. Because change buffer merge will typically
take place as a result of executing SQL statements, there should be
a clearer connection between the crash and the SQL statements that
were executed when the server crashed.
In many cases, a slight performance improvement was observed.
This is joint work with Thirunarayanan Balathandayuthapani
and was tested by Axel Schwenke and Matthias Leich.
The InnoDB monitor counter innodb_ibuf_merge_usec will be removed.
On slow shutdown (innodb_fast_shutdown=0), we will continue to
merge all buffered changes (and purge all undo log history).
Two InnoDB configuration parameters will be changed as follows:
innodb_disable_background_merge: Removed.
This parameter existed only in debug builds.
All change buffer merges will use synchronous reads.
innodb_force_recovery will be changed as follows:
* innodb_force_recovery=4 will be the same as innodb_force_recovery=3
(the change buffer merge cannot be disabled; it can only happen as
a result of an operation that accesses a secondary index leaf page).
The option used to be capable of corrupting secondary index leaf pages.
Now that capability is removed, and innodb_force_recovery=4 becomes 'safe'.
* innodb_force_recovery=5 (which essentially hard-wires
SET GLOBAL TRANSACTION ISOLATION LEVEL READ UNCOMMITTED)
becomes safe to use. Bogus data can be returned to SQL, but
persistent InnoDB data files will not be corrupted further.
* innodb_force_recovery=6 (ignore the redo log files)
will be the only option that can potentially cause
persistent corruption of InnoDB data files.
Code changes:
buf_page_t::ibuf_exist: New flag, to indicate whether buffered
changes exist for a buffer pool page. Pages with pending changes
can be returned by buf_page_get_gen(). Previously, the changes
were always merged inside buf_page_get_gen() if needed.
ibuf_page_exists(const buf_page_t&): Check if a buffered changes
exist for an X-latched or read-fixed page.
buf_page_get_gen(): Add the parameter allow_ibuf_merge=false.
All callers that know that they may be accessing a secondary index
leaf page must pass this parameter as allow_ibuf_merge=true,
unless it does not matter for that caller whether all buffered
changes have been applied. Assert that whenever allow_ibuf_merge
holds, the page actually is a leaf page. Attempt change buffer
merge only to secondary B-tree index leaf pages.
btr_block_get(): Add parameter 'bool merge'.
All callers of btr_block_get() should know whether the page could be
a secondary index leaf page. If it is not, we should avoid consulting
the change buffer bitmap to even consider a merge. This is the main
interface to requesting index pages from the buffer pool.
ibuf_merge_or_delete_for_page(), recv_recover_page(): Replace
buf_page_get_known_nowait() with much simpler logic, because
it is now guaranteed that that the block is x-latched or read-fixed.
mlog_init_t::mark_ibuf_exist(): Renamed from mlog_init_t::ibuf_merge().
On crash recovery, we will no longer merge any buffered changes
for the pages that we read into the buffer pool during the last batch
of applying log records.
buf_page_get_gen_known_nowait(), BUF_MAKE_YOUNG, BUF_KEEP_OLD: Remove.
btr_search_guess_on_hash(): Merge buf_page_get_gen_known_nowait()
to its only remaining caller.
buf_page_make_young_if_needed(): Define as an inline function.
Add the parameter buf_pool.
buf_page_peek_if_young(), buf_page_peek_if_too_old(): Add the
parameter buf_pool.
fil_space_validate_for_mtr_commit(): Remove a bogus comment
about background merge of the change buffer.
btr_cur_open_at_rnd_pos_func(), btr_cur_search_to_nth_level_func(),
btr_cur_open_at_index_side_func(): Use narrower data types and scopes.
ibuf_read_merge_pages(): Replaces buf_read_ibuf_merge_pages().
Merge the change buffer by invoking buf_page_get_gen().
6 years ago MDEV-19514 Defer change buffer merge until pages are requested
We will remove the InnoDB background operation of merging buffered
changes to secondary index leaf pages. Changes will only be merged as a
result of an operation that accesses a secondary index leaf page,
such as a SQL statement that performs a lookup via that index,
or is modifying the index. Also ROLLBACK and some background operations,
such as purging the history of committed transactions, or computing
index cardinality statistics, can cause change buffer merge.
Encryption key rotation will not perform change buffer merge.
The motivation of this change is to simplify the I/O logic and to
allow crash recovery to happen in the background (MDEV-14481).
We also hope that this will reduce the number of "mystery" crashes
due to corrupted data. Because change buffer merge will typically
take place as a result of executing SQL statements, there should be
a clearer connection between the crash and the SQL statements that
were executed when the server crashed.
In many cases, a slight performance improvement was observed.
This is joint work with Thirunarayanan Balathandayuthapani
and was tested by Axel Schwenke and Matthias Leich.
The InnoDB monitor counter innodb_ibuf_merge_usec will be removed.
On slow shutdown (innodb_fast_shutdown=0), we will continue to
merge all buffered changes (and purge all undo log history).
Two InnoDB configuration parameters will be changed as follows:
innodb_disable_background_merge: Removed.
This parameter existed only in debug builds.
All change buffer merges will use synchronous reads.
innodb_force_recovery will be changed as follows:
* innodb_force_recovery=4 will be the same as innodb_force_recovery=3
(the change buffer merge cannot be disabled; it can only happen as
a result of an operation that accesses a secondary index leaf page).
The option used to be capable of corrupting secondary index leaf pages.
Now that capability is removed, and innodb_force_recovery=4 becomes 'safe'.
* innodb_force_recovery=5 (which essentially hard-wires
SET GLOBAL TRANSACTION ISOLATION LEVEL READ UNCOMMITTED)
becomes safe to use. Bogus data can be returned to SQL, but
persistent InnoDB data files will not be corrupted further.
* innodb_force_recovery=6 (ignore the redo log files)
will be the only option that can potentially cause
persistent corruption of InnoDB data files.
Code changes:
buf_page_t::ibuf_exist: New flag, to indicate whether buffered
changes exist for a buffer pool page. Pages with pending changes
can be returned by buf_page_get_gen(). Previously, the changes
were always merged inside buf_page_get_gen() if needed.
ibuf_page_exists(const buf_page_t&): Check if a buffered changes
exist for an X-latched or read-fixed page.
buf_page_get_gen(): Add the parameter allow_ibuf_merge=false.
All callers that know that they may be accessing a secondary index
leaf page must pass this parameter as allow_ibuf_merge=true,
unless it does not matter for that caller whether all buffered
changes have been applied. Assert that whenever allow_ibuf_merge
holds, the page actually is a leaf page. Attempt change buffer
merge only to secondary B-tree index leaf pages.
btr_block_get(): Add parameter 'bool merge'.
All callers of btr_block_get() should know whether the page could be
a secondary index leaf page. If it is not, we should avoid consulting
the change buffer bitmap to even consider a merge. This is the main
interface to requesting index pages from the buffer pool.
ibuf_merge_or_delete_for_page(), recv_recover_page(): Replace
buf_page_get_known_nowait() with much simpler logic, because
it is now guaranteed that that the block is x-latched or read-fixed.
mlog_init_t::mark_ibuf_exist(): Renamed from mlog_init_t::ibuf_merge().
On crash recovery, we will no longer merge any buffered changes
for the pages that we read into the buffer pool during the last batch
of applying log records.
buf_page_get_gen_known_nowait(), BUF_MAKE_YOUNG, BUF_KEEP_OLD: Remove.
btr_search_guess_on_hash(): Merge buf_page_get_gen_known_nowait()
to its only remaining caller.
buf_page_make_young_if_needed(): Define as an inline function.
Add the parameter buf_pool.
buf_page_peek_if_young(), buf_page_peek_if_too_old(): Add the
parameter buf_pool.
fil_space_validate_for_mtr_commit(): Remove a bogus comment
about background merge of the change buffer.
btr_cur_open_at_rnd_pos_func(), btr_cur_search_to_nth_level_func(),
btr_cur_open_at_index_side_func(): Use narrower data types and scopes.
ibuf_read_merge_pages(): Replaces buf_read_ibuf_merge_pages().
Merge the change buffer by invoking buf_page_get_gen().
6 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-26827 Make page flushing even faster
For more convenient monitoring of something that could greatly affect
the volume of page writes, we add the status variable
Innodb_buffer_pool_pages_split that was previously only available
via information_schema.innodb_metrics as "innodb_page_splits".
This was suggested by Axel Schwenke.
buf_flush_page_count: Replaced with buf_pool.stat.n_pages_written.
We protect buf_pool.stat (except n_page_gets) with buf_pool.mutex
and remove unnecessary export_vars indirection.
buf_pool.flush_list_bytes: Moved from buf_pool.stat.flush_list_bytes.
Protected by buf_pool.flush_list_mutex.
buf_pool_t::page_cleaner_status: Replaces buf_pool_t::n_flush_LRU_,
buf_pool_t::n_flush_list_, and buf_pool_t::page_cleaner_is_idle.
Protected by buf_pool.flush_list_mutex. We will exclusively broadcast
buf_pool.done_flush_list by the buf_flush_page_cleaner thread,
and only wait for it when communicating with buf_flush_page_cleaner.
There is no need to keep a count of pending writes by the
buf_pool.flush_list processing. A single flag suffices for that.
Waits for page write completion can be performed by
simply waiting on block->page.lock, or by invoking
buf_dblwr.wait_for_page_writes().
buf_LRU_block_free_non_file_page(): Broadcast buf_pool.done_free and
set buf_pool.try_LRU_scan when freeing a page. This would be
executed also as part of buf_page_write_complete().
buf_page_write_complete(): Do not broadcast buf_pool.done_flush_list,
and do not acquire buf_pool.mutex unless buf_pool.LRU eviction is needed.
Let buf_dblwr count all writes to persistent pages and broadcast a
condition variable when no outstanding writes remain.
buf_flush_page_cleaner(): Prioritize LRU flushing and eviction right after
"furious flushing" (lsn_limit). Simplify the conditions and reduce the
hold time of buf_pool.flush_list_mutex. Refuse to shut down
or sleep if buf_pool.ran_out(), that is, LRU eviction is needed.
buf_pool_t::page_cleaner_wakeup(): Add the optional parameter for_LRU.
buf_LRU_get_free_block(): Protect buf_lru_free_blocks_error_printed
with buf_pool.mutex. Invoke buf_pool.page_cleaner_wakeup(true) to
to ensure that buf_flush_page_cleaner() will process the LRU flush
request.
buf_do_LRU_batch(), buf_flush_list(), buf_flush_list_space():
Update buf_pool.stat.n_pages_written when submitting writes
(while holding buf_pool.mutex), not when completing them.
buf_page_t::flush(), buf_flush_discard_page(): Require that
the page U-latch be acquired upfront, and remove
buf_page_t::ready_for_flush().
buf_pool_t::delete_from_flush_list(): Remove the parameter "bool clear".
buf_flush_page(): Count pending page writes via buf_dblwr.
buf_flush_try_neighbors(): Take the block of page_id as a parameter.
If the tablespace is dropped before our page has been written out,
release the page U-latch.
buf_pool_invalidate(): Let the caller ensure that there are no
outstanding writes.
buf_flush_wait_batch_end(false),
buf_flush_wait_batch_end_acquiring_mutex(false):
Replaced with buf_dblwr.wait_for_page_writes().
buf_flush_wait_LRU_batch_end(): Replaces buf_flush_wait_batch_end(true).
buf_flush_list(): Remove some broadcast of buf_pool.done_flush_list.
buf_flush_buffer_pool(): Invoke also buf_dblwr.wait_for_page_writes().
buf_pool_t::io_pending(), buf_pool_t::n_flush_list(): Remove.
Outstanding writes are reflected by buf_dblwr.pending_writes().
buf_dblwr_t::init(): New function, to initialize the mutex and
the condition variables, but not the backing store.
buf_dblwr_t::is_created(): Replaces buf_dblwr_t::is_initialised().
buf_dblwr_t::pending_writes(), buf_dblwr_t::writes_pending:
Keeps track of writes of persistent data pages.
buf_flush_LRU(): Allow calls while LRU flushing may be in progress
in another thread.
Tested by Matthias Leich (correctness) and Axel Schwenke (performance)
3 years ago MDEV-26827 Make page flushing even faster
For more convenient monitoring of something that could greatly affect
the volume of page writes, we add the status variable
Innodb_buffer_pool_pages_split that was previously only available
via information_schema.innodb_metrics as "innodb_page_splits".
This was suggested by Axel Schwenke.
buf_flush_page_count: Replaced with buf_pool.stat.n_pages_written.
We protect buf_pool.stat (except n_page_gets) with buf_pool.mutex
and remove unnecessary export_vars indirection.
buf_pool.flush_list_bytes: Moved from buf_pool.stat.flush_list_bytes.
Protected by buf_pool.flush_list_mutex.
buf_pool_t::page_cleaner_status: Replaces buf_pool_t::n_flush_LRU_,
buf_pool_t::n_flush_list_, and buf_pool_t::page_cleaner_is_idle.
Protected by buf_pool.flush_list_mutex. We will exclusively broadcast
buf_pool.done_flush_list by the buf_flush_page_cleaner thread,
and only wait for it when communicating with buf_flush_page_cleaner.
There is no need to keep a count of pending writes by the
buf_pool.flush_list processing. A single flag suffices for that.
Waits for page write completion can be performed by
simply waiting on block->page.lock, or by invoking
buf_dblwr.wait_for_page_writes().
buf_LRU_block_free_non_file_page(): Broadcast buf_pool.done_free and
set buf_pool.try_LRU_scan when freeing a page. This would be
executed also as part of buf_page_write_complete().
buf_page_write_complete(): Do not broadcast buf_pool.done_flush_list,
and do not acquire buf_pool.mutex unless buf_pool.LRU eviction is needed.
Let buf_dblwr count all writes to persistent pages and broadcast a
condition variable when no outstanding writes remain.
buf_flush_page_cleaner(): Prioritize LRU flushing and eviction right after
"furious flushing" (lsn_limit). Simplify the conditions and reduce the
hold time of buf_pool.flush_list_mutex. Refuse to shut down
or sleep if buf_pool.ran_out(), that is, LRU eviction is needed.
buf_pool_t::page_cleaner_wakeup(): Add the optional parameter for_LRU.
buf_LRU_get_free_block(): Protect buf_lru_free_blocks_error_printed
with buf_pool.mutex. Invoke buf_pool.page_cleaner_wakeup(true) to
to ensure that buf_flush_page_cleaner() will process the LRU flush
request.
buf_do_LRU_batch(), buf_flush_list(), buf_flush_list_space():
Update buf_pool.stat.n_pages_written when submitting writes
(while holding buf_pool.mutex), not when completing them.
buf_page_t::flush(), buf_flush_discard_page(): Require that
the page U-latch be acquired upfront, and remove
buf_page_t::ready_for_flush().
buf_pool_t::delete_from_flush_list(): Remove the parameter "bool clear".
buf_flush_page(): Count pending page writes via buf_dblwr.
buf_flush_try_neighbors(): Take the block of page_id as a parameter.
If the tablespace is dropped before our page has been written out,
release the page U-latch.
buf_pool_invalidate(): Let the caller ensure that there are no
outstanding writes.
buf_flush_wait_batch_end(false),
buf_flush_wait_batch_end_acquiring_mutex(false):
Replaced with buf_dblwr.wait_for_page_writes().
buf_flush_wait_LRU_batch_end(): Replaces buf_flush_wait_batch_end(true).
buf_flush_list(): Remove some broadcast of buf_pool.done_flush_list.
buf_flush_buffer_pool(): Invoke also buf_dblwr.wait_for_page_writes().
buf_pool_t::io_pending(), buf_pool_t::n_flush_list(): Remove.
Outstanding writes are reflected by buf_dblwr.pending_writes().
buf_dblwr_t::init(): New function, to initialize the mutex and
the condition variables, but not the backing store.
buf_dblwr_t::is_created(): Replaces buf_dblwr_t::is_initialised().
buf_dblwr_t::pending_writes(), buf_dblwr_t::writes_pending:
Keeps track of writes of persistent data pages.
buf_flush_LRU(): Allow calls while LRU flushing may be in progress
in another thread.
Tested by Matthias Leich (correctness) and Axel Schwenke (performance)
3 years ago MDEV-13542: Crashing on corrupted page is unhelpful
The approach to handling corruption that was chosen by Oracle in
commit 177d8b0c125b841c0650d27d735e3b87509dc286
is not really useful. Not only did it actually fail to prevent InnoDB
from crashing, but it is making things worse by blocking attempts to
rescue data from or rebuild a partially readable table.
We will try to prevent crashes in a different way: by propagating
errors up the call stack. We will never mark the clustered index
persistently corrupted, so that data recovery may be attempted by
reading from the table, or by rebuilding the table.
This should also fix MDEV-13680 (crash on btr_page_alloc() failure);
it was extensively tested with innodb_file_per_table=0 and a
non-autoextend system tablespace.
We should now avoid crashes in many cases, such as when a page
cannot be read or allocated, or an inconsistency is detected when
attempting to update multiple pages. We will not crash on double-free,
such as on the recovery of DDL in system tablespace in case something
was corrupted.
Crashes on corrupted data are still possible. The fault injection mechanism
that is introduced in the subsequent commit may help catch more of them.
buf_page_import_corrupt_failure: Remove the fault injection, and instead
corrupt some pages using Perl code in the tests.
btr_cur_pessimistic_insert(): Always reserve extents (except for the
change buffer), in order to prevent a subsequent allocation failure.
btr_pcur_open_at_rnd_pos(): Merged to the only caller ibuf_merge_pages().
btr_assert_not_corrupted(), btr_corruption_report(): Remove.
Similar checks are already part of btr_block_get().
FSEG_MAGIC_N_BYTES: Replaces FSEG_MAGIC_N_VALUE.
dict_hdr_get(), trx_rsegf_get_new(), trx_undo_page_get(),
trx_undo_page_get_s_latched(): Replaced with error-checking calls.
trx_rseg_t::get(mtr_t*): Replaces trx_rsegf_get().
trx_rseg_header_create(): Let the caller update the TRX_SYS page if needed.
trx_sys_create_sys_pages(): Merged with trx_sysf_create().
dict_check_tablespaces_and_store_max_id(): Do not access
DICT_HDR_MAX_SPACE_ID, because it was already recovered in dict_boot().
Merge dict_check_sys_tables() with this function.
dir_pathname(): Replaces os_file_make_new_pathname().
row_undo_ins_remove_sec(): Do not modify the undo page by adding
a terminating NUL byte to the record.
btr_decryption_failed(): Report decryption failures
dict_set_corrupted_by_space(), dict_set_encrypted_by_space(),
dict_set_corrupted_index_cache_only(): Remove.
dict_set_corrupted(): Remove the constant parameter dict_locked=false.
Never flag the clustered index corrupted in SYS_INDEXES, because
that would deny further access to the table. It might be possible to
repair the table by executing ALTER TABLE or OPTIMIZE TABLE, in case
no B-tree leaf page is corrupted.
dict_table_skip_corrupt_index(), dict_table_next_uncorrupted_index(),
row_purge_skip_uncommitted_virtual_index(): Remove, and refactor
the callers to read dict_index_t::type only once.
dict_table_is_corrupted(): Remove.
dict_index_t::is_btree(): Determine if the index is a valid B-tree.
BUF_GET_NO_LATCH, BUF_EVICT_IF_IN_POOL: Remove.
UNIV_BTR_DEBUG: Remove. Any inconsistency will no longer trigger
assertion failures, but error codes being returned.
buf_corrupt_page_release(): Replaced with a direct call to
buf_pool.corrupted_evict().
fil_invalid_page_access_msg(): Never crash on an invalid read;
let the caller of buf_page_get_gen() decide.
btr_pcur_t::restore_position(): Propagate failure status to the caller
by returning CORRUPTED.
opt_search_plan_for_table(): Simplify the code.
row_purge_del_mark(), row_purge_upd_exist_or_extern_func(),
row_undo_ins_remove_sec_rec(), row_undo_mod_upd_del_sec(),
row_undo_mod_del_mark_sec(): Avoid mem_heap_create()/mem_heap_free()
when no secondary indexes exist.
row_undo_mod_upd_exist_sec(): Simplify the code.
row_upd_clust_step(), dict_load_table_one(): Return DB_TABLE_CORRUPT
if the clustered index (and therefore the table) is corrupted, similar
to what we do in row_insert_for_mysql().
fut_get_ptr(): Replace with buf_page_get_gen() calls.
buf_page_get_gen(): Return nullptr and *err=DB_CORRUPTION
if the page is marked as freed. For other modes than
BUF_GET_POSSIBLY_FREED or BUF_PEEK_IF_IN_POOL this will
trigger a debug assertion failure. For BUF_GET_POSSIBLY_FREED,
we will return nullptr for freed pages, so that the callers
can be simplified. The purge of transaction history will be
a new user of BUF_GET_POSSIBLY_FREED, to avoid crashes on
corrupted data.
buf_page_get_low(): Never crash on a corrupted page, but simply
return nullptr.
fseg_page_is_allocated(): Replaces fseg_page_is_free().
fts_drop_common_tables(): Return an error if the transaction
was rolled back.
fil_space_t::set_corrupted(): Report a tablespace as corrupted if
it was not reported already.
fil_space_t::io(): Invoke fil_space_t::set_corrupted() to report
out-of-bounds page access or other errors.
Clean up mtr_t::page_lock()
buf_page_get_low(): Validate the page identifier (to check for
recently read corrupted pages) after acquiring the page latch.
buf_page_t::read_complete(): Flag uninitialized (all-zero) pages
with DB_FAIL. Return DB_PAGE_CORRUPTED on page number mismatch.
mtr_t::defer_drop_ahi(): Renamed from mtr_defer_drop_ahi().
recv_sys_t::free_corrupted_page(): Only set_corrupt_fs()
if any log records exist for the page. We do not mind if read-ahead
produces corrupted (or all-zero) pages that were not actually needed
during recovery.
recv_recover_page(): Return whether the operation succeeded.
recv_sys_t::recover_low(): Simplify the logic. Check for recovery error.
Thanks to Matthias Leich for testing this extensively and to the
authors of https://rr-project.org for making it easy to diagnose
and fix any failures that were found during the testing.
3 years ago MDEV-26827 Make page flushing even faster
For more convenient monitoring of something that could greatly affect
the volume of page writes, we add the status variable
Innodb_buffer_pool_pages_split that was previously only available
via information_schema.innodb_metrics as "innodb_page_splits".
This was suggested by Axel Schwenke.
buf_flush_page_count: Replaced with buf_pool.stat.n_pages_written.
We protect buf_pool.stat (except n_page_gets) with buf_pool.mutex
and remove unnecessary export_vars indirection.
buf_pool.flush_list_bytes: Moved from buf_pool.stat.flush_list_bytes.
Protected by buf_pool.flush_list_mutex.
buf_pool_t::page_cleaner_status: Replaces buf_pool_t::n_flush_LRU_,
buf_pool_t::n_flush_list_, and buf_pool_t::page_cleaner_is_idle.
Protected by buf_pool.flush_list_mutex. We will exclusively broadcast
buf_pool.done_flush_list by the buf_flush_page_cleaner thread,
and only wait for it when communicating with buf_flush_page_cleaner.
There is no need to keep a count of pending writes by the
buf_pool.flush_list processing. A single flag suffices for that.
Waits for page write completion can be performed by
simply waiting on block->page.lock, or by invoking
buf_dblwr.wait_for_page_writes().
buf_LRU_block_free_non_file_page(): Broadcast buf_pool.done_free and
set buf_pool.try_LRU_scan when freeing a page. This would be
executed also as part of buf_page_write_complete().
buf_page_write_complete(): Do not broadcast buf_pool.done_flush_list,
and do not acquire buf_pool.mutex unless buf_pool.LRU eviction is needed.
Let buf_dblwr count all writes to persistent pages and broadcast a
condition variable when no outstanding writes remain.
buf_flush_page_cleaner(): Prioritize LRU flushing and eviction right after
"furious flushing" (lsn_limit). Simplify the conditions and reduce the
hold time of buf_pool.flush_list_mutex. Refuse to shut down
or sleep if buf_pool.ran_out(), that is, LRU eviction is needed.
buf_pool_t::page_cleaner_wakeup(): Add the optional parameter for_LRU.
buf_LRU_get_free_block(): Protect buf_lru_free_blocks_error_printed
with buf_pool.mutex. Invoke buf_pool.page_cleaner_wakeup(true) to
to ensure that buf_flush_page_cleaner() will process the LRU flush
request.
buf_do_LRU_batch(), buf_flush_list(), buf_flush_list_space():
Update buf_pool.stat.n_pages_written when submitting writes
(while holding buf_pool.mutex), not when completing them.
buf_page_t::flush(), buf_flush_discard_page(): Require that
the page U-latch be acquired upfront, and remove
buf_page_t::ready_for_flush().
buf_pool_t::delete_from_flush_list(): Remove the parameter "bool clear".
buf_flush_page(): Count pending page writes via buf_dblwr.
buf_flush_try_neighbors(): Take the block of page_id as a parameter.
If the tablespace is dropped before our page has been written out,
release the page U-latch.
buf_pool_invalidate(): Let the caller ensure that there are no
outstanding writes.
buf_flush_wait_batch_end(false),
buf_flush_wait_batch_end_acquiring_mutex(false):
Replaced with buf_dblwr.wait_for_page_writes().
buf_flush_wait_LRU_batch_end(): Replaces buf_flush_wait_batch_end(true).
buf_flush_list(): Remove some broadcast of buf_pool.done_flush_list.
buf_flush_buffer_pool(): Invoke also buf_dblwr.wait_for_page_writes().
buf_pool_t::io_pending(), buf_pool_t::n_flush_list(): Remove.
Outstanding writes are reflected by buf_dblwr.pending_writes().
buf_dblwr_t::init(): New function, to initialize the mutex and
the condition variables, but not the backing store.
buf_dblwr_t::is_created(): Replaces buf_dblwr_t::is_initialised().
buf_dblwr_t::pending_writes(), buf_dblwr_t::writes_pending:
Keeps track of writes of persistent data pages.
buf_flush_LRU(): Allow calls while LRU flushing may be in progress
in another thread.
Tested by Matthias Leich (correctness) and Axel Schwenke (performance)
3 years ago MDEV-13542: Crashing on corrupted page is unhelpful
The approach to handling corruption that was chosen by Oracle in
commit 177d8b0c125b841c0650d27d735e3b87509dc286
is not really useful. Not only did it actually fail to prevent InnoDB
from crashing, but it is making things worse by blocking attempts to
rescue data from or rebuild a partially readable table.
We will try to prevent crashes in a different way: by propagating
errors up the call stack. We will never mark the clustered index
persistently corrupted, so that data recovery may be attempted by
reading from the table, or by rebuilding the table.
This should also fix MDEV-13680 (crash on btr_page_alloc() failure);
it was extensively tested with innodb_file_per_table=0 and a
non-autoextend system tablespace.
We should now avoid crashes in many cases, such as when a page
cannot be read or allocated, or an inconsistency is detected when
attempting to update multiple pages. We will not crash on double-free,
such as on the recovery of DDL in system tablespace in case something
was corrupted.
Crashes on corrupted data are still possible. The fault injection mechanism
that is introduced in the subsequent commit may help catch more of them.
buf_page_import_corrupt_failure: Remove the fault injection, and instead
corrupt some pages using Perl code in the tests.
btr_cur_pessimistic_insert(): Always reserve extents (except for the
change buffer), in order to prevent a subsequent allocation failure.
btr_pcur_open_at_rnd_pos(): Merged to the only caller ibuf_merge_pages().
btr_assert_not_corrupted(), btr_corruption_report(): Remove.
Similar checks are already part of btr_block_get().
FSEG_MAGIC_N_BYTES: Replaces FSEG_MAGIC_N_VALUE.
dict_hdr_get(), trx_rsegf_get_new(), trx_undo_page_get(),
trx_undo_page_get_s_latched(): Replaced with error-checking calls.
trx_rseg_t::get(mtr_t*): Replaces trx_rsegf_get().
trx_rseg_header_create(): Let the caller update the TRX_SYS page if needed.
trx_sys_create_sys_pages(): Merged with trx_sysf_create().
dict_check_tablespaces_and_store_max_id(): Do not access
DICT_HDR_MAX_SPACE_ID, because it was already recovered in dict_boot().
Merge dict_check_sys_tables() with this function.
dir_pathname(): Replaces os_file_make_new_pathname().
row_undo_ins_remove_sec(): Do not modify the undo page by adding
a terminating NUL byte to the record.
btr_decryption_failed(): Report decryption failures
dict_set_corrupted_by_space(), dict_set_encrypted_by_space(),
dict_set_corrupted_index_cache_only(): Remove.
dict_set_corrupted(): Remove the constant parameter dict_locked=false.
Never flag the clustered index corrupted in SYS_INDEXES, because
that would deny further access to the table. It might be possible to
repair the table by executing ALTER TABLE or OPTIMIZE TABLE, in case
no B-tree leaf page is corrupted.
dict_table_skip_corrupt_index(), dict_table_next_uncorrupted_index(),
row_purge_skip_uncommitted_virtual_index(): Remove, and refactor
the callers to read dict_index_t::type only once.
dict_table_is_corrupted(): Remove.
dict_index_t::is_btree(): Determine if the index is a valid B-tree.
BUF_GET_NO_LATCH, BUF_EVICT_IF_IN_POOL: Remove.
UNIV_BTR_DEBUG: Remove. Any inconsistency will no longer trigger
assertion failures, but error codes being returned.
buf_corrupt_page_release(): Replaced with a direct call to
buf_pool.corrupted_evict().
fil_invalid_page_access_msg(): Never crash on an invalid read;
let the caller of buf_page_get_gen() decide.
btr_pcur_t::restore_position(): Propagate failure status to the caller
by returning CORRUPTED.
opt_search_plan_for_table(): Simplify the code.
row_purge_del_mark(), row_purge_upd_exist_or_extern_func(),
row_undo_ins_remove_sec_rec(), row_undo_mod_upd_del_sec(),
row_undo_mod_del_mark_sec(): Avoid mem_heap_create()/mem_heap_free()
when no secondary indexes exist.
row_undo_mod_upd_exist_sec(): Simplify the code.
row_upd_clust_step(), dict_load_table_one(): Return DB_TABLE_CORRUPT
if the clustered index (and therefore the table) is corrupted, similar
to what we do in row_insert_for_mysql().
fut_get_ptr(): Replace with buf_page_get_gen() calls.
buf_page_get_gen(): Return nullptr and *err=DB_CORRUPTION
if the page is marked as freed. For other modes than
BUF_GET_POSSIBLY_FREED or BUF_PEEK_IF_IN_POOL this will
trigger a debug assertion failure. For BUF_GET_POSSIBLY_FREED,
we will return nullptr for freed pages, so that the callers
can be simplified. The purge of transaction history will be
a new user of BUF_GET_POSSIBLY_FREED, to avoid crashes on
corrupted data.
buf_page_get_low(): Never crash on a corrupted page, but simply
return nullptr.
fseg_page_is_allocated(): Replaces fseg_page_is_free().
fts_drop_common_tables(): Return an error if the transaction
was rolled back.
fil_space_t::set_corrupted(): Report a tablespace as corrupted if
it was not reported already.
fil_space_t::io(): Invoke fil_space_t::set_corrupted() to report
out-of-bounds page access or other errors.
Clean up mtr_t::page_lock()
buf_page_get_low(): Validate the page identifier (to check for
recently read corrupted pages) after acquiring the page latch.
buf_page_t::read_complete(): Flag uninitialized (all-zero) pages
with DB_FAIL. Return DB_PAGE_CORRUPTED on page number mismatch.
mtr_t::defer_drop_ahi(): Renamed from mtr_defer_drop_ahi().
recv_sys_t::free_corrupted_page(): Only set_corrupt_fs()
if any log records exist for the page. We do not mind if read-ahead
produces corrupted (or all-zero) pages that were not actually needed
during recovery.
recv_recover_page(): Return whether the operation succeeded.
recv_sys_t::recover_low(): Simplify the logic. Check for recovery error.
Thanks to Matthias Leich for testing this extensively and to the
authors of https://rr-project.org for making it easy to diagnose
and fix any failures that were found during the testing.
3 years ago MDEV-13542: Crashing on corrupted page is unhelpful
The approach to handling corruption that was chosen by Oracle in
commit 177d8b0c125b841c0650d27d735e3b87509dc286
is not really useful. Not only did it actually fail to prevent InnoDB
from crashing, but it is making things worse by blocking attempts to
rescue data from or rebuild a partially readable table.
We will try to prevent crashes in a different way: by propagating
errors up the call stack. We will never mark the clustered index
persistently corrupted, so that data recovery may be attempted by
reading from the table, or by rebuilding the table.
This should also fix MDEV-13680 (crash on btr_page_alloc() failure);
it was extensively tested with innodb_file_per_table=0 and a
non-autoextend system tablespace.
We should now avoid crashes in many cases, such as when a page
cannot be read or allocated, or an inconsistency is detected when
attempting to update multiple pages. We will not crash on double-free,
such as on the recovery of DDL in system tablespace in case something
was corrupted.
Crashes on corrupted data are still possible. The fault injection mechanism
that is introduced in the subsequent commit may help catch more of them.
buf_page_import_corrupt_failure: Remove the fault injection, and instead
corrupt some pages using Perl code in the tests.
btr_cur_pessimistic_insert(): Always reserve extents (except for the
change buffer), in order to prevent a subsequent allocation failure.
btr_pcur_open_at_rnd_pos(): Merged to the only caller ibuf_merge_pages().
btr_assert_not_corrupted(), btr_corruption_report(): Remove.
Similar checks are already part of btr_block_get().
FSEG_MAGIC_N_BYTES: Replaces FSEG_MAGIC_N_VALUE.
dict_hdr_get(), trx_rsegf_get_new(), trx_undo_page_get(),
trx_undo_page_get_s_latched(): Replaced with error-checking calls.
trx_rseg_t::get(mtr_t*): Replaces trx_rsegf_get().
trx_rseg_header_create(): Let the caller update the TRX_SYS page if needed.
trx_sys_create_sys_pages(): Merged with trx_sysf_create().
dict_check_tablespaces_and_store_max_id(): Do not access
DICT_HDR_MAX_SPACE_ID, because it was already recovered in dict_boot().
Merge dict_check_sys_tables() with this function.
dir_pathname(): Replaces os_file_make_new_pathname().
row_undo_ins_remove_sec(): Do not modify the undo page by adding
a terminating NUL byte to the record.
btr_decryption_failed(): Report decryption failures
dict_set_corrupted_by_space(), dict_set_encrypted_by_space(),
dict_set_corrupted_index_cache_only(): Remove.
dict_set_corrupted(): Remove the constant parameter dict_locked=false.
Never flag the clustered index corrupted in SYS_INDEXES, because
that would deny further access to the table. It might be possible to
repair the table by executing ALTER TABLE or OPTIMIZE TABLE, in case
no B-tree leaf page is corrupted.
dict_table_skip_corrupt_index(), dict_table_next_uncorrupted_index(),
row_purge_skip_uncommitted_virtual_index(): Remove, and refactor
the callers to read dict_index_t::type only once.
dict_table_is_corrupted(): Remove.
dict_index_t::is_btree(): Determine if the index is a valid B-tree.
BUF_GET_NO_LATCH, BUF_EVICT_IF_IN_POOL: Remove.
UNIV_BTR_DEBUG: Remove. Any inconsistency will no longer trigger
assertion failures, but error codes being returned.
buf_corrupt_page_release(): Replaced with a direct call to
buf_pool.corrupted_evict().
fil_invalid_page_access_msg(): Never crash on an invalid read;
let the caller of buf_page_get_gen() decide.
btr_pcur_t::restore_position(): Propagate failure status to the caller
by returning CORRUPTED.
opt_search_plan_for_table(): Simplify the code.
row_purge_del_mark(), row_purge_upd_exist_or_extern_func(),
row_undo_ins_remove_sec_rec(), row_undo_mod_upd_del_sec(),
row_undo_mod_del_mark_sec(): Avoid mem_heap_create()/mem_heap_free()
when no secondary indexes exist.
row_undo_mod_upd_exist_sec(): Simplify the code.
row_upd_clust_step(), dict_load_table_one(): Return DB_TABLE_CORRUPT
if the clustered index (and therefore the table) is corrupted, similar
to what we do in row_insert_for_mysql().
fut_get_ptr(): Replace with buf_page_get_gen() calls.
buf_page_get_gen(): Return nullptr and *err=DB_CORRUPTION
if the page is marked as freed. For other modes than
BUF_GET_POSSIBLY_FREED or BUF_PEEK_IF_IN_POOL this will
trigger a debug assertion failure. For BUF_GET_POSSIBLY_FREED,
we will return nullptr for freed pages, so that the callers
can be simplified. The purge of transaction history will be
a new user of BUF_GET_POSSIBLY_FREED, to avoid crashes on
corrupted data.
buf_page_get_low(): Never crash on a corrupted page, but simply
return nullptr.
fseg_page_is_allocated(): Replaces fseg_page_is_free().
fts_drop_common_tables(): Return an error if the transaction
was rolled back.
fil_space_t::set_corrupted(): Report a tablespace as corrupted if
it was not reported already.
fil_space_t::io(): Invoke fil_space_t::set_corrupted() to report
out-of-bounds page access or other errors.
Clean up mtr_t::page_lock()
buf_page_get_low(): Validate the page identifier (to check for
recently read corrupted pages) after acquiring the page latch.
buf_page_t::read_complete(): Flag uninitialized (all-zero) pages
with DB_FAIL. Return DB_PAGE_CORRUPTED on page number mismatch.
mtr_t::defer_drop_ahi(): Renamed from mtr_defer_drop_ahi().
recv_sys_t::free_corrupted_page(): Only set_corrupt_fs()
if any log records exist for the page. We do not mind if read-ahead
produces corrupted (or all-zero) pages that were not actually needed
during recovery.
recv_recover_page(): Return whether the operation succeeded.
recv_sys_t::recover_low(): Simplify the logic. Check for recovery error.
Thanks to Matthias Leich for testing this extensively and to the
authors of https://rr-project.org for making it easy to diagnose
and fix any failures that were found during the testing.
3 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-13542: Crashing on corrupted page is unhelpful
The approach to handling corruption that was chosen by Oracle in
commit 177d8b0c125b841c0650d27d735e3b87509dc286
is not really useful. Not only did it actually fail to prevent InnoDB
from crashing, but it is making things worse by blocking attempts to
rescue data from or rebuild a partially readable table.
We will try to prevent crashes in a different way: by propagating
errors up the call stack. We will never mark the clustered index
persistently corrupted, so that data recovery may be attempted by
reading from the table, or by rebuilding the table.
This should also fix MDEV-13680 (crash on btr_page_alloc() failure);
it was extensively tested with innodb_file_per_table=0 and a
non-autoextend system tablespace.
We should now avoid crashes in many cases, such as when a page
cannot be read or allocated, or an inconsistency is detected when
attempting to update multiple pages. We will not crash on double-free,
such as on the recovery of DDL in system tablespace in case something
was corrupted.
Crashes on corrupted data are still possible. The fault injection mechanism
that is introduced in the subsequent commit may help catch more of them.
buf_page_import_corrupt_failure: Remove the fault injection, and instead
corrupt some pages using Perl code in the tests.
btr_cur_pessimistic_insert(): Always reserve extents (except for the
change buffer), in order to prevent a subsequent allocation failure.
btr_pcur_open_at_rnd_pos(): Merged to the only caller ibuf_merge_pages().
btr_assert_not_corrupted(), btr_corruption_report(): Remove.
Similar checks are already part of btr_block_get().
FSEG_MAGIC_N_BYTES: Replaces FSEG_MAGIC_N_VALUE.
dict_hdr_get(), trx_rsegf_get_new(), trx_undo_page_get(),
trx_undo_page_get_s_latched(): Replaced with error-checking calls.
trx_rseg_t::get(mtr_t*): Replaces trx_rsegf_get().
trx_rseg_header_create(): Let the caller update the TRX_SYS page if needed.
trx_sys_create_sys_pages(): Merged with trx_sysf_create().
dict_check_tablespaces_and_store_max_id(): Do not access
DICT_HDR_MAX_SPACE_ID, because it was already recovered in dict_boot().
Merge dict_check_sys_tables() with this function.
dir_pathname(): Replaces os_file_make_new_pathname().
row_undo_ins_remove_sec(): Do not modify the undo page by adding
a terminating NUL byte to the record.
btr_decryption_failed(): Report decryption failures
dict_set_corrupted_by_space(), dict_set_encrypted_by_space(),
dict_set_corrupted_index_cache_only(): Remove.
dict_set_corrupted(): Remove the constant parameter dict_locked=false.
Never flag the clustered index corrupted in SYS_INDEXES, because
that would deny further access to the table. It might be possible to
repair the table by executing ALTER TABLE or OPTIMIZE TABLE, in case
no B-tree leaf page is corrupted.
dict_table_skip_corrupt_index(), dict_table_next_uncorrupted_index(),
row_purge_skip_uncommitted_virtual_index(): Remove, and refactor
the callers to read dict_index_t::type only once.
dict_table_is_corrupted(): Remove.
dict_index_t::is_btree(): Determine if the index is a valid B-tree.
BUF_GET_NO_LATCH, BUF_EVICT_IF_IN_POOL: Remove.
UNIV_BTR_DEBUG: Remove. Any inconsistency will no longer trigger
assertion failures, but error codes being returned.
buf_corrupt_page_release(): Replaced with a direct call to
buf_pool.corrupted_evict().
fil_invalid_page_access_msg(): Never crash on an invalid read;
let the caller of buf_page_get_gen() decide.
btr_pcur_t::restore_position(): Propagate failure status to the caller
by returning CORRUPTED.
opt_search_plan_for_table(): Simplify the code.
row_purge_del_mark(), row_purge_upd_exist_or_extern_func(),
row_undo_ins_remove_sec_rec(), row_undo_mod_upd_del_sec(),
row_undo_mod_del_mark_sec(): Avoid mem_heap_create()/mem_heap_free()
when no secondary indexes exist.
row_undo_mod_upd_exist_sec(): Simplify the code.
row_upd_clust_step(), dict_load_table_one(): Return DB_TABLE_CORRUPT
if the clustered index (and therefore the table) is corrupted, similar
to what we do in row_insert_for_mysql().
fut_get_ptr(): Replace with buf_page_get_gen() calls.
buf_page_get_gen(): Return nullptr and *err=DB_CORRUPTION
if the page is marked as freed. For other modes than
BUF_GET_POSSIBLY_FREED or BUF_PEEK_IF_IN_POOL this will
trigger a debug assertion failure. For BUF_GET_POSSIBLY_FREED,
we will return nullptr for freed pages, so that the callers
can be simplified. The purge of transaction history will be
a new user of BUF_GET_POSSIBLY_FREED, to avoid crashes on
corrupted data.
buf_page_get_low(): Never crash on a corrupted page, but simply
return nullptr.
fseg_page_is_allocated(): Replaces fseg_page_is_free().
fts_drop_common_tables(): Return an error if the transaction
was rolled back.
fil_space_t::set_corrupted(): Report a tablespace as corrupted if
it was not reported already.
fil_space_t::io(): Invoke fil_space_t::set_corrupted() to report
out-of-bounds page access or other errors.
Clean up mtr_t::page_lock()
buf_page_get_low(): Validate the page identifier (to check for
recently read corrupted pages) after acquiring the page latch.
buf_page_t::read_complete(): Flag uninitialized (all-zero) pages
with DB_FAIL. Return DB_PAGE_CORRUPTED on page number mismatch.
mtr_t::defer_drop_ahi(): Renamed from mtr_defer_drop_ahi().
recv_sys_t::free_corrupted_page(): Only set_corrupt_fs()
if any log records exist for the page. We do not mind if read-ahead
produces corrupted (or all-zero) pages that were not actually needed
during recovery.
recv_recover_page(): Return whether the operation succeeded.
recv_sys_t::recover_low(): Simplify the logic. Check for recovery error.
Thanks to Matthias Leich for testing this extensively and to the
authors of https://rr-project.org for making it easy to diagnose
and fix any failures that were found during the testing.
3 years ago MDEV-13542: Crashing on corrupted page is unhelpful
The approach to handling corruption that was chosen by Oracle in
commit 177d8b0c125b841c0650d27d735e3b87509dc286
is not really useful. Not only did it actually fail to prevent InnoDB
from crashing, but it is making things worse by blocking attempts to
rescue data from or rebuild a partially readable table.
We will try to prevent crashes in a different way: by propagating
errors up the call stack. We will never mark the clustered index
persistently corrupted, so that data recovery may be attempted by
reading from the table, or by rebuilding the table.
This should also fix MDEV-13680 (crash on btr_page_alloc() failure);
it was extensively tested with innodb_file_per_table=0 and a
non-autoextend system tablespace.
We should now avoid crashes in many cases, such as when a page
cannot be read or allocated, or an inconsistency is detected when
attempting to update multiple pages. We will not crash on double-free,
such as on the recovery of DDL in system tablespace in case something
was corrupted.
Crashes on corrupted data are still possible. The fault injection mechanism
that is introduced in the subsequent commit may help catch more of them.
buf_page_import_corrupt_failure: Remove the fault injection, and instead
corrupt some pages using Perl code in the tests.
btr_cur_pessimistic_insert(): Always reserve extents (except for the
change buffer), in order to prevent a subsequent allocation failure.
btr_pcur_open_at_rnd_pos(): Merged to the only caller ibuf_merge_pages().
btr_assert_not_corrupted(), btr_corruption_report(): Remove.
Similar checks are already part of btr_block_get().
FSEG_MAGIC_N_BYTES: Replaces FSEG_MAGIC_N_VALUE.
dict_hdr_get(), trx_rsegf_get_new(), trx_undo_page_get(),
trx_undo_page_get_s_latched(): Replaced with error-checking calls.
trx_rseg_t::get(mtr_t*): Replaces trx_rsegf_get().
trx_rseg_header_create(): Let the caller update the TRX_SYS page if needed.
trx_sys_create_sys_pages(): Merged with trx_sysf_create().
dict_check_tablespaces_and_store_max_id(): Do not access
DICT_HDR_MAX_SPACE_ID, because it was already recovered in dict_boot().
Merge dict_check_sys_tables() with this function.
dir_pathname(): Replaces os_file_make_new_pathname().
row_undo_ins_remove_sec(): Do not modify the undo page by adding
a terminating NUL byte to the record.
btr_decryption_failed(): Report decryption failures
dict_set_corrupted_by_space(), dict_set_encrypted_by_space(),
dict_set_corrupted_index_cache_only(): Remove.
dict_set_corrupted(): Remove the constant parameter dict_locked=false.
Never flag the clustered index corrupted in SYS_INDEXES, because
that would deny further access to the table. It might be possible to
repair the table by executing ALTER TABLE or OPTIMIZE TABLE, in case
no B-tree leaf page is corrupted.
dict_table_skip_corrupt_index(), dict_table_next_uncorrupted_index(),
row_purge_skip_uncommitted_virtual_index(): Remove, and refactor
the callers to read dict_index_t::type only once.
dict_table_is_corrupted(): Remove.
dict_index_t::is_btree(): Determine if the index is a valid B-tree.
BUF_GET_NO_LATCH, BUF_EVICT_IF_IN_POOL: Remove.
UNIV_BTR_DEBUG: Remove. Any inconsistency will no longer trigger
assertion failures, but error codes being returned.
buf_corrupt_page_release(): Replaced with a direct call to
buf_pool.corrupted_evict().
fil_invalid_page_access_msg(): Never crash on an invalid read;
let the caller of buf_page_get_gen() decide.
btr_pcur_t::restore_position(): Propagate failure status to the caller
by returning CORRUPTED.
opt_search_plan_for_table(): Simplify the code.
row_purge_del_mark(), row_purge_upd_exist_or_extern_func(),
row_undo_ins_remove_sec_rec(), row_undo_mod_upd_del_sec(),
row_undo_mod_del_mark_sec(): Avoid mem_heap_create()/mem_heap_free()
when no secondary indexes exist.
row_undo_mod_upd_exist_sec(): Simplify the code.
row_upd_clust_step(), dict_load_table_one(): Return DB_TABLE_CORRUPT
if the clustered index (and therefore the table) is corrupted, similar
to what we do in row_insert_for_mysql().
fut_get_ptr(): Replace with buf_page_get_gen() calls.
buf_page_get_gen(): Return nullptr and *err=DB_CORRUPTION
if the page is marked as freed. For other modes than
BUF_GET_POSSIBLY_FREED or BUF_PEEK_IF_IN_POOL this will
trigger a debug assertion failure. For BUF_GET_POSSIBLY_FREED,
we will return nullptr for freed pages, so that the callers
can be simplified. The purge of transaction history will be
a new user of BUF_GET_POSSIBLY_FREED, to avoid crashes on
corrupted data.
buf_page_get_low(): Never crash on a corrupted page, but simply
return nullptr.
fseg_page_is_allocated(): Replaces fseg_page_is_free().
fts_drop_common_tables(): Return an error if the transaction
was rolled back.
fil_space_t::set_corrupted(): Report a tablespace as corrupted if
it was not reported already.
fil_space_t::io(): Invoke fil_space_t::set_corrupted() to report
out-of-bounds page access or other errors.
Clean up mtr_t::page_lock()
buf_page_get_low(): Validate the page identifier (to check for
recently read corrupted pages) after acquiring the page latch.
buf_page_t::read_complete(): Flag uninitialized (all-zero) pages
with DB_FAIL. Return DB_PAGE_CORRUPTED on page number mismatch.
mtr_t::defer_drop_ahi(): Renamed from mtr_defer_drop_ahi().
recv_sys_t::free_corrupted_page(): Only set_corrupt_fs()
if any log records exist for the page. We do not mind if read-ahead
produces corrupted (or all-zero) pages that were not actually needed
during recovery.
recv_recover_page(): Return whether the operation succeeded.
recv_sys_t::recover_low(): Simplify the logic. Check for recovery error.
Thanks to Matthias Leich for testing this extensively and to the
authors of https://rr-project.org for making it easy to diagnose
and fix any failures that were found during the testing.
3 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-13542: Crashing on corrupted page is unhelpful
The approach to handling corruption that was chosen by Oracle in
commit 177d8b0c125b841c0650d27d735e3b87509dc286
is not really useful. Not only did it actually fail to prevent InnoDB
from crashing, but it is making things worse by blocking attempts to
rescue data from or rebuild a partially readable table.
We will try to prevent crashes in a different way: by propagating
errors up the call stack. We will never mark the clustered index
persistently corrupted, so that data recovery may be attempted by
reading from the table, or by rebuilding the table.
This should also fix MDEV-13680 (crash on btr_page_alloc() failure);
it was extensively tested with innodb_file_per_table=0 and a
non-autoextend system tablespace.
We should now avoid crashes in many cases, such as when a page
cannot be read or allocated, or an inconsistency is detected when
attempting to update multiple pages. We will not crash on double-free,
such as on the recovery of DDL in system tablespace in case something
was corrupted.
Crashes on corrupted data are still possible. The fault injection mechanism
that is introduced in the subsequent commit may help catch more of them.
buf_page_import_corrupt_failure: Remove the fault injection, and instead
corrupt some pages using Perl code in the tests.
btr_cur_pessimistic_insert(): Always reserve extents (except for the
change buffer), in order to prevent a subsequent allocation failure.
btr_pcur_open_at_rnd_pos(): Merged to the only caller ibuf_merge_pages().
btr_assert_not_corrupted(), btr_corruption_report(): Remove.
Similar checks are already part of btr_block_get().
FSEG_MAGIC_N_BYTES: Replaces FSEG_MAGIC_N_VALUE.
dict_hdr_get(), trx_rsegf_get_new(), trx_undo_page_get(),
trx_undo_page_get_s_latched(): Replaced with error-checking calls.
trx_rseg_t::get(mtr_t*): Replaces trx_rsegf_get().
trx_rseg_header_create(): Let the caller update the TRX_SYS page if needed.
trx_sys_create_sys_pages(): Merged with trx_sysf_create().
dict_check_tablespaces_and_store_max_id(): Do not access
DICT_HDR_MAX_SPACE_ID, because it was already recovered in dict_boot().
Merge dict_check_sys_tables() with this function.
dir_pathname(): Replaces os_file_make_new_pathname().
row_undo_ins_remove_sec(): Do not modify the undo page by adding
a terminating NUL byte to the record.
btr_decryption_failed(): Report decryption failures
dict_set_corrupted_by_space(), dict_set_encrypted_by_space(),
dict_set_corrupted_index_cache_only(): Remove.
dict_set_corrupted(): Remove the constant parameter dict_locked=false.
Never flag the clustered index corrupted in SYS_INDEXES, because
that would deny further access to the table. It might be possible to
repair the table by executing ALTER TABLE or OPTIMIZE TABLE, in case
no B-tree leaf page is corrupted.
dict_table_skip_corrupt_index(), dict_table_next_uncorrupted_index(),
row_purge_skip_uncommitted_virtual_index(): Remove, and refactor
the callers to read dict_index_t::type only once.
dict_table_is_corrupted(): Remove.
dict_index_t::is_btree(): Determine if the index is a valid B-tree.
BUF_GET_NO_LATCH, BUF_EVICT_IF_IN_POOL: Remove.
UNIV_BTR_DEBUG: Remove. Any inconsistency will no longer trigger
assertion failures, but error codes being returned.
buf_corrupt_page_release(): Replaced with a direct call to
buf_pool.corrupted_evict().
fil_invalid_page_access_msg(): Never crash on an invalid read;
let the caller of buf_page_get_gen() decide.
btr_pcur_t::restore_position(): Propagate failure status to the caller
by returning CORRUPTED.
opt_search_plan_for_table(): Simplify the code.
row_purge_del_mark(), row_purge_upd_exist_or_extern_func(),
row_undo_ins_remove_sec_rec(), row_undo_mod_upd_del_sec(),
row_undo_mod_del_mark_sec(): Avoid mem_heap_create()/mem_heap_free()
when no secondary indexes exist.
row_undo_mod_upd_exist_sec(): Simplify the code.
row_upd_clust_step(), dict_load_table_one(): Return DB_TABLE_CORRUPT
if the clustered index (and therefore the table) is corrupted, similar
to what we do in row_insert_for_mysql().
fut_get_ptr(): Replace with buf_page_get_gen() calls.
buf_page_get_gen(): Return nullptr and *err=DB_CORRUPTION
if the page is marked as freed. For other modes than
BUF_GET_POSSIBLY_FREED or BUF_PEEK_IF_IN_POOL this will
trigger a debug assertion failure. For BUF_GET_POSSIBLY_FREED,
we will return nullptr for freed pages, so that the callers
can be simplified. The purge of transaction history will be
a new user of BUF_GET_POSSIBLY_FREED, to avoid crashes on
corrupted data.
buf_page_get_low(): Never crash on a corrupted page, but simply
return nullptr.
fseg_page_is_allocated(): Replaces fseg_page_is_free().
fts_drop_common_tables(): Return an error if the transaction
was rolled back.
fil_space_t::set_corrupted(): Report a tablespace as corrupted if
it was not reported already.
fil_space_t::io(): Invoke fil_space_t::set_corrupted() to report
out-of-bounds page access or other errors.
Clean up mtr_t::page_lock()
buf_page_get_low(): Validate the page identifier (to check for
recently read corrupted pages) after acquiring the page latch.
buf_page_t::read_complete(): Flag uninitialized (all-zero) pages
with DB_FAIL. Return DB_PAGE_CORRUPTED on page number mismatch.
mtr_t::defer_drop_ahi(): Renamed from mtr_defer_drop_ahi().
recv_sys_t::free_corrupted_page(): Only set_corrupt_fs()
if any log records exist for the page. We do not mind if read-ahead
produces corrupted (or all-zero) pages that were not actually needed
during recovery.
recv_recover_page(): Return whether the operation succeeded.
recv_sys_t::recover_low(): Simplify the logic. Check for recovery error.
Thanks to Matthias Leich for testing this extensively and to the
authors of https://rr-project.org for making it easy to diagnose
and fix any failures that were found during the testing.
3 years ago MDEV-23399: Performance regression with write workloads
The buffer pool refactoring in MDEV-15053 and MDEV-22871 shifted
the performance bottleneck to the page flushing.
The configuration parameters will be changed as follows:
innodb_lru_flush_size=32 (new: how many pages to flush on LRU eviction)
innodb_lru_scan_depth=1536 (old: 1024)
innodb_max_dirty_pages_pct=90 (old: 75)
innodb_max_dirty_pages_pct_lwm=75 (old: 0)
Note: The parameter innodb_lru_scan_depth will only affect LRU
eviction of buffer pool pages when a new page is being allocated. The
page cleaner thread will no longer evict any pages. It used to
guarantee that some pages will remain free in the buffer pool. Now, we
perform that eviction 'on demand' in buf_LRU_get_free_block().
The parameter innodb_lru_scan_depth(srv_LRU_scan_depth) is used as follows:
* When the buffer pool is being shrunk in buf_pool_t::withdraw_blocks()
* As a buf_pool.free limit in buf_LRU_list_batch() for terminating
the flushing that is initiated e.g., by buf_LRU_get_free_block()
The parameter also used to serve as an initial limit for unzip_LRU
eviction (evicting uncompressed page frames while retaining
ROW_FORMAT=COMPRESSED pages), but now we will use a hard-coded limit
of 100 or unlimited for invoking buf_LRU_scan_and_free_block().
The status variables will be changed as follows:
innodb_buffer_pool_pages_flushed: This includes also the count of
innodb_buffer_pool_pages_LRU_flushed and should work reliably,
updated one by one in buf_flush_page() to give more real-time
statistics. The function buf_flush_stats(), which we are removing,
was not called in every code path. For both counters, we will use
regular variables that are incremented in a critical section of
buf_pool.mutex. Note that show_innodb_vars() directly links to the
variables, and reads of the counters will *not* be protected by
buf_pool.mutex, so you cannot get a consistent snapshot of both variables.
The following INFORMATION_SCHEMA.INNODB_METRICS counters will be
removed, because the page cleaner no longer deals with writing or
evicting least recently used pages, and because the single-page writes
have been removed:
* buffer_LRU_batch_flush_avg_time_slot
* buffer_LRU_batch_flush_avg_time_thread
* buffer_LRU_batch_flush_avg_time_est
* buffer_LRU_batch_flush_avg_pass
* buffer_LRU_single_flush_scanned
* buffer_LRU_single_flush_num_scan
* buffer_LRU_single_flush_scanned_per_call
When moving to a single buffer pool instance in MDEV-15058, we missed
some opportunity to simplify the buf_flush_page_cleaner thread. It was
unnecessarily using a mutex and some complex data structures, even
though we always have a single page cleaner thread.
Furthermore, the buf_flush_page_cleaner thread had separate 'recovery'
and 'shutdown' modes where it was waiting to be triggered by some
other thread, adding unnecessary latency and potential for hangs in
relatively rarely executed startup or shutdown code.
The page cleaner was also running two kinds of batches in an
interleaved fashion: "LRU flush" (writing out some least recently used
pages and evicting them on write completion) and the normal batches
that aim to increase the MIN(oldest_modification) in the buffer pool,
to help the log checkpoint advance.
The buf_pool.flush_list flushing was being blocked by
buf_block_t::lock for no good reason. Furthermore, if the FIL_PAGE_LSN
of a page is ahead of log_sys.get_flushed_lsn(), that is, what has
been persistently written to the redo log, we would trigger a log
flush and then resume the page flushing. This would unnecessarily
limit the performance of the page cleaner thread and trigger the
infamous messages "InnoDB: page_cleaner: 1000ms intended loop took 4450ms.
The settings might not be optimal" that were suppressed in
commit d1ab89037a518fcffbc50c24e4bd94e4ec33aed0 unless log_warnings>2.
Our revised algorithm will make log_sys.get_flushed_lsn() advance at
the start of buf_flush_lists(), and then execute a 'best effort' to
write out all pages. The flush batches will skip pages that were modified
since the log was written, or are are currently exclusively locked.
The MDEV-13670 message "page_cleaner: 1000ms intended loop took" message
will be removed, because by design, the buf_flush_page_cleaner() should
not be blocked during a batch for extended periods of time.
We will remove the single-page flushing altogether. Related to this,
the debug parameter innodb_doublewrite_batch_size will be removed,
because all of the doublewrite buffer will be used for flushing
batches. If a page needs to be evicted from the buffer pool and all
100 least recently used pages in the buffer pool have unflushed
changes, buf_LRU_get_free_block() will execute buf_flush_lists() to
write out and evict innodb_lru_flush_size pages. At most one thread
will execute buf_flush_lists() in buf_LRU_get_free_block(); other
threads will wait for that LRU flushing batch to finish.
To improve concurrency, we will replace the InnoDB ib_mutex_t and
os_event_t native mutexes and condition variables in this area of code.
Most notably, this means that the buffer pool mutex (buf_pool.mutex)
is no longer instrumented via any InnoDB interfaces. It will continue
to be instrumented via PERFORMANCE_SCHEMA.
For now, both buf_pool.flush_list_mutex and buf_pool.mutex will be
declared with MY_MUTEX_INIT_FAST (PTHREAD_MUTEX_ADAPTIVE_NP). The critical
sections of buf_pool.flush_list_mutex should be shorter than those for
buf_pool.mutex, because in the worst case, they cover a linear scan of
buf_pool.flush_list, while the worst case of a critical section of
buf_pool.mutex covers a linear scan of the potentially much longer
buf_pool.LRU list.
mysql_mutex_is_owner(), safe_mutex_is_owner(): New predicate, usable
with SAFE_MUTEX. Some InnoDB debug assertions need this predicate
instead of mysql_mutex_assert_owner() or mysql_mutex_assert_not_owner().
buf_pool_t::n_flush_LRU, buf_pool_t::n_flush_list:
Replaces buf_pool_t::init_flush[] and buf_pool_t::n_flush[].
The number of active flush operations.
buf_pool_t::mutex, buf_pool_t::flush_list_mutex: Use mysql_mutex_t
instead of ib_mutex_t, to have native mutexes with PERFORMANCE_SCHEMA
and SAFE_MUTEX instrumentation.
buf_pool_t::done_flush_LRU: Condition variable for !n_flush_LRU.
buf_pool_t::done_flush_list: Condition variable for !n_flush_list.
buf_pool_t::do_flush_list: Condition variable to wake up the
buf_flush_page_cleaner when a log checkpoint needs to be written
or the server is being shut down. Replaces buf_flush_event.
We will keep using timed waits (the page cleaner thread will wake
_at least_ once per second), because the calculations for
innodb_adaptive_flushing depend on fixed time intervals.
buf_dblwr: Allocate statically, and move all code to member functions.
Use a native mutex and condition variable. Remove code to deal with
single-page flushing.
buf_dblwr_check_block(): Make the check debug-only. We were spending
a significant amount of execution time in page_simple_validate_new().
flush_counters_t::unzip_LRU_evicted: Remove.
IORequest: Make more members const. FIXME: m_fil_node should be removed.
buf_flush_sync_lsn: Protect by std::atomic, not page_cleaner.mutex
(which we are removing).
page_cleaner_slot_t, page_cleaner_t: Remove many redundant members.
pc_request_flush_slot(): Replaces pc_request() and pc_flush_slot().
recv_writer_thread: Remove. Recovery works just fine without it, if we
simply invoke buf_flush_sync() at the end of each batch in
recv_sys_t::apply().
recv_recovery_from_checkpoint_finish(): Remove. We can simply call
recv_sys.debug_free() directly.
srv_started_redo: Replaces srv_start_state.
SRV_SHUTDOWN_FLUSH_PHASE: Remove. logs_empty_and_mark_files_at_shutdown()
can communicate with the normal page cleaner loop via the new function
flush_buffer_pool().
buf_flush_remove(): Assert that the calling thread is holding
buf_pool.flush_list_mutex. This removes unnecessary mutex operations
from buf_flush_remove_pages() and buf_flush_dirty_pages(),
which replace buf_LRU_flush_or_remove_pages().
buf_flush_lists(): Renamed from buf_flush_batch(), with simplified
interface. Return the number of flushed pages. Clarified comments and
renamed min_n to max_n. Identify LRU batch by lsn=0. Merge all the functions
buf_flush_start(), buf_flush_batch(), buf_flush_end() directly to this
function, which was their only caller, and remove 2 unnecessary
buf_pool.mutex release/re-acquisition that we used to perform around
the buf_flush_batch() call. At the start, if not all log has been
durably written, wait for a background task to do it, or start a new
task to do it. This allows the log write to run concurrently with our
page flushing batch. Any pages that were skipped due to too recent
FIL_PAGE_LSN or due to them being latched by a writer should be flushed
during the next batch, unless there are further modifications to those
pages. It is possible that a page that we must flush due to small
oldest_modification also carries a recent FIL_PAGE_LSN or is being
constantly modified. In the worst case, all writers would then end up
waiting in log_free_check() to allow the flushing and the checkpoint
to complete.
buf_do_flush_list_batch(): Clarify comments, and rename min_n to max_n.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_flush_space(): Auxiliary function to look up a tablespace for
page flushing.
buf_flush_page(): Defer the computation of space->full_crc32(). Never
call log_write_up_to(), but instead skip persistent pages whose latest
modification (FIL_PAGE_LSN) is newer than the redo log. Also skip
pages on which we cannot acquire a shared latch without waiting.
buf_flush_try_neighbors(): Do not bother checking buf_fix_count
because buf_flush_page() will no longer wait for the page latch.
Take the tablespace as a parameter, and only execute this function
when innodb_flush_neighbors>0. Avoid repeated calls of page_id_t::fold().
buf_flush_relocate_on_flush_list(): Declare as cold, and push down
a condition from the callers.
buf_flush_check_neighbor(): Take id.fold() as a parameter.
buf_flush_sync(): Ensure that the buf_pool.flush_list is empty,
because the flushing batch will skip pages whose modifications have
not yet been written to the log or were latched for modification.
buf_free_from_unzip_LRU_list_batch(): Remove redundant local variables.
buf_flush_LRU_list_batch(): Let the caller buf_do_LRU_batch() initialize
the counters, and report n->evicted.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_do_LRU_batch(): Return the number of pages flushed.
buf_LRU_free_page(): Only release and re-acquire buf_pool.mutex if
adaptive hash index entries are pointing to the block.
buf_LRU_get_free_block(): Do not wake up the page cleaner, because it
will no longer perform any useful work for us, and we do not want it
to compete for I/O while buf_flush_lists(innodb_lru_flush_size, 0)
writes out and evicts at most innodb_lru_flush_size pages. (The
function buf_do_LRU_batch() may complete after writing fewer pages if
more than innodb_lru_scan_depth pages end up in buf_pool.free list.)
Eliminate some mutex release-acquire cycles, and wait for the LRU
flush batch to complete before rescanning.
buf_LRU_check_size_of_non_data_objects(): Simplify the code.
buf_page_write_complete(): Remove the parameter evict, and always
evict pages that were part of an LRU flush.
buf_page_create(): Take a pre-allocated page as a parameter.
buf_pool_t::free_block(): Free a pre-allocated block.
recv_sys_t::recover_low(), recv_sys_t::apply(): Preallocate the block
while not holding recv_sys.mutex. During page allocation, we may
initiate a page flush, which in turn may initiate a log flush, which
would require acquiring log_sys.mutex, which should always be acquired
before recv_sys.mutex in order to avoid deadlocks. Therefore, we must
not be holding recv_sys.mutex while allocating a buffer pool block.
BtrBulk::logFreeCheck(): Skip a redundant condition.
row_undo_step(): Do not invoke srv_inc_activity_count() for every row
that is being rolled back. It should suffice to invoke the function in
trx_flush_log_if_needed() during trx_t::commit_in_memory() when the
rollback completes.
sync_check_enable(): Remove. We will enable innodb_sync_debug from the
very beginning.
Reviewed by: Vladislav Vaintroub
5 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-30400 Assertion height == btr_page_get_level(...) on INSERT
This also fixes part of MDEV-29835 Partial server freeze
which is caused by violations of the latching order that was
defined in https://dev.mysql.com/worklog/task/?id=6326
(WL#6326: InnoDB: fix index->lock contention). Unless the
current thread is holding an exclusive dict_index_t::lock,
it must acquire page latches in a strict parent-to-child,
left-to-right order. Not all cases of MDEV-29835 are fixed yet.
Failure to follow the correct latching order will cause deadlocks
of threads due to lock order inversion.
As part of these changes, the BTR_MODIFY_TREE mode is modified
so that an Update latch (U a.k.a. SX) will be acquired on the
root page, and eXclusive latches (X) will be acquired on all pages
leading to the leaf page, as well as any left and right siblings
of the pages along the path. The DEBUG_SYNC test innodb.innodb_wl6326
will be removed, because at the time the DEBUG_SYNC point is hit,
the thread is actually holding several page latches that will be
blocking a concurrent SELECT statement.
We also remove double bookkeeping that was caused due to excessive
information hiding in mtr_t::m_memo. We simply let mtr_t::m_memo
store information of latched pages, and ensure that
mtr_memo_slot_t::object is never a null pointer.
The tree_blocks[] and tree_savepoints[] were redundant.
buf_page_get_low(): If innodb_change_buffering_debug=1, to avoid
a hang, do not try to evict blocks if we are holding a latch on
a modified page. The test innodb.innodb-change-buffer-recovery
will be removed, because change buffering may no longer be forced
by debug injection when the change buffer comprises multiple pages.
Remove a debug assertion that could fail when
innodb_change_buffering_debug=1 fails to evict a page.
For other cases, the assertion is redundant, because we already
checked that right after the got_block: label. The test
innodb.innodb-change-buffering-recovery will be removed, because
due to this change, we will be unable to evict the desired page.
mtr_t::lock_register(): Register a change of a page latch
on an unmodified buffer-fixed block.
mtr_t::x_latch_at_savepoint(), mtr_t::sx_latch_at_savepoint():
Replaced by the use of mtr_t::upgrade_buffer_fix(), which now
also handles RW_S_LATCH.
mtr_t::set_modified(): For temporary tables, invoke
buf_page_t::set_modified() here and not in mtr_t::commit().
We will never set the MTR_MEMO_MODIFY flag on other than
persistent data pages, nor set mtr_t::m_modifications when
temporary data pages are modified.
mtr_t::commit(): Only invoke the buf_flush_note_modification() loop
if persistent data pages were modified.
mtr_t::get_already_latched(): Look up a latched page in mtr_t::m_memo.
This avoids many redundant entries in mtr_t::m_memo, as well as
redundant calls to buf_page_get_gen() for blocks that had already
been looked up in a mini-transaction.
btr_get_latched_root(): Return a pointer to an already latched root page.
This replaces btr_root_block_get() in cases where the mini-transaction
has already latched the root page.
btr_page_get_parent(): Fetch a parent page that was already latched
in BTR_MODIFY_TREE, by invoking mtr_t::get_already_latched().
If needed, upgrade the root page U latch to X.
This avoids bloating mtr_t::m_memo as well as performing redundant
buf_pool.page_hash lookups. For non-QUICK CHECK TABLE as well as for
B-tree defragmentation, we will invoke btr_cur_search_to_nth_level().
btr_cur_search_to_nth_level(): This will only be used for non-leaf
(level>0) B-tree searches that were formerly named BTR_CONT_SEARCH_TREE
or BTR_CONT_MODIFY_TREE. In MDEV-29835, this function could be
removed altogether, or retained for the case of
CHECK TABLE without QUICK.
btr_cur_t::left_block: Remove. btr_pcur_move_backward_from_page()
can retrieve the left sibling from the end of mtr_t::m_memo.
btr_cur_t::open_leaf(): Some clean-up.
btr_cur_t::search_leaf(): Replaces btr_cur_search_to_nth_level()
for searches to level=0 (the leaf level). We will never release
parent page latches before acquiring leaf page latches. If we need to
temporarily release the level=1 page latch in the BTR_SEARCH_PREV or
BTR_MODIFY_PREV latch_mode, we will reposition the cursor on the
child node pointer so that we will land on the correct leaf page.
btr_cur_t::pessimistic_search_leaf(): Implement new BTR_MODIFY_TREE
latching logic in the case that page splits or merges will be needed.
The parent pages (and their siblings) should already be latched on
the first dive to the leaf and be present in mtr_t::m_memo; there
should be no need for BTR_CONT_MODIFY_TREE. This pre-latching almost
suffices; it must be revised in MDEV-29835 and work-arounds removed
for cases where mtr_t::get_already_latched() fails to find a block.
rtr_search_to_nth_level(): A SPATIAL INDEX version of
btr_search_to_nth_level() that can search to any level
(including the leaf level).
rtr_search_leaf(), rtr_insert_leaf(): Wrappers for
rtr_search_to_nth_level().
rtr_search(): Replaces rtr_pcur_open().
rtr_latch_leaves(): Replaces btr_cur_latch_leaves(). Note that unlike
in the B-tree code, there is no error handling in case the sibling
pages are corrupted.
rtr_cur_restore_position(): Remove an unused constant parameter.
btr_pcur_open_on_user_rec(): Remove the constant parameter
mode=PAGE_CUR_GE.
row_ins_clust_index_entry_low(): Use a new
mode=BTR_MODIFY_ROOT_AND_LEAF to gain access to the root page
when mode!=BTR_MODIFY_TREE, to write the PAGE_ROOT_AUTO_INC.
BTR_SEARCH_TREE, BTR_CONT_SEARCH_TREE: Remove.
BTR_CONT_MODIFY_TREE: Note that this is only used by
rtr_search_to_nth_level().
btr_pcur_optimistic_latch_leaves(): Replaces
btr_cur_optimistic_latch_leaves().
ibuf_delete_rec(): Acquire exclusive ibuf.index->lock in order
to avoid a deadlock with ibuf_insert_low(BTR_MODIFY_PREV).
btr_blob_log_check_t(): Acquire a U latch on the root page,
so that btr_page_alloc() in btr_store_big_rec_extern_fields()
will avoid a deadlock.
btr_store_big_rec_extern_fields(): Assert that the root page latch
is being held.
Tested by: Matthias Leich
Reviewed by: Vladislav Lesin
3 years ago MDEV-23399: Performance regression with write workloads
The buffer pool refactoring in MDEV-15053 and MDEV-22871 shifted
the performance bottleneck to the page flushing.
The configuration parameters will be changed as follows:
innodb_lru_flush_size=32 (new: how many pages to flush on LRU eviction)
innodb_lru_scan_depth=1536 (old: 1024)
innodb_max_dirty_pages_pct=90 (old: 75)
innodb_max_dirty_pages_pct_lwm=75 (old: 0)
Note: The parameter innodb_lru_scan_depth will only affect LRU
eviction of buffer pool pages when a new page is being allocated. The
page cleaner thread will no longer evict any pages. It used to
guarantee that some pages will remain free in the buffer pool. Now, we
perform that eviction 'on demand' in buf_LRU_get_free_block().
The parameter innodb_lru_scan_depth(srv_LRU_scan_depth) is used as follows:
* When the buffer pool is being shrunk in buf_pool_t::withdraw_blocks()
* As a buf_pool.free limit in buf_LRU_list_batch() for terminating
the flushing that is initiated e.g., by buf_LRU_get_free_block()
The parameter also used to serve as an initial limit for unzip_LRU
eviction (evicting uncompressed page frames while retaining
ROW_FORMAT=COMPRESSED pages), but now we will use a hard-coded limit
of 100 or unlimited for invoking buf_LRU_scan_and_free_block().
The status variables will be changed as follows:
innodb_buffer_pool_pages_flushed: This includes also the count of
innodb_buffer_pool_pages_LRU_flushed and should work reliably,
updated one by one in buf_flush_page() to give more real-time
statistics. The function buf_flush_stats(), which we are removing,
was not called in every code path. For both counters, we will use
regular variables that are incremented in a critical section of
buf_pool.mutex. Note that show_innodb_vars() directly links to the
variables, and reads of the counters will *not* be protected by
buf_pool.mutex, so you cannot get a consistent snapshot of both variables.
The following INFORMATION_SCHEMA.INNODB_METRICS counters will be
removed, because the page cleaner no longer deals with writing or
evicting least recently used pages, and because the single-page writes
have been removed:
* buffer_LRU_batch_flush_avg_time_slot
* buffer_LRU_batch_flush_avg_time_thread
* buffer_LRU_batch_flush_avg_time_est
* buffer_LRU_batch_flush_avg_pass
* buffer_LRU_single_flush_scanned
* buffer_LRU_single_flush_num_scan
* buffer_LRU_single_flush_scanned_per_call
When moving to a single buffer pool instance in MDEV-15058, we missed
some opportunity to simplify the buf_flush_page_cleaner thread. It was
unnecessarily using a mutex and some complex data structures, even
though we always have a single page cleaner thread.
Furthermore, the buf_flush_page_cleaner thread had separate 'recovery'
and 'shutdown' modes where it was waiting to be triggered by some
other thread, adding unnecessary latency and potential for hangs in
relatively rarely executed startup or shutdown code.
The page cleaner was also running two kinds of batches in an
interleaved fashion: "LRU flush" (writing out some least recently used
pages and evicting them on write completion) and the normal batches
that aim to increase the MIN(oldest_modification) in the buffer pool,
to help the log checkpoint advance.
The buf_pool.flush_list flushing was being blocked by
buf_block_t::lock for no good reason. Furthermore, if the FIL_PAGE_LSN
of a page is ahead of log_sys.get_flushed_lsn(), that is, what has
been persistently written to the redo log, we would trigger a log
flush and then resume the page flushing. This would unnecessarily
limit the performance of the page cleaner thread and trigger the
infamous messages "InnoDB: page_cleaner: 1000ms intended loop took 4450ms.
The settings might not be optimal" that were suppressed in
commit d1ab89037a518fcffbc50c24e4bd94e4ec33aed0 unless log_warnings>2.
Our revised algorithm will make log_sys.get_flushed_lsn() advance at
the start of buf_flush_lists(), and then execute a 'best effort' to
write out all pages. The flush batches will skip pages that were modified
since the log was written, or are are currently exclusively locked.
The MDEV-13670 message "page_cleaner: 1000ms intended loop took" message
will be removed, because by design, the buf_flush_page_cleaner() should
not be blocked during a batch for extended periods of time.
We will remove the single-page flushing altogether. Related to this,
the debug parameter innodb_doublewrite_batch_size will be removed,
because all of the doublewrite buffer will be used for flushing
batches. If a page needs to be evicted from the buffer pool and all
100 least recently used pages in the buffer pool have unflushed
changes, buf_LRU_get_free_block() will execute buf_flush_lists() to
write out and evict innodb_lru_flush_size pages. At most one thread
will execute buf_flush_lists() in buf_LRU_get_free_block(); other
threads will wait for that LRU flushing batch to finish.
To improve concurrency, we will replace the InnoDB ib_mutex_t and
os_event_t native mutexes and condition variables in this area of code.
Most notably, this means that the buffer pool mutex (buf_pool.mutex)
is no longer instrumented via any InnoDB interfaces. It will continue
to be instrumented via PERFORMANCE_SCHEMA.
For now, both buf_pool.flush_list_mutex and buf_pool.mutex will be
declared with MY_MUTEX_INIT_FAST (PTHREAD_MUTEX_ADAPTIVE_NP). The critical
sections of buf_pool.flush_list_mutex should be shorter than those for
buf_pool.mutex, because in the worst case, they cover a linear scan of
buf_pool.flush_list, while the worst case of a critical section of
buf_pool.mutex covers a linear scan of the potentially much longer
buf_pool.LRU list.
mysql_mutex_is_owner(), safe_mutex_is_owner(): New predicate, usable
with SAFE_MUTEX. Some InnoDB debug assertions need this predicate
instead of mysql_mutex_assert_owner() or mysql_mutex_assert_not_owner().
buf_pool_t::n_flush_LRU, buf_pool_t::n_flush_list:
Replaces buf_pool_t::init_flush[] and buf_pool_t::n_flush[].
The number of active flush operations.
buf_pool_t::mutex, buf_pool_t::flush_list_mutex: Use mysql_mutex_t
instead of ib_mutex_t, to have native mutexes with PERFORMANCE_SCHEMA
and SAFE_MUTEX instrumentation.
buf_pool_t::done_flush_LRU: Condition variable for !n_flush_LRU.
buf_pool_t::done_flush_list: Condition variable for !n_flush_list.
buf_pool_t::do_flush_list: Condition variable to wake up the
buf_flush_page_cleaner when a log checkpoint needs to be written
or the server is being shut down. Replaces buf_flush_event.
We will keep using timed waits (the page cleaner thread will wake
_at least_ once per second), because the calculations for
innodb_adaptive_flushing depend on fixed time intervals.
buf_dblwr: Allocate statically, and move all code to member functions.
Use a native mutex and condition variable. Remove code to deal with
single-page flushing.
buf_dblwr_check_block(): Make the check debug-only. We were spending
a significant amount of execution time in page_simple_validate_new().
flush_counters_t::unzip_LRU_evicted: Remove.
IORequest: Make more members const. FIXME: m_fil_node should be removed.
buf_flush_sync_lsn: Protect by std::atomic, not page_cleaner.mutex
(which we are removing).
page_cleaner_slot_t, page_cleaner_t: Remove many redundant members.
pc_request_flush_slot(): Replaces pc_request() and pc_flush_slot().
recv_writer_thread: Remove. Recovery works just fine without it, if we
simply invoke buf_flush_sync() at the end of each batch in
recv_sys_t::apply().
recv_recovery_from_checkpoint_finish(): Remove. We can simply call
recv_sys.debug_free() directly.
srv_started_redo: Replaces srv_start_state.
SRV_SHUTDOWN_FLUSH_PHASE: Remove. logs_empty_and_mark_files_at_shutdown()
can communicate with the normal page cleaner loop via the new function
flush_buffer_pool().
buf_flush_remove(): Assert that the calling thread is holding
buf_pool.flush_list_mutex. This removes unnecessary mutex operations
from buf_flush_remove_pages() and buf_flush_dirty_pages(),
which replace buf_LRU_flush_or_remove_pages().
buf_flush_lists(): Renamed from buf_flush_batch(), with simplified
interface. Return the number of flushed pages. Clarified comments and
renamed min_n to max_n. Identify LRU batch by lsn=0. Merge all the functions
buf_flush_start(), buf_flush_batch(), buf_flush_end() directly to this
function, which was their only caller, and remove 2 unnecessary
buf_pool.mutex release/re-acquisition that we used to perform around
the buf_flush_batch() call. At the start, if not all log has been
durably written, wait for a background task to do it, or start a new
task to do it. This allows the log write to run concurrently with our
page flushing batch. Any pages that were skipped due to too recent
FIL_PAGE_LSN or due to them being latched by a writer should be flushed
during the next batch, unless there are further modifications to those
pages. It is possible that a page that we must flush due to small
oldest_modification also carries a recent FIL_PAGE_LSN or is being
constantly modified. In the worst case, all writers would then end up
waiting in log_free_check() to allow the flushing and the checkpoint
to complete.
buf_do_flush_list_batch(): Clarify comments, and rename min_n to max_n.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_flush_space(): Auxiliary function to look up a tablespace for
page flushing.
buf_flush_page(): Defer the computation of space->full_crc32(). Never
call log_write_up_to(), but instead skip persistent pages whose latest
modification (FIL_PAGE_LSN) is newer than the redo log. Also skip
pages on which we cannot acquire a shared latch without waiting.
buf_flush_try_neighbors(): Do not bother checking buf_fix_count
because buf_flush_page() will no longer wait for the page latch.
Take the tablespace as a parameter, and only execute this function
when innodb_flush_neighbors>0. Avoid repeated calls of page_id_t::fold().
buf_flush_relocate_on_flush_list(): Declare as cold, and push down
a condition from the callers.
buf_flush_check_neighbor(): Take id.fold() as a parameter.
buf_flush_sync(): Ensure that the buf_pool.flush_list is empty,
because the flushing batch will skip pages whose modifications have
not yet been written to the log or were latched for modification.
buf_free_from_unzip_LRU_list_batch(): Remove redundant local variables.
buf_flush_LRU_list_batch(): Let the caller buf_do_LRU_batch() initialize
the counters, and report n->evicted.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_do_LRU_batch(): Return the number of pages flushed.
buf_LRU_free_page(): Only release and re-acquire buf_pool.mutex if
adaptive hash index entries are pointing to the block.
buf_LRU_get_free_block(): Do not wake up the page cleaner, because it
will no longer perform any useful work for us, and we do not want it
to compete for I/O while buf_flush_lists(innodb_lru_flush_size, 0)
writes out and evicts at most innodb_lru_flush_size pages. (The
function buf_do_LRU_batch() may complete after writing fewer pages if
more than innodb_lru_scan_depth pages end up in buf_pool.free list.)
Eliminate some mutex release-acquire cycles, and wait for the LRU
flush batch to complete before rescanning.
buf_LRU_check_size_of_non_data_objects(): Simplify the code.
buf_page_write_complete(): Remove the parameter evict, and always
evict pages that were part of an LRU flush.
buf_page_create(): Take a pre-allocated page as a parameter.
buf_pool_t::free_block(): Free a pre-allocated block.
recv_sys_t::recover_low(), recv_sys_t::apply(): Preallocate the block
while not holding recv_sys.mutex. During page allocation, we may
initiate a page flush, which in turn may initiate a log flush, which
would require acquiring log_sys.mutex, which should always be acquired
before recv_sys.mutex in order to avoid deadlocks. Therefore, we must
not be holding recv_sys.mutex while allocating a buffer pool block.
BtrBulk::logFreeCheck(): Skip a redundant condition.
row_undo_step(): Do not invoke srv_inc_activity_count() for every row
that is being rolled back. It should suffice to invoke the function in
trx_flush_log_if_needed() during trx_t::commit_in_memory() when the
rollback completes.
sync_check_enable(): Remove. We will enable innodb_sync_debug from the
very beginning.
Reviewed by: Vladislav Vaintroub
5 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-23399: Performance regression with write workloads
The buffer pool refactoring in MDEV-15053 and MDEV-22871 shifted
the performance bottleneck to the page flushing.
The configuration parameters will be changed as follows:
innodb_lru_flush_size=32 (new: how many pages to flush on LRU eviction)
innodb_lru_scan_depth=1536 (old: 1024)
innodb_max_dirty_pages_pct=90 (old: 75)
innodb_max_dirty_pages_pct_lwm=75 (old: 0)
Note: The parameter innodb_lru_scan_depth will only affect LRU
eviction of buffer pool pages when a new page is being allocated. The
page cleaner thread will no longer evict any pages. It used to
guarantee that some pages will remain free in the buffer pool. Now, we
perform that eviction 'on demand' in buf_LRU_get_free_block().
The parameter innodb_lru_scan_depth(srv_LRU_scan_depth) is used as follows:
* When the buffer pool is being shrunk in buf_pool_t::withdraw_blocks()
* As a buf_pool.free limit in buf_LRU_list_batch() for terminating
the flushing that is initiated e.g., by buf_LRU_get_free_block()
The parameter also used to serve as an initial limit for unzip_LRU
eviction (evicting uncompressed page frames while retaining
ROW_FORMAT=COMPRESSED pages), but now we will use a hard-coded limit
of 100 or unlimited for invoking buf_LRU_scan_and_free_block().
The status variables will be changed as follows:
innodb_buffer_pool_pages_flushed: This includes also the count of
innodb_buffer_pool_pages_LRU_flushed and should work reliably,
updated one by one in buf_flush_page() to give more real-time
statistics. The function buf_flush_stats(), which we are removing,
was not called in every code path. For both counters, we will use
regular variables that are incremented in a critical section of
buf_pool.mutex. Note that show_innodb_vars() directly links to the
variables, and reads of the counters will *not* be protected by
buf_pool.mutex, so you cannot get a consistent snapshot of both variables.
The following INFORMATION_SCHEMA.INNODB_METRICS counters will be
removed, because the page cleaner no longer deals with writing or
evicting least recently used pages, and because the single-page writes
have been removed:
* buffer_LRU_batch_flush_avg_time_slot
* buffer_LRU_batch_flush_avg_time_thread
* buffer_LRU_batch_flush_avg_time_est
* buffer_LRU_batch_flush_avg_pass
* buffer_LRU_single_flush_scanned
* buffer_LRU_single_flush_num_scan
* buffer_LRU_single_flush_scanned_per_call
When moving to a single buffer pool instance in MDEV-15058, we missed
some opportunity to simplify the buf_flush_page_cleaner thread. It was
unnecessarily using a mutex and some complex data structures, even
though we always have a single page cleaner thread.
Furthermore, the buf_flush_page_cleaner thread had separate 'recovery'
and 'shutdown' modes where it was waiting to be triggered by some
other thread, adding unnecessary latency and potential for hangs in
relatively rarely executed startup or shutdown code.
The page cleaner was also running two kinds of batches in an
interleaved fashion: "LRU flush" (writing out some least recently used
pages and evicting them on write completion) and the normal batches
that aim to increase the MIN(oldest_modification) in the buffer pool,
to help the log checkpoint advance.
The buf_pool.flush_list flushing was being blocked by
buf_block_t::lock for no good reason. Furthermore, if the FIL_PAGE_LSN
of a page is ahead of log_sys.get_flushed_lsn(), that is, what has
been persistently written to the redo log, we would trigger a log
flush and then resume the page flushing. This would unnecessarily
limit the performance of the page cleaner thread and trigger the
infamous messages "InnoDB: page_cleaner: 1000ms intended loop took 4450ms.
The settings might not be optimal" that were suppressed in
commit d1ab89037a518fcffbc50c24e4bd94e4ec33aed0 unless log_warnings>2.
Our revised algorithm will make log_sys.get_flushed_lsn() advance at
the start of buf_flush_lists(), and then execute a 'best effort' to
write out all pages. The flush batches will skip pages that were modified
since the log was written, or are are currently exclusively locked.
The MDEV-13670 message "page_cleaner: 1000ms intended loop took" message
will be removed, because by design, the buf_flush_page_cleaner() should
not be blocked during a batch for extended periods of time.
We will remove the single-page flushing altogether. Related to this,
the debug parameter innodb_doublewrite_batch_size will be removed,
because all of the doublewrite buffer will be used for flushing
batches. If a page needs to be evicted from the buffer pool and all
100 least recently used pages in the buffer pool have unflushed
changes, buf_LRU_get_free_block() will execute buf_flush_lists() to
write out and evict innodb_lru_flush_size pages. At most one thread
will execute buf_flush_lists() in buf_LRU_get_free_block(); other
threads will wait for that LRU flushing batch to finish.
To improve concurrency, we will replace the InnoDB ib_mutex_t and
os_event_t native mutexes and condition variables in this area of code.
Most notably, this means that the buffer pool mutex (buf_pool.mutex)
is no longer instrumented via any InnoDB interfaces. It will continue
to be instrumented via PERFORMANCE_SCHEMA.
For now, both buf_pool.flush_list_mutex and buf_pool.mutex will be
declared with MY_MUTEX_INIT_FAST (PTHREAD_MUTEX_ADAPTIVE_NP). The critical
sections of buf_pool.flush_list_mutex should be shorter than those for
buf_pool.mutex, because in the worst case, they cover a linear scan of
buf_pool.flush_list, while the worst case of a critical section of
buf_pool.mutex covers a linear scan of the potentially much longer
buf_pool.LRU list.
mysql_mutex_is_owner(), safe_mutex_is_owner(): New predicate, usable
with SAFE_MUTEX. Some InnoDB debug assertions need this predicate
instead of mysql_mutex_assert_owner() or mysql_mutex_assert_not_owner().
buf_pool_t::n_flush_LRU, buf_pool_t::n_flush_list:
Replaces buf_pool_t::init_flush[] and buf_pool_t::n_flush[].
The number of active flush operations.
buf_pool_t::mutex, buf_pool_t::flush_list_mutex: Use mysql_mutex_t
instead of ib_mutex_t, to have native mutexes with PERFORMANCE_SCHEMA
and SAFE_MUTEX instrumentation.
buf_pool_t::done_flush_LRU: Condition variable for !n_flush_LRU.
buf_pool_t::done_flush_list: Condition variable for !n_flush_list.
buf_pool_t::do_flush_list: Condition variable to wake up the
buf_flush_page_cleaner when a log checkpoint needs to be written
or the server is being shut down. Replaces buf_flush_event.
We will keep using timed waits (the page cleaner thread will wake
_at least_ once per second), because the calculations for
innodb_adaptive_flushing depend on fixed time intervals.
buf_dblwr: Allocate statically, and move all code to member functions.
Use a native mutex and condition variable. Remove code to deal with
single-page flushing.
buf_dblwr_check_block(): Make the check debug-only. We were spending
a significant amount of execution time in page_simple_validate_new().
flush_counters_t::unzip_LRU_evicted: Remove.
IORequest: Make more members const. FIXME: m_fil_node should be removed.
buf_flush_sync_lsn: Protect by std::atomic, not page_cleaner.mutex
(which we are removing).
page_cleaner_slot_t, page_cleaner_t: Remove many redundant members.
pc_request_flush_slot(): Replaces pc_request() and pc_flush_slot().
recv_writer_thread: Remove. Recovery works just fine without it, if we
simply invoke buf_flush_sync() at the end of each batch in
recv_sys_t::apply().
recv_recovery_from_checkpoint_finish(): Remove. We can simply call
recv_sys.debug_free() directly.
srv_started_redo: Replaces srv_start_state.
SRV_SHUTDOWN_FLUSH_PHASE: Remove. logs_empty_and_mark_files_at_shutdown()
can communicate with the normal page cleaner loop via the new function
flush_buffer_pool().
buf_flush_remove(): Assert that the calling thread is holding
buf_pool.flush_list_mutex. This removes unnecessary mutex operations
from buf_flush_remove_pages() and buf_flush_dirty_pages(),
which replace buf_LRU_flush_or_remove_pages().
buf_flush_lists(): Renamed from buf_flush_batch(), with simplified
interface. Return the number of flushed pages. Clarified comments and
renamed min_n to max_n. Identify LRU batch by lsn=0. Merge all the functions
buf_flush_start(), buf_flush_batch(), buf_flush_end() directly to this
function, which was their only caller, and remove 2 unnecessary
buf_pool.mutex release/re-acquisition that we used to perform around
the buf_flush_batch() call. At the start, if not all log has been
durably written, wait for a background task to do it, or start a new
task to do it. This allows the log write to run concurrently with our
page flushing batch. Any pages that were skipped due to too recent
FIL_PAGE_LSN or due to them being latched by a writer should be flushed
during the next batch, unless there are further modifications to those
pages. It is possible that a page that we must flush due to small
oldest_modification also carries a recent FIL_PAGE_LSN or is being
constantly modified. In the worst case, all writers would then end up
waiting in log_free_check() to allow the flushing and the checkpoint
to complete.
buf_do_flush_list_batch(): Clarify comments, and rename min_n to max_n.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_flush_space(): Auxiliary function to look up a tablespace for
page flushing.
buf_flush_page(): Defer the computation of space->full_crc32(). Never
call log_write_up_to(), but instead skip persistent pages whose latest
modification (FIL_PAGE_LSN) is newer than the redo log. Also skip
pages on which we cannot acquire a shared latch without waiting.
buf_flush_try_neighbors(): Do not bother checking buf_fix_count
because buf_flush_page() will no longer wait for the page latch.
Take the tablespace as a parameter, and only execute this function
when innodb_flush_neighbors>0. Avoid repeated calls of page_id_t::fold().
buf_flush_relocate_on_flush_list(): Declare as cold, and push down
a condition from the callers.
buf_flush_check_neighbor(): Take id.fold() as a parameter.
buf_flush_sync(): Ensure that the buf_pool.flush_list is empty,
because the flushing batch will skip pages whose modifications have
not yet been written to the log or were latched for modification.
buf_free_from_unzip_LRU_list_batch(): Remove redundant local variables.
buf_flush_LRU_list_batch(): Let the caller buf_do_LRU_batch() initialize
the counters, and report n->evicted.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_do_LRU_batch(): Return the number of pages flushed.
buf_LRU_free_page(): Only release and re-acquire buf_pool.mutex if
adaptive hash index entries are pointing to the block.
buf_LRU_get_free_block(): Do not wake up the page cleaner, because it
will no longer perform any useful work for us, and we do not want it
to compete for I/O while buf_flush_lists(innodb_lru_flush_size, 0)
writes out and evicts at most innodb_lru_flush_size pages. (The
function buf_do_LRU_batch() may complete after writing fewer pages if
more than innodb_lru_scan_depth pages end up in buf_pool.free list.)
Eliminate some mutex release-acquire cycles, and wait for the LRU
flush batch to complete before rescanning.
buf_LRU_check_size_of_non_data_objects(): Simplify the code.
buf_page_write_complete(): Remove the parameter evict, and always
evict pages that were part of an LRU flush.
buf_page_create(): Take a pre-allocated page as a parameter.
buf_pool_t::free_block(): Free a pre-allocated block.
recv_sys_t::recover_low(), recv_sys_t::apply(): Preallocate the block
while not holding recv_sys.mutex. During page allocation, we may
initiate a page flush, which in turn may initiate a log flush, which
would require acquiring log_sys.mutex, which should always be acquired
before recv_sys.mutex in order to avoid deadlocks. Therefore, we must
not be holding recv_sys.mutex while allocating a buffer pool block.
BtrBulk::logFreeCheck(): Skip a redundant condition.
row_undo_step(): Do not invoke srv_inc_activity_count() for every row
that is being rolled back. It should suffice to invoke the function in
trx_flush_log_if_needed() during trx_t::commit_in_memory() when the
rollback completes.
sync_check_enable(): Remove. We will enable innodb_sync_debug from the
very beginning.
Reviewed by: Vladislav Vaintroub
5 years ago MDEV-27416 InnoDB hang in buf_flush_wait_flushed(), on log checkpoint
InnoDB could sometimes hang when triggering a log checkpoint. This is
due to commit 7b1252c03d7131754d9503560fe507b33ca1f8b4 (MDEV-24278),
which introduced an untimed wait to buf_flush_page_cleaner().
The hang was noticed by occasional failures of IMPORT TABLESPACE tests,
such as innodb.innodb-wl5522, which would (unnecessarily) invoke
log_make_checkpoint() from row_import_cleanup().
The reason of the hang was that buf_flush_page_cleaner() would enter
untimed sleep despite buf_flush_sync_lsn being set. The exact failure
scenario is unclear, because buf_flush_sync_lsn should actually be
protected by buf_pool.flush_list_mutex. We prevent the hang by
invoking buf_pool.page_cleaner_set_idle(false) whenever we are
setting buf_flush_sync_lsn and signaling buf_pool.do_flush_list.
The bulk of these changes was originally developed as a preparation
for MDEV-26827, to invoke buf_flush_list() from fewer threads,
and tested on 10.6 by Matthias Leich.
This fix was tested by running 100 repetitions of 100 concurrent instances
of the test innodb.innodb-wl5522 on a RelWithDebInfo build, using ext4fs
and innodb_flush_method=O_DIRECT on a SATA SSD with 4096-byte block size.
During the test, the call to log_make_checkpoint() in row_import_cleanup()
was present.
buf_flush_list(): Make static.
buf_flush_wait(): Wait for buf_pool.get_oldest_modification()
to reach a target, by work done in the buf_flush_page_cleaner.
If buf_flush_sync_lsn is going to be set, we will invoke
buf_pool.page_cleaner_set_idle(false).
buf_flush_ahead(): If buf_flush_sync_lsn or buf_flush_async_lsn
is going to be set and the page cleaner woken up, we will invoke
buf_pool.page_cleaner_set_idle(false).
buf_flush_wait_flushed(): Invoke buf_flush_wait().
buf_flush_sync(): Invoke recv_sys.apply() at the start in case
crash recovery is active. Invoke buf_flush_wait().
buf_flush_sync_batch(): A lower-level variant of buf_flush_sync()
that is only called by recv_sys_t::apply().
buf_flush_sync_for_checkpoint(): Do not trigger log apply
or checkpoint during recovery.
buf_dblwr_t::create(): Only initiate a buffer pool flush, not
a checkpoint.
row_import_cleanup(): Do not unnecessarily invoke log_make_checkpoint().
Invoking buf_flush_list_space() before starting to generate redo log
for the imported tablespace should suffice.
srv_prepare_to_delete_redo_log_file():
Set recv_sys.recovery_on in order to prevent
buf_flush_sync_for_checkpoint() from initiating a checkpoint
while the log is inaccessible. Remove a wait loop that is already
part of buf_flush_sync().
Do not invoke fil_names_clear() if the log is being upgraded,
because the FILE_MODIFY record is specific to the latest format.
create_log_file(): Clear recv_sys.recovery_on only after calling
log_make_checkpoint(), to prevent buf_flush_page_cleaner from
invoking a checkpoint.
innodb_shutdown(): Simplify the logic in mariadb-backup --prepare.
os_aio_wait_until_no_pending_writes(): Update the function comment.
Apart from row_quiesce_table_start() during FLUSH TABLES...FOR EXPORT,
this is being called by buf_flush_list_space(), which is invoked
by ALTER TABLE...IMPORT TABLESPACE as well as some encryption operations.
4 years ago MDEV-23399: Performance regression with write workloads
The buffer pool refactoring in MDEV-15053 and MDEV-22871 shifted
the performance bottleneck to the page flushing.
The configuration parameters will be changed as follows:
innodb_lru_flush_size=32 (new: how many pages to flush on LRU eviction)
innodb_lru_scan_depth=1536 (old: 1024)
innodb_max_dirty_pages_pct=90 (old: 75)
innodb_max_dirty_pages_pct_lwm=75 (old: 0)
Note: The parameter innodb_lru_scan_depth will only affect LRU
eviction of buffer pool pages when a new page is being allocated. The
page cleaner thread will no longer evict any pages. It used to
guarantee that some pages will remain free in the buffer pool. Now, we
perform that eviction 'on demand' in buf_LRU_get_free_block().
The parameter innodb_lru_scan_depth(srv_LRU_scan_depth) is used as follows:
* When the buffer pool is being shrunk in buf_pool_t::withdraw_blocks()
* As a buf_pool.free limit in buf_LRU_list_batch() for terminating
the flushing that is initiated e.g., by buf_LRU_get_free_block()
The parameter also used to serve as an initial limit for unzip_LRU
eviction (evicting uncompressed page frames while retaining
ROW_FORMAT=COMPRESSED pages), but now we will use a hard-coded limit
of 100 or unlimited for invoking buf_LRU_scan_and_free_block().
The status variables will be changed as follows:
innodb_buffer_pool_pages_flushed: This includes also the count of
innodb_buffer_pool_pages_LRU_flushed and should work reliably,
updated one by one in buf_flush_page() to give more real-time
statistics. The function buf_flush_stats(), which we are removing,
was not called in every code path. For both counters, we will use
regular variables that are incremented in a critical section of
buf_pool.mutex. Note that show_innodb_vars() directly links to the
variables, and reads of the counters will *not* be protected by
buf_pool.mutex, so you cannot get a consistent snapshot of both variables.
The following INFORMATION_SCHEMA.INNODB_METRICS counters will be
removed, because the page cleaner no longer deals with writing or
evicting least recently used pages, and because the single-page writes
have been removed:
* buffer_LRU_batch_flush_avg_time_slot
* buffer_LRU_batch_flush_avg_time_thread
* buffer_LRU_batch_flush_avg_time_est
* buffer_LRU_batch_flush_avg_pass
* buffer_LRU_single_flush_scanned
* buffer_LRU_single_flush_num_scan
* buffer_LRU_single_flush_scanned_per_call
When moving to a single buffer pool instance in MDEV-15058, we missed
some opportunity to simplify the buf_flush_page_cleaner thread. It was
unnecessarily using a mutex and some complex data structures, even
though we always have a single page cleaner thread.
Furthermore, the buf_flush_page_cleaner thread had separate 'recovery'
and 'shutdown' modes where it was waiting to be triggered by some
other thread, adding unnecessary latency and potential for hangs in
relatively rarely executed startup or shutdown code.
The page cleaner was also running two kinds of batches in an
interleaved fashion: "LRU flush" (writing out some least recently used
pages and evicting them on write completion) and the normal batches
that aim to increase the MIN(oldest_modification) in the buffer pool,
to help the log checkpoint advance.
The buf_pool.flush_list flushing was being blocked by
buf_block_t::lock for no good reason. Furthermore, if the FIL_PAGE_LSN
of a page is ahead of log_sys.get_flushed_lsn(), that is, what has
been persistently written to the redo log, we would trigger a log
flush and then resume the page flushing. This would unnecessarily
limit the performance of the page cleaner thread and trigger the
infamous messages "InnoDB: page_cleaner: 1000ms intended loop took 4450ms.
The settings might not be optimal" that were suppressed in
commit d1ab89037a518fcffbc50c24e4bd94e4ec33aed0 unless log_warnings>2.
Our revised algorithm will make log_sys.get_flushed_lsn() advance at
the start of buf_flush_lists(), and then execute a 'best effort' to
write out all pages. The flush batches will skip pages that were modified
since the log was written, or are are currently exclusively locked.
The MDEV-13670 message "page_cleaner: 1000ms intended loop took" message
will be removed, because by design, the buf_flush_page_cleaner() should
not be blocked during a batch for extended periods of time.
We will remove the single-page flushing altogether. Related to this,
the debug parameter innodb_doublewrite_batch_size will be removed,
because all of the doublewrite buffer will be used for flushing
batches. If a page needs to be evicted from the buffer pool and all
100 least recently used pages in the buffer pool have unflushed
changes, buf_LRU_get_free_block() will execute buf_flush_lists() to
write out and evict innodb_lru_flush_size pages. At most one thread
will execute buf_flush_lists() in buf_LRU_get_free_block(); other
threads will wait for that LRU flushing batch to finish.
To improve concurrency, we will replace the InnoDB ib_mutex_t and
os_event_t native mutexes and condition variables in this area of code.
Most notably, this means that the buffer pool mutex (buf_pool.mutex)
is no longer instrumented via any InnoDB interfaces. It will continue
to be instrumented via PERFORMANCE_SCHEMA.
For now, both buf_pool.flush_list_mutex and buf_pool.mutex will be
declared with MY_MUTEX_INIT_FAST (PTHREAD_MUTEX_ADAPTIVE_NP). The critical
sections of buf_pool.flush_list_mutex should be shorter than those for
buf_pool.mutex, because in the worst case, they cover a linear scan of
buf_pool.flush_list, while the worst case of a critical section of
buf_pool.mutex covers a linear scan of the potentially much longer
buf_pool.LRU list.
mysql_mutex_is_owner(), safe_mutex_is_owner(): New predicate, usable
with SAFE_MUTEX. Some InnoDB debug assertions need this predicate
instead of mysql_mutex_assert_owner() or mysql_mutex_assert_not_owner().
buf_pool_t::n_flush_LRU, buf_pool_t::n_flush_list:
Replaces buf_pool_t::init_flush[] and buf_pool_t::n_flush[].
The number of active flush operations.
buf_pool_t::mutex, buf_pool_t::flush_list_mutex: Use mysql_mutex_t
instead of ib_mutex_t, to have native mutexes with PERFORMANCE_SCHEMA
and SAFE_MUTEX instrumentation.
buf_pool_t::done_flush_LRU: Condition variable for !n_flush_LRU.
buf_pool_t::done_flush_list: Condition variable for !n_flush_list.
buf_pool_t::do_flush_list: Condition variable to wake up the
buf_flush_page_cleaner when a log checkpoint needs to be written
or the server is being shut down. Replaces buf_flush_event.
We will keep using timed waits (the page cleaner thread will wake
_at least_ once per second), because the calculations for
innodb_adaptive_flushing depend on fixed time intervals.
buf_dblwr: Allocate statically, and move all code to member functions.
Use a native mutex and condition variable. Remove code to deal with
single-page flushing.
buf_dblwr_check_block(): Make the check debug-only. We were spending
a significant amount of execution time in page_simple_validate_new().
flush_counters_t::unzip_LRU_evicted: Remove.
IORequest: Make more members const. FIXME: m_fil_node should be removed.
buf_flush_sync_lsn: Protect by std::atomic, not page_cleaner.mutex
(which we are removing).
page_cleaner_slot_t, page_cleaner_t: Remove many redundant members.
pc_request_flush_slot(): Replaces pc_request() and pc_flush_slot().
recv_writer_thread: Remove. Recovery works just fine without it, if we
simply invoke buf_flush_sync() at the end of each batch in
recv_sys_t::apply().
recv_recovery_from_checkpoint_finish(): Remove. We can simply call
recv_sys.debug_free() directly.
srv_started_redo: Replaces srv_start_state.
SRV_SHUTDOWN_FLUSH_PHASE: Remove. logs_empty_and_mark_files_at_shutdown()
can communicate with the normal page cleaner loop via the new function
flush_buffer_pool().
buf_flush_remove(): Assert that the calling thread is holding
buf_pool.flush_list_mutex. This removes unnecessary mutex operations
from buf_flush_remove_pages() and buf_flush_dirty_pages(),
which replace buf_LRU_flush_or_remove_pages().
buf_flush_lists(): Renamed from buf_flush_batch(), with simplified
interface. Return the number of flushed pages. Clarified comments and
renamed min_n to max_n. Identify LRU batch by lsn=0. Merge all the functions
buf_flush_start(), buf_flush_batch(), buf_flush_end() directly to this
function, which was their only caller, and remove 2 unnecessary
buf_pool.mutex release/re-acquisition that we used to perform around
the buf_flush_batch() call. At the start, if not all log has been
durably written, wait for a background task to do it, or start a new
task to do it. This allows the log write to run concurrently with our
page flushing batch. Any pages that were skipped due to too recent
FIL_PAGE_LSN or due to them being latched by a writer should be flushed
during the next batch, unless there are further modifications to those
pages. It is possible that a page that we must flush due to small
oldest_modification also carries a recent FIL_PAGE_LSN or is being
constantly modified. In the worst case, all writers would then end up
waiting in log_free_check() to allow the flushing and the checkpoint
to complete.
buf_do_flush_list_batch(): Clarify comments, and rename min_n to max_n.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_flush_space(): Auxiliary function to look up a tablespace for
page flushing.
buf_flush_page(): Defer the computation of space->full_crc32(). Never
call log_write_up_to(), but instead skip persistent pages whose latest
modification (FIL_PAGE_LSN) is newer than the redo log. Also skip
pages on which we cannot acquire a shared latch without waiting.
buf_flush_try_neighbors(): Do not bother checking buf_fix_count
because buf_flush_page() will no longer wait for the page latch.
Take the tablespace as a parameter, and only execute this function
when innodb_flush_neighbors>0. Avoid repeated calls of page_id_t::fold().
buf_flush_relocate_on_flush_list(): Declare as cold, and push down
a condition from the callers.
buf_flush_check_neighbor(): Take id.fold() as a parameter.
buf_flush_sync(): Ensure that the buf_pool.flush_list is empty,
because the flushing batch will skip pages whose modifications have
not yet been written to the log or were latched for modification.
buf_free_from_unzip_LRU_list_batch(): Remove redundant local variables.
buf_flush_LRU_list_batch(): Let the caller buf_do_LRU_batch() initialize
the counters, and report n->evicted.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_do_LRU_batch(): Return the number of pages flushed.
buf_LRU_free_page(): Only release and re-acquire buf_pool.mutex if
adaptive hash index entries are pointing to the block.
buf_LRU_get_free_block(): Do not wake up the page cleaner, because it
will no longer perform any useful work for us, and we do not want it
to compete for I/O while buf_flush_lists(innodb_lru_flush_size, 0)
writes out and evicts at most innodb_lru_flush_size pages. (The
function buf_do_LRU_batch() may complete after writing fewer pages if
more than innodb_lru_scan_depth pages end up in buf_pool.free list.)
Eliminate some mutex release-acquire cycles, and wait for the LRU
flush batch to complete before rescanning.
buf_LRU_check_size_of_non_data_objects(): Simplify the code.
buf_page_write_complete(): Remove the parameter evict, and always
evict pages that were part of an LRU flush.
buf_page_create(): Take a pre-allocated page as a parameter.
buf_pool_t::free_block(): Free a pre-allocated block.
recv_sys_t::recover_low(), recv_sys_t::apply(): Preallocate the block
while not holding recv_sys.mutex. During page allocation, we may
initiate a page flush, which in turn may initiate a log flush, which
would require acquiring log_sys.mutex, which should always be acquired
before recv_sys.mutex in order to avoid deadlocks. Therefore, we must
not be holding recv_sys.mutex while allocating a buffer pool block.
BtrBulk::logFreeCheck(): Skip a redundant condition.
row_undo_step(): Do not invoke srv_inc_activity_count() for every row
that is being rolled back. It should suffice to invoke the function in
trx_flush_log_if_needed() during trx_t::commit_in_memory() when the
rollback completes.
sync_check_enable(): Remove. We will enable innodb_sync_debug from the
very beginning.
Reviewed by: Vladislav Vaintroub
5 years ago MDEV-23399: Performance regression with write workloads
The buffer pool refactoring in MDEV-15053 and MDEV-22871 shifted
the performance bottleneck to the page flushing.
The configuration parameters will be changed as follows:
innodb_lru_flush_size=32 (new: how many pages to flush on LRU eviction)
innodb_lru_scan_depth=1536 (old: 1024)
innodb_max_dirty_pages_pct=90 (old: 75)
innodb_max_dirty_pages_pct_lwm=75 (old: 0)
Note: The parameter innodb_lru_scan_depth will only affect LRU
eviction of buffer pool pages when a new page is being allocated. The
page cleaner thread will no longer evict any pages. It used to
guarantee that some pages will remain free in the buffer pool. Now, we
perform that eviction 'on demand' in buf_LRU_get_free_block().
The parameter innodb_lru_scan_depth(srv_LRU_scan_depth) is used as follows:
* When the buffer pool is being shrunk in buf_pool_t::withdraw_blocks()
* As a buf_pool.free limit in buf_LRU_list_batch() for terminating
the flushing that is initiated e.g., by buf_LRU_get_free_block()
The parameter also used to serve as an initial limit for unzip_LRU
eviction (evicting uncompressed page frames while retaining
ROW_FORMAT=COMPRESSED pages), but now we will use a hard-coded limit
of 100 or unlimited for invoking buf_LRU_scan_and_free_block().
The status variables will be changed as follows:
innodb_buffer_pool_pages_flushed: This includes also the count of
innodb_buffer_pool_pages_LRU_flushed and should work reliably,
updated one by one in buf_flush_page() to give more real-time
statistics. The function buf_flush_stats(), which we are removing,
was not called in every code path. For both counters, we will use
regular variables that are incremented in a critical section of
buf_pool.mutex. Note that show_innodb_vars() directly links to the
variables, and reads of the counters will *not* be protected by
buf_pool.mutex, so you cannot get a consistent snapshot of both variables.
The following INFORMATION_SCHEMA.INNODB_METRICS counters will be
removed, because the page cleaner no longer deals with writing or
evicting least recently used pages, and because the single-page writes
have been removed:
* buffer_LRU_batch_flush_avg_time_slot
* buffer_LRU_batch_flush_avg_time_thread
* buffer_LRU_batch_flush_avg_time_est
* buffer_LRU_batch_flush_avg_pass
* buffer_LRU_single_flush_scanned
* buffer_LRU_single_flush_num_scan
* buffer_LRU_single_flush_scanned_per_call
When moving to a single buffer pool instance in MDEV-15058, we missed
some opportunity to simplify the buf_flush_page_cleaner thread. It was
unnecessarily using a mutex and some complex data structures, even
though we always have a single page cleaner thread.
Furthermore, the buf_flush_page_cleaner thread had separate 'recovery'
and 'shutdown' modes where it was waiting to be triggered by some
other thread, adding unnecessary latency and potential for hangs in
relatively rarely executed startup or shutdown code.
The page cleaner was also running two kinds of batches in an
interleaved fashion: "LRU flush" (writing out some least recently used
pages and evicting them on write completion) and the normal batches
that aim to increase the MIN(oldest_modification) in the buffer pool,
to help the log checkpoint advance.
The buf_pool.flush_list flushing was being blocked by
buf_block_t::lock for no good reason. Furthermore, if the FIL_PAGE_LSN
of a page is ahead of log_sys.get_flushed_lsn(), that is, what has
been persistently written to the redo log, we would trigger a log
flush and then resume the page flushing. This would unnecessarily
limit the performance of the page cleaner thread and trigger the
infamous messages "InnoDB: page_cleaner: 1000ms intended loop took 4450ms.
The settings might not be optimal" that were suppressed in
commit d1ab89037a518fcffbc50c24e4bd94e4ec33aed0 unless log_warnings>2.
Our revised algorithm will make log_sys.get_flushed_lsn() advance at
the start of buf_flush_lists(), and then execute a 'best effort' to
write out all pages. The flush batches will skip pages that were modified
since the log was written, or are are currently exclusively locked.
The MDEV-13670 message "page_cleaner: 1000ms intended loop took" message
will be removed, because by design, the buf_flush_page_cleaner() should
not be blocked during a batch for extended periods of time.
We will remove the single-page flushing altogether. Related to this,
the debug parameter innodb_doublewrite_batch_size will be removed,
because all of the doublewrite buffer will be used for flushing
batches. If a page needs to be evicted from the buffer pool and all
100 least recently used pages in the buffer pool have unflushed
changes, buf_LRU_get_free_block() will execute buf_flush_lists() to
write out and evict innodb_lru_flush_size pages. At most one thread
will execute buf_flush_lists() in buf_LRU_get_free_block(); other
threads will wait for that LRU flushing batch to finish.
To improve concurrency, we will replace the InnoDB ib_mutex_t and
os_event_t native mutexes and condition variables in this area of code.
Most notably, this means that the buffer pool mutex (buf_pool.mutex)
is no longer instrumented via any InnoDB interfaces. It will continue
to be instrumented via PERFORMANCE_SCHEMA.
For now, both buf_pool.flush_list_mutex and buf_pool.mutex will be
declared with MY_MUTEX_INIT_FAST (PTHREAD_MUTEX_ADAPTIVE_NP). The critical
sections of buf_pool.flush_list_mutex should be shorter than those for
buf_pool.mutex, because in the worst case, they cover a linear scan of
buf_pool.flush_list, while the worst case of a critical section of
buf_pool.mutex covers a linear scan of the potentially much longer
buf_pool.LRU list.
mysql_mutex_is_owner(), safe_mutex_is_owner(): New predicate, usable
with SAFE_MUTEX. Some InnoDB debug assertions need this predicate
instead of mysql_mutex_assert_owner() or mysql_mutex_assert_not_owner().
buf_pool_t::n_flush_LRU, buf_pool_t::n_flush_list:
Replaces buf_pool_t::init_flush[] and buf_pool_t::n_flush[].
The number of active flush operations.
buf_pool_t::mutex, buf_pool_t::flush_list_mutex: Use mysql_mutex_t
instead of ib_mutex_t, to have native mutexes with PERFORMANCE_SCHEMA
and SAFE_MUTEX instrumentation.
buf_pool_t::done_flush_LRU: Condition variable for !n_flush_LRU.
buf_pool_t::done_flush_list: Condition variable for !n_flush_list.
buf_pool_t::do_flush_list: Condition variable to wake up the
buf_flush_page_cleaner when a log checkpoint needs to be written
or the server is being shut down. Replaces buf_flush_event.
We will keep using timed waits (the page cleaner thread will wake
_at least_ once per second), because the calculations for
innodb_adaptive_flushing depend on fixed time intervals.
buf_dblwr: Allocate statically, and move all code to member functions.
Use a native mutex and condition variable. Remove code to deal with
single-page flushing.
buf_dblwr_check_block(): Make the check debug-only. We were spending
a significant amount of execution time in page_simple_validate_new().
flush_counters_t::unzip_LRU_evicted: Remove.
IORequest: Make more members const. FIXME: m_fil_node should be removed.
buf_flush_sync_lsn: Protect by std::atomic, not page_cleaner.mutex
(which we are removing).
page_cleaner_slot_t, page_cleaner_t: Remove many redundant members.
pc_request_flush_slot(): Replaces pc_request() and pc_flush_slot().
recv_writer_thread: Remove. Recovery works just fine without it, if we
simply invoke buf_flush_sync() at the end of each batch in
recv_sys_t::apply().
recv_recovery_from_checkpoint_finish(): Remove. We can simply call
recv_sys.debug_free() directly.
srv_started_redo: Replaces srv_start_state.
SRV_SHUTDOWN_FLUSH_PHASE: Remove. logs_empty_and_mark_files_at_shutdown()
can communicate with the normal page cleaner loop via the new function
flush_buffer_pool().
buf_flush_remove(): Assert that the calling thread is holding
buf_pool.flush_list_mutex. This removes unnecessary mutex operations
from buf_flush_remove_pages() and buf_flush_dirty_pages(),
which replace buf_LRU_flush_or_remove_pages().
buf_flush_lists(): Renamed from buf_flush_batch(), with simplified
interface. Return the number of flushed pages. Clarified comments and
renamed min_n to max_n. Identify LRU batch by lsn=0. Merge all the functions
buf_flush_start(), buf_flush_batch(), buf_flush_end() directly to this
function, which was their only caller, and remove 2 unnecessary
buf_pool.mutex release/re-acquisition that we used to perform around
the buf_flush_batch() call. At the start, if not all log has been
durably written, wait for a background task to do it, or start a new
task to do it. This allows the log write to run concurrently with our
page flushing batch. Any pages that were skipped due to too recent
FIL_PAGE_LSN or due to them being latched by a writer should be flushed
during the next batch, unless there are further modifications to those
pages. It is possible that a page that we must flush due to small
oldest_modification also carries a recent FIL_PAGE_LSN or is being
constantly modified. In the worst case, all writers would then end up
waiting in log_free_check() to allow the flushing and the checkpoint
to complete.
buf_do_flush_list_batch(): Clarify comments, and rename min_n to max_n.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_flush_space(): Auxiliary function to look up a tablespace for
page flushing.
buf_flush_page(): Defer the computation of space->full_crc32(). Never
call log_write_up_to(), but instead skip persistent pages whose latest
modification (FIL_PAGE_LSN) is newer than the redo log. Also skip
pages on which we cannot acquire a shared latch without waiting.
buf_flush_try_neighbors(): Do not bother checking buf_fix_count
because buf_flush_page() will no longer wait for the page latch.
Take the tablespace as a parameter, and only execute this function
when innodb_flush_neighbors>0. Avoid repeated calls of page_id_t::fold().
buf_flush_relocate_on_flush_list(): Declare as cold, and push down
a condition from the callers.
buf_flush_check_neighbor(): Take id.fold() as a parameter.
buf_flush_sync(): Ensure that the buf_pool.flush_list is empty,
because the flushing batch will skip pages whose modifications have
not yet been written to the log or were latched for modification.
buf_free_from_unzip_LRU_list_batch(): Remove redundant local variables.
buf_flush_LRU_list_batch(): Let the caller buf_do_LRU_batch() initialize
the counters, and report n->evicted.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_do_LRU_batch(): Return the number of pages flushed.
buf_LRU_free_page(): Only release and re-acquire buf_pool.mutex if
adaptive hash index entries are pointing to the block.
buf_LRU_get_free_block(): Do not wake up the page cleaner, because it
will no longer perform any useful work for us, and we do not want it
to compete for I/O while buf_flush_lists(innodb_lru_flush_size, 0)
writes out and evicts at most innodb_lru_flush_size pages. (The
function buf_do_LRU_batch() may complete after writing fewer pages if
more than innodb_lru_scan_depth pages end up in buf_pool.free list.)
Eliminate some mutex release-acquire cycles, and wait for the LRU
flush batch to complete before rescanning.
buf_LRU_check_size_of_non_data_objects(): Simplify the code.
buf_page_write_complete(): Remove the parameter evict, and always
evict pages that were part of an LRU flush.
buf_page_create(): Take a pre-allocated page as a parameter.
buf_pool_t::free_block(): Free a pre-allocated block.
recv_sys_t::recover_low(), recv_sys_t::apply(): Preallocate the block
while not holding recv_sys.mutex. During page allocation, we may
initiate a page flush, which in turn may initiate a log flush, which
would require acquiring log_sys.mutex, which should always be acquired
before recv_sys.mutex in order to avoid deadlocks. Therefore, we must
not be holding recv_sys.mutex while allocating a buffer pool block.
BtrBulk::logFreeCheck(): Skip a redundant condition.
row_undo_step(): Do not invoke srv_inc_activity_count() for every row
that is being rolled back. It should suffice to invoke the function in
trx_flush_log_if_needed() during trx_t::commit_in_memory() when the
rollback completes.
sync_check_enable(): Remove. We will enable innodb_sync_debug from the
very beginning.
Reviewed by: Vladislav Vaintroub
5 years ago MDEV-30400 Assertion height == btr_page_get_level(...) on INSERT
This also fixes part of MDEV-29835 Partial server freeze
which is caused by violations of the latching order that was
defined in https://dev.mysql.com/worklog/task/?id=6326
(WL#6326: InnoDB: fix index->lock contention). Unless the
current thread is holding an exclusive dict_index_t::lock,
it must acquire page latches in a strict parent-to-child,
left-to-right order. Not all cases of MDEV-29835 are fixed yet.
Failure to follow the correct latching order will cause deadlocks
of threads due to lock order inversion.
As part of these changes, the BTR_MODIFY_TREE mode is modified
so that an Update latch (U a.k.a. SX) will be acquired on the
root page, and eXclusive latches (X) will be acquired on all pages
leading to the leaf page, as well as any left and right siblings
of the pages along the path. The DEBUG_SYNC test innodb.innodb_wl6326
will be removed, because at the time the DEBUG_SYNC point is hit,
the thread is actually holding several page latches that will be
blocking a concurrent SELECT statement.
We also remove double bookkeeping that was caused due to excessive
information hiding in mtr_t::m_memo. We simply let mtr_t::m_memo
store information of latched pages, and ensure that
mtr_memo_slot_t::object is never a null pointer.
The tree_blocks[] and tree_savepoints[] were redundant.
buf_page_get_low(): If innodb_change_buffering_debug=1, to avoid
a hang, do not try to evict blocks if we are holding a latch on
a modified page. The test innodb.innodb-change-buffer-recovery
will be removed, because change buffering may no longer be forced
by debug injection when the change buffer comprises multiple pages.
Remove a debug assertion that could fail when
innodb_change_buffering_debug=1 fails to evict a page.
For other cases, the assertion is redundant, because we already
checked that right after the got_block: label. The test
innodb.innodb-change-buffering-recovery will be removed, because
due to this change, we will be unable to evict the desired page.
mtr_t::lock_register(): Register a change of a page latch
on an unmodified buffer-fixed block.
mtr_t::x_latch_at_savepoint(), mtr_t::sx_latch_at_savepoint():
Replaced by the use of mtr_t::upgrade_buffer_fix(), which now
also handles RW_S_LATCH.
mtr_t::set_modified(): For temporary tables, invoke
buf_page_t::set_modified() here and not in mtr_t::commit().
We will never set the MTR_MEMO_MODIFY flag on other than
persistent data pages, nor set mtr_t::m_modifications when
temporary data pages are modified.
mtr_t::commit(): Only invoke the buf_flush_note_modification() loop
if persistent data pages were modified.
mtr_t::get_already_latched(): Look up a latched page in mtr_t::m_memo.
This avoids many redundant entries in mtr_t::m_memo, as well as
redundant calls to buf_page_get_gen() for blocks that had already
been looked up in a mini-transaction.
btr_get_latched_root(): Return a pointer to an already latched root page.
This replaces btr_root_block_get() in cases where the mini-transaction
has already latched the root page.
btr_page_get_parent(): Fetch a parent page that was already latched
in BTR_MODIFY_TREE, by invoking mtr_t::get_already_latched().
If needed, upgrade the root page U latch to X.
This avoids bloating mtr_t::m_memo as well as performing redundant
buf_pool.page_hash lookups. For non-QUICK CHECK TABLE as well as for
B-tree defragmentation, we will invoke btr_cur_search_to_nth_level().
btr_cur_search_to_nth_level(): This will only be used for non-leaf
(level>0) B-tree searches that were formerly named BTR_CONT_SEARCH_TREE
or BTR_CONT_MODIFY_TREE. In MDEV-29835, this function could be
removed altogether, or retained for the case of
CHECK TABLE without QUICK.
btr_cur_t::left_block: Remove. btr_pcur_move_backward_from_page()
can retrieve the left sibling from the end of mtr_t::m_memo.
btr_cur_t::open_leaf(): Some clean-up.
btr_cur_t::search_leaf(): Replaces btr_cur_search_to_nth_level()
for searches to level=0 (the leaf level). We will never release
parent page latches before acquiring leaf page latches. If we need to
temporarily release the level=1 page latch in the BTR_SEARCH_PREV or
BTR_MODIFY_PREV latch_mode, we will reposition the cursor on the
child node pointer so that we will land on the correct leaf page.
btr_cur_t::pessimistic_search_leaf(): Implement new BTR_MODIFY_TREE
latching logic in the case that page splits or merges will be needed.
The parent pages (and their siblings) should already be latched on
the first dive to the leaf and be present in mtr_t::m_memo; there
should be no need for BTR_CONT_MODIFY_TREE. This pre-latching almost
suffices; it must be revised in MDEV-29835 and work-arounds removed
for cases where mtr_t::get_already_latched() fails to find a block.
rtr_search_to_nth_level(): A SPATIAL INDEX version of
btr_search_to_nth_level() that can search to any level
(including the leaf level).
rtr_search_leaf(), rtr_insert_leaf(): Wrappers for
rtr_search_to_nth_level().
rtr_search(): Replaces rtr_pcur_open().
rtr_latch_leaves(): Replaces btr_cur_latch_leaves(). Note that unlike
in the B-tree code, there is no error handling in case the sibling
pages are corrupted.
rtr_cur_restore_position(): Remove an unused constant parameter.
btr_pcur_open_on_user_rec(): Remove the constant parameter
mode=PAGE_CUR_GE.
row_ins_clust_index_entry_low(): Use a new
mode=BTR_MODIFY_ROOT_AND_LEAF to gain access to the root page
when mode!=BTR_MODIFY_TREE, to write the PAGE_ROOT_AUTO_INC.
BTR_SEARCH_TREE, BTR_CONT_SEARCH_TREE: Remove.
BTR_CONT_MODIFY_TREE: Note that this is only used by
rtr_search_to_nth_level().
btr_pcur_optimistic_latch_leaves(): Replaces
btr_cur_optimistic_latch_leaves().
ibuf_delete_rec(): Acquire exclusive ibuf.index->lock in order
to avoid a deadlock with ibuf_insert_low(BTR_MODIFY_PREV).
btr_blob_log_check_t(): Acquire a U latch on the root page,
so that btr_page_alloc() in btr_store_big_rec_extern_fields()
will avoid a deadlock.
btr_store_big_rec_extern_fields(): Assert that the root page latch
is being held.
Tested by: Matthias Leich
Reviewed by: Vladislav Lesin
3 years ago MDEV-13542: Crashing on corrupted page is unhelpful
The approach to handling corruption that was chosen by Oracle in
commit 177d8b0c125b841c0650d27d735e3b87509dc286
is not really useful. Not only did it actually fail to prevent InnoDB
from crashing, but it is making things worse by blocking attempts to
rescue data from or rebuild a partially readable table.
We will try to prevent crashes in a different way: by propagating
errors up the call stack. We will never mark the clustered index
persistently corrupted, so that data recovery may be attempted by
reading from the table, or by rebuilding the table.
This should also fix MDEV-13680 (crash on btr_page_alloc() failure);
it was extensively tested with innodb_file_per_table=0 and a
non-autoextend system tablespace.
We should now avoid crashes in many cases, such as when a page
cannot be read or allocated, or an inconsistency is detected when
attempting to update multiple pages. We will not crash on double-free,
such as on the recovery of DDL in system tablespace in case something
was corrupted.
Crashes on corrupted data are still possible. The fault injection mechanism
that is introduced in the subsequent commit may help catch more of them.
buf_page_import_corrupt_failure: Remove the fault injection, and instead
corrupt some pages using Perl code in the tests.
btr_cur_pessimistic_insert(): Always reserve extents (except for the
change buffer), in order to prevent a subsequent allocation failure.
btr_pcur_open_at_rnd_pos(): Merged to the only caller ibuf_merge_pages().
btr_assert_not_corrupted(), btr_corruption_report(): Remove.
Similar checks are already part of btr_block_get().
FSEG_MAGIC_N_BYTES: Replaces FSEG_MAGIC_N_VALUE.
dict_hdr_get(), trx_rsegf_get_new(), trx_undo_page_get(),
trx_undo_page_get_s_latched(): Replaced with error-checking calls.
trx_rseg_t::get(mtr_t*): Replaces trx_rsegf_get().
trx_rseg_header_create(): Let the caller update the TRX_SYS page if needed.
trx_sys_create_sys_pages(): Merged with trx_sysf_create().
dict_check_tablespaces_and_store_max_id(): Do not access
DICT_HDR_MAX_SPACE_ID, because it was already recovered in dict_boot().
Merge dict_check_sys_tables() with this function.
dir_pathname(): Replaces os_file_make_new_pathname().
row_undo_ins_remove_sec(): Do not modify the undo page by adding
a terminating NUL byte to the record.
btr_decryption_failed(): Report decryption failures
dict_set_corrupted_by_space(), dict_set_encrypted_by_space(),
dict_set_corrupted_index_cache_only(): Remove.
dict_set_corrupted(): Remove the constant parameter dict_locked=false.
Never flag the clustered index corrupted in SYS_INDEXES, because
that would deny further access to the table. It might be possible to
repair the table by executing ALTER TABLE or OPTIMIZE TABLE, in case
no B-tree leaf page is corrupted.
dict_table_skip_corrupt_index(), dict_table_next_uncorrupted_index(),
row_purge_skip_uncommitted_virtual_index(): Remove, and refactor
the callers to read dict_index_t::type only once.
dict_table_is_corrupted(): Remove.
dict_index_t::is_btree(): Determine if the index is a valid B-tree.
BUF_GET_NO_LATCH, BUF_EVICT_IF_IN_POOL: Remove.
UNIV_BTR_DEBUG: Remove. Any inconsistency will no longer trigger
assertion failures, but error codes being returned.
buf_corrupt_page_release(): Replaced with a direct call to
buf_pool.corrupted_evict().
fil_invalid_page_access_msg(): Never crash on an invalid read;
let the caller of buf_page_get_gen() decide.
btr_pcur_t::restore_position(): Propagate failure status to the caller
by returning CORRUPTED.
opt_search_plan_for_table(): Simplify the code.
row_purge_del_mark(), row_purge_upd_exist_or_extern_func(),
row_undo_ins_remove_sec_rec(), row_undo_mod_upd_del_sec(),
row_undo_mod_del_mark_sec(): Avoid mem_heap_create()/mem_heap_free()
when no secondary indexes exist.
row_undo_mod_upd_exist_sec(): Simplify the code.
row_upd_clust_step(), dict_load_table_one(): Return DB_TABLE_CORRUPT
if the clustered index (and therefore the table) is corrupted, similar
to what we do in row_insert_for_mysql().
fut_get_ptr(): Replace with buf_page_get_gen() calls.
buf_page_get_gen(): Return nullptr and *err=DB_CORRUPTION
if the page is marked as freed. For other modes than
BUF_GET_POSSIBLY_FREED or BUF_PEEK_IF_IN_POOL this will
trigger a debug assertion failure. For BUF_GET_POSSIBLY_FREED,
we will return nullptr for freed pages, so that the callers
can be simplified. The purge of transaction history will be
a new user of BUF_GET_POSSIBLY_FREED, to avoid crashes on
corrupted data.
buf_page_get_low(): Never crash on a corrupted page, but simply
return nullptr.
fseg_page_is_allocated(): Replaces fseg_page_is_free().
fts_drop_common_tables(): Return an error if the transaction
was rolled back.
fil_space_t::set_corrupted(): Report a tablespace as corrupted if
it was not reported already.
fil_space_t::io(): Invoke fil_space_t::set_corrupted() to report
out-of-bounds page access or other errors.
Clean up mtr_t::page_lock()
buf_page_get_low(): Validate the page identifier (to check for
recently read corrupted pages) after acquiring the page latch.
buf_page_t::read_complete(): Flag uninitialized (all-zero) pages
with DB_FAIL. Return DB_PAGE_CORRUPTED on page number mismatch.
mtr_t::defer_drop_ahi(): Renamed from mtr_defer_drop_ahi().
recv_sys_t::free_corrupted_page(): Only set_corrupt_fs()
if any log records exist for the page. We do not mind if read-ahead
produces corrupted (or all-zero) pages that were not actually needed
during recovery.
recv_recover_page(): Return whether the operation succeeded.
recv_sys_t::recover_low(): Simplify the logic. Check for recovery error.
Thanks to Matthias Leich for testing this extensively and to the
authors of https://rr-project.org for making it easy to diagnose
and fix any failures that were found during the testing.
3 years ago MDEV-13542: Crashing on corrupted page is unhelpful
The approach to handling corruption that was chosen by Oracle in
commit 177d8b0c125b841c0650d27d735e3b87509dc286
is not really useful. Not only did it actually fail to prevent InnoDB
from crashing, but it is making things worse by blocking attempts to
rescue data from or rebuild a partially readable table.
We will try to prevent crashes in a different way: by propagating
errors up the call stack. We will never mark the clustered index
persistently corrupted, so that data recovery may be attempted by
reading from the table, or by rebuilding the table.
This should also fix MDEV-13680 (crash on btr_page_alloc() failure);
it was extensively tested with innodb_file_per_table=0 and a
non-autoextend system tablespace.
We should now avoid crashes in many cases, such as when a page
cannot be read or allocated, or an inconsistency is detected when
attempting to update multiple pages. We will not crash on double-free,
such as on the recovery of DDL in system tablespace in case something
was corrupted.
Crashes on corrupted data are still possible. The fault injection mechanism
that is introduced in the subsequent commit may help catch more of them.
buf_page_import_corrupt_failure: Remove the fault injection, and instead
corrupt some pages using Perl code in the tests.
btr_cur_pessimistic_insert(): Always reserve extents (except for the
change buffer), in order to prevent a subsequent allocation failure.
btr_pcur_open_at_rnd_pos(): Merged to the only caller ibuf_merge_pages().
btr_assert_not_corrupted(), btr_corruption_report(): Remove.
Similar checks are already part of btr_block_get().
FSEG_MAGIC_N_BYTES: Replaces FSEG_MAGIC_N_VALUE.
dict_hdr_get(), trx_rsegf_get_new(), trx_undo_page_get(),
trx_undo_page_get_s_latched(): Replaced with error-checking calls.
trx_rseg_t::get(mtr_t*): Replaces trx_rsegf_get().
trx_rseg_header_create(): Let the caller update the TRX_SYS page if needed.
trx_sys_create_sys_pages(): Merged with trx_sysf_create().
dict_check_tablespaces_and_store_max_id(): Do not access
DICT_HDR_MAX_SPACE_ID, because it was already recovered in dict_boot().
Merge dict_check_sys_tables() with this function.
dir_pathname(): Replaces os_file_make_new_pathname().
row_undo_ins_remove_sec(): Do not modify the undo page by adding
a terminating NUL byte to the record.
btr_decryption_failed(): Report decryption failures
dict_set_corrupted_by_space(), dict_set_encrypted_by_space(),
dict_set_corrupted_index_cache_only(): Remove.
dict_set_corrupted(): Remove the constant parameter dict_locked=false.
Never flag the clustered index corrupted in SYS_INDEXES, because
that would deny further access to the table. It might be possible to
repair the table by executing ALTER TABLE or OPTIMIZE TABLE, in case
no B-tree leaf page is corrupted.
dict_table_skip_corrupt_index(), dict_table_next_uncorrupted_index(),
row_purge_skip_uncommitted_virtual_index(): Remove, and refactor
the callers to read dict_index_t::type only once.
dict_table_is_corrupted(): Remove.
dict_index_t::is_btree(): Determine if the index is a valid B-tree.
BUF_GET_NO_LATCH, BUF_EVICT_IF_IN_POOL: Remove.
UNIV_BTR_DEBUG: Remove. Any inconsistency will no longer trigger
assertion failures, but error codes being returned.
buf_corrupt_page_release(): Replaced with a direct call to
buf_pool.corrupted_evict().
fil_invalid_page_access_msg(): Never crash on an invalid read;
let the caller of buf_page_get_gen() decide.
btr_pcur_t::restore_position(): Propagate failure status to the caller
by returning CORRUPTED.
opt_search_plan_for_table(): Simplify the code.
row_purge_del_mark(), row_purge_upd_exist_or_extern_func(),
row_undo_ins_remove_sec_rec(), row_undo_mod_upd_del_sec(),
row_undo_mod_del_mark_sec(): Avoid mem_heap_create()/mem_heap_free()
when no secondary indexes exist.
row_undo_mod_upd_exist_sec(): Simplify the code.
row_upd_clust_step(), dict_load_table_one(): Return DB_TABLE_CORRUPT
if the clustered index (and therefore the table) is corrupted, similar
to what we do in row_insert_for_mysql().
fut_get_ptr(): Replace with buf_page_get_gen() calls.
buf_page_get_gen(): Return nullptr and *err=DB_CORRUPTION
if the page is marked as freed. For other modes than
BUF_GET_POSSIBLY_FREED or BUF_PEEK_IF_IN_POOL this will
trigger a debug assertion failure. For BUF_GET_POSSIBLY_FREED,
we will return nullptr for freed pages, so that the callers
can be simplified. The purge of transaction history will be
a new user of BUF_GET_POSSIBLY_FREED, to avoid crashes on
corrupted data.
buf_page_get_low(): Never crash on a corrupted page, but simply
return nullptr.
fseg_page_is_allocated(): Replaces fseg_page_is_free().
fts_drop_common_tables(): Return an error if the transaction
was rolled back.
fil_space_t::set_corrupted(): Report a tablespace as corrupted if
it was not reported already.
fil_space_t::io(): Invoke fil_space_t::set_corrupted() to report
out-of-bounds page access or other errors.
Clean up mtr_t::page_lock()
buf_page_get_low(): Validate the page identifier (to check for
recently read corrupted pages) after acquiring the page latch.
buf_page_t::read_complete(): Flag uninitialized (all-zero) pages
with DB_FAIL. Return DB_PAGE_CORRUPTED on page number mismatch.
mtr_t::defer_drop_ahi(): Renamed from mtr_defer_drop_ahi().
recv_sys_t::free_corrupted_page(): Only set_corrupt_fs()
if any log records exist for the page. We do not mind if read-ahead
produces corrupted (or all-zero) pages that were not actually needed
during recovery.
recv_recover_page(): Return whether the operation succeeded.
recv_sys_t::recover_low(): Simplify the logic. Check for recovery error.
Thanks to Matthias Leich for testing this extensively and to the
authors of https://rr-project.org for making it easy to diagnose
and fix any failures that were found during the testing.
3 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-13542: Crashing on corrupted page is unhelpful
The approach to handling corruption that was chosen by Oracle in
commit 177d8b0c125b841c0650d27d735e3b87509dc286
is not really useful. Not only did it actually fail to prevent InnoDB
from crashing, but it is making things worse by blocking attempts to
rescue data from or rebuild a partially readable table.
We will try to prevent crashes in a different way: by propagating
errors up the call stack. We will never mark the clustered index
persistently corrupted, so that data recovery may be attempted by
reading from the table, or by rebuilding the table.
This should also fix MDEV-13680 (crash on btr_page_alloc() failure);
it was extensively tested with innodb_file_per_table=0 and a
non-autoextend system tablespace.
We should now avoid crashes in many cases, such as when a page
cannot be read or allocated, or an inconsistency is detected when
attempting to update multiple pages. We will not crash on double-free,
such as on the recovery of DDL in system tablespace in case something
was corrupted.
Crashes on corrupted data are still possible. The fault injection mechanism
that is introduced in the subsequent commit may help catch more of them.
buf_page_import_corrupt_failure: Remove the fault injection, and instead
corrupt some pages using Perl code in the tests.
btr_cur_pessimistic_insert(): Always reserve extents (except for the
change buffer), in order to prevent a subsequent allocation failure.
btr_pcur_open_at_rnd_pos(): Merged to the only caller ibuf_merge_pages().
btr_assert_not_corrupted(), btr_corruption_report(): Remove.
Similar checks are already part of btr_block_get().
FSEG_MAGIC_N_BYTES: Replaces FSEG_MAGIC_N_VALUE.
dict_hdr_get(), trx_rsegf_get_new(), trx_undo_page_get(),
trx_undo_page_get_s_latched(): Replaced with error-checking calls.
trx_rseg_t::get(mtr_t*): Replaces trx_rsegf_get().
trx_rseg_header_create(): Let the caller update the TRX_SYS page if needed.
trx_sys_create_sys_pages(): Merged with trx_sysf_create().
dict_check_tablespaces_and_store_max_id(): Do not access
DICT_HDR_MAX_SPACE_ID, because it was already recovered in dict_boot().
Merge dict_check_sys_tables() with this function.
dir_pathname(): Replaces os_file_make_new_pathname().
row_undo_ins_remove_sec(): Do not modify the undo page by adding
a terminating NUL byte to the record.
btr_decryption_failed(): Report decryption failures
dict_set_corrupted_by_space(), dict_set_encrypted_by_space(),
dict_set_corrupted_index_cache_only(): Remove.
dict_set_corrupted(): Remove the constant parameter dict_locked=false.
Never flag the clustered index corrupted in SYS_INDEXES, because
that would deny further access to the table. It might be possible to
repair the table by executing ALTER TABLE or OPTIMIZE TABLE, in case
no B-tree leaf page is corrupted.
dict_table_skip_corrupt_index(), dict_table_next_uncorrupted_index(),
row_purge_skip_uncommitted_virtual_index(): Remove, and refactor
the callers to read dict_index_t::type only once.
dict_table_is_corrupted(): Remove.
dict_index_t::is_btree(): Determine if the index is a valid B-tree.
BUF_GET_NO_LATCH, BUF_EVICT_IF_IN_POOL: Remove.
UNIV_BTR_DEBUG: Remove. Any inconsistency will no longer trigger
assertion failures, but error codes being returned.
buf_corrupt_page_release(): Replaced with a direct call to
buf_pool.corrupted_evict().
fil_invalid_page_access_msg(): Never crash on an invalid read;
let the caller of buf_page_get_gen() decide.
btr_pcur_t::restore_position(): Propagate failure status to the caller
by returning CORRUPTED.
opt_search_plan_for_table(): Simplify the code.
row_purge_del_mark(), row_purge_upd_exist_or_extern_func(),
row_undo_ins_remove_sec_rec(), row_undo_mod_upd_del_sec(),
row_undo_mod_del_mark_sec(): Avoid mem_heap_create()/mem_heap_free()
when no secondary indexes exist.
row_undo_mod_upd_exist_sec(): Simplify the code.
row_upd_clust_step(), dict_load_table_one(): Return DB_TABLE_CORRUPT
if the clustered index (and therefore the table) is corrupted, similar
to what we do in row_insert_for_mysql().
fut_get_ptr(): Replace with buf_page_get_gen() calls.
buf_page_get_gen(): Return nullptr and *err=DB_CORRUPTION
if the page is marked as freed. For other modes than
BUF_GET_POSSIBLY_FREED or BUF_PEEK_IF_IN_POOL this will
trigger a debug assertion failure. For BUF_GET_POSSIBLY_FREED,
we will return nullptr for freed pages, so that the callers
can be simplified. The purge of transaction history will be
a new user of BUF_GET_POSSIBLY_FREED, to avoid crashes on
corrupted data.
buf_page_get_low(): Never crash on a corrupted page, but simply
return nullptr.
fseg_page_is_allocated(): Replaces fseg_page_is_free().
fts_drop_common_tables(): Return an error if the transaction
was rolled back.
fil_space_t::set_corrupted(): Report a tablespace as corrupted if
it was not reported already.
fil_space_t::io(): Invoke fil_space_t::set_corrupted() to report
out-of-bounds page access or other errors.
Clean up mtr_t::page_lock()
buf_page_get_low(): Validate the page identifier (to check for
recently read corrupted pages) after acquiring the page latch.
buf_page_t::read_complete(): Flag uninitialized (all-zero) pages
with DB_FAIL. Return DB_PAGE_CORRUPTED on page number mismatch.
mtr_t::defer_drop_ahi(): Renamed from mtr_defer_drop_ahi().
recv_sys_t::free_corrupted_page(): Only set_corrupt_fs()
if any log records exist for the page. We do not mind if read-ahead
produces corrupted (or all-zero) pages that were not actually needed
during recovery.
recv_recover_page(): Return whether the operation succeeded.
recv_sys_t::recover_low(): Simplify the logic. Check for recovery error.
Thanks to Matthias Leich for testing this extensively and to the
authors of https://rr-project.org for making it easy to diagnose
and fix any failures that were found during the testing.
3 years ago MDEV-13542: Crashing on corrupted page is unhelpful
The approach to handling corruption that was chosen by Oracle in
commit 177d8b0c125b841c0650d27d735e3b87509dc286
is not really useful. Not only did it actually fail to prevent InnoDB
from crashing, but it is making things worse by blocking attempts to
rescue data from or rebuild a partially readable table.
We will try to prevent crashes in a different way: by propagating
errors up the call stack. We will never mark the clustered index
persistently corrupted, so that data recovery may be attempted by
reading from the table, or by rebuilding the table.
This should also fix MDEV-13680 (crash on btr_page_alloc() failure);
it was extensively tested with innodb_file_per_table=0 and a
non-autoextend system tablespace.
We should now avoid crashes in many cases, such as when a page
cannot be read or allocated, or an inconsistency is detected when
attempting to update multiple pages. We will not crash on double-free,
such as on the recovery of DDL in system tablespace in case something
was corrupted.
Crashes on corrupted data are still possible. The fault injection mechanism
that is introduced in the subsequent commit may help catch more of them.
buf_page_import_corrupt_failure: Remove the fault injection, and instead
corrupt some pages using Perl code in the tests.
btr_cur_pessimistic_insert(): Always reserve extents (except for the
change buffer), in order to prevent a subsequent allocation failure.
btr_pcur_open_at_rnd_pos(): Merged to the only caller ibuf_merge_pages().
btr_assert_not_corrupted(), btr_corruption_report(): Remove.
Similar checks are already part of btr_block_get().
FSEG_MAGIC_N_BYTES: Replaces FSEG_MAGIC_N_VALUE.
dict_hdr_get(), trx_rsegf_get_new(), trx_undo_page_get(),
trx_undo_page_get_s_latched(): Replaced with error-checking calls.
trx_rseg_t::get(mtr_t*): Replaces trx_rsegf_get().
trx_rseg_header_create(): Let the caller update the TRX_SYS page if needed.
trx_sys_create_sys_pages(): Merged with trx_sysf_create().
dict_check_tablespaces_and_store_max_id(): Do not access
DICT_HDR_MAX_SPACE_ID, because it was already recovered in dict_boot().
Merge dict_check_sys_tables() with this function.
dir_pathname(): Replaces os_file_make_new_pathname().
row_undo_ins_remove_sec(): Do not modify the undo page by adding
a terminating NUL byte to the record.
btr_decryption_failed(): Report decryption failures
dict_set_corrupted_by_space(), dict_set_encrypted_by_space(),
dict_set_corrupted_index_cache_only(): Remove.
dict_set_corrupted(): Remove the constant parameter dict_locked=false.
Never flag the clustered index corrupted in SYS_INDEXES, because
that would deny further access to the table. It might be possible to
repair the table by executing ALTER TABLE or OPTIMIZE TABLE, in case
no B-tree leaf page is corrupted.
dict_table_skip_corrupt_index(), dict_table_next_uncorrupted_index(),
row_purge_skip_uncommitted_virtual_index(): Remove, and refactor
the callers to read dict_index_t::type only once.
dict_table_is_corrupted(): Remove.
dict_index_t::is_btree(): Determine if the index is a valid B-tree.
BUF_GET_NO_LATCH, BUF_EVICT_IF_IN_POOL: Remove.
UNIV_BTR_DEBUG: Remove. Any inconsistency will no longer trigger
assertion failures, but error codes being returned.
buf_corrupt_page_release(): Replaced with a direct call to
buf_pool.corrupted_evict().
fil_invalid_page_access_msg(): Never crash on an invalid read;
let the caller of buf_page_get_gen() decide.
btr_pcur_t::restore_position(): Propagate failure status to the caller
by returning CORRUPTED.
opt_search_plan_for_table(): Simplify the code.
row_purge_del_mark(), row_purge_upd_exist_or_extern_func(),
row_undo_ins_remove_sec_rec(), row_undo_mod_upd_del_sec(),
row_undo_mod_del_mark_sec(): Avoid mem_heap_create()/mem_heap_free()
when no secondary indexes exist.
row_undo_mod_upd_exist_sec(): Simplify the code.
row_upd_clust_step(), dict_load_table_one(): Return DB_TABLE_CORRUPT
if the clustered index (and therefore the table) is corrupted, similar
to what we do in row_insert_for_mysql().
fut_get_ptr(): Replace with buf_page_get_gen() calls.
buf_page_get_gen(): Return nullptr and *err=DB_CORRUPTION
if the page is marked as freed. For other modes than
BUF_GET_POSSIBLY_FREED or BUF_PEEK_IF_IN_POOL this will
trigger a debug assertion failure. For BUF_GET_POSSIBLY_FREED,
we will return nullptr for freed pages, so that the callers
can be simplified. The purge of transaction history will be
a new user of BUF_GET_POSSIBLY_FREED, to avoid crashes on
corrupted data.
buf_page_get_low(): Never crash on a corrupted page, but simply
return nullptr.
fseg_page_is_allocated(): Replaces fseg_page_is_free().
fts_drop_common_tables(): Return an error if the transaction
was rolled back.
fil_space_t::set_corrupted(): Report a tablespace as corrupted if
it was not reported already.
fil_space_t::io(): Invoke fil_space_t::set_corrupted() to report
out-of-bounds page access or other errors.
Clean up mtr_t::page_lock()
buf_page_get_low(): Validate the page identifier (to check for
recently read corrupted pages) after acquiring the page latch.
buf_page_t::read_complete(): Flag uninitialized (all-zero) pages
with DB_FAIL. Return DB_PAGE_CORRUPTED on page number mismatch.
mtr_t::defer_drop_ahi(): Renamed from mtr_defer_drop_ahi().
recv_sys_t::free_corrupted_page(): Only set_corrupt_fs()
if any log records exist for the page. We do not mind if read-ahead
produces corrupted (or all-zero) pages that were not actually needed
during recovery.
recv_recover_page(): Return whether the operation succeeded.
recv_sys_t::recover_low(): Simplify the logic. Check for recovery error.
Thanks to Matthias Leich for testing this extensively and to the
authors of https://rr-project.org for making it easy to diagnose
and fix any failures that were found during the testing.
3 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-13542: Crashing on corrupted page is unhelpful
The approach to handling corruption that was chosen by Oracle in
commit 177d8b0c125b841c0650d27d735e3b87509dc286
is not really useful. Not only did it actually fail to prevent InnoDB
from crashing, but it is making things worse by blocking attempts to
rescue data from or rebuild a partially readable table.
We will try to prevent crashes in a different way: by propagating
errors up the call stack. We will never mark the clustered index
persistently corrupted, so that data recovery may be attempted by
reading from the table, or by rebuilding the table.
This should also fix MDEV-13680 (crash on btr_page_alloc() failure);
it was extensively tested with innodb_file_per_table=0 and a
non-autoextend system tablespace.
We should now avoid crashes in many cases, such as when a page
cannot be read or allocated, or an inconsistency is detected when
attempting to update multiple pages. We will not crash on double-free,
such as on the recovery of DDL in system tablespace in case something
was corrupted.
Crashes on corrupted data are still possible. The fault injection mechanism
that is introduced in the subsequent commit may help catch more of them.
buf_page_import_corrupt_failure: Remove the fault injection, and instead
corrupt some pages using Perl code in the tests.
btr_cur_pessimistic_insert(): Always reserve extents (except for the
change buffer), in order to prevent a subsequent allocation failure.
btr_pcur_open_at_rnd_pos(): Merged to the only caller ibuf_merge_pages().
btr_assert_not_corrupted(), btr_corruption_report(): Remove.
Similar checks are already part of btr_block_get().
FSEG_MAGIC_N_BYTES: Replaces FSEG_MAGIC_N_VALUE.
dict_hdr_get(), trx_rsegf_get_new(), trx_undo_page_get(),
trx_undo_page_get_s_latched(): Replaced with error-checking calls.
trx_rseg_t::get(mtr_t*): Replaces trx_rsegf_get().
trx_rseg_header_create(): Let the caller update the TRX_SYS page if needed.
trx_sys_create_sys_pages(): Merged with trx_sysf_create().
dict_check_tablespaces_and_store_max_id(): Do not access
DICT_HDR_MAX_SPACE_ID, because it was already recovered in dict_boot().
Merge dict_check_sys_tables() with this function.
dir_pathname(): Replaces os_file_make_new_pathname().
row_undo_ins_remove_sec(): Do not modify the undo page by adding
a terminating NUL byte to the record.
btr_decryption_failed(): Report decryption failures
dict_set_corrupted_by_space(), dict_set_encrypted_by_space(),
dict_set_corrupted_index_cache_only(): Remove.
dict_set_corrupted(): Remove the constant parameter dict_locked=false.
Never flag the clustered index corrupted in SYS_INDEXES, because
that would deny further access to the table. It might be possible to
repair the table by executing ALTER TABLE or OPTIMIZE TABLE, in case
no B-tree leaf page is corrupted.
dict_table_skip_corrupt_index(), dict_table_next_uncorrupted_index(),
row_purge_skip_uncommitted_virtual_index(): Remove, and refactor
the callers to read dict_index_t::type only once.
dict_table_is_corrupted(): Remove.
dict_index_t::is_btree(): Determine if the index is a valid B-tree.
BUF_GET_NO_LATCH, BUF_EVICT_IF_IN_POOL: Remove.
UNIV_BTR_DEBUG: Remove. Any inconsistency will no longer trigger
assertion failures, but error codes being returned.
buf_corrupt_page_release(): Replaced with a direct call to
buf_pool.corrupted_evict().
fil_invalid_page_access_msg(): Never crash on an invalid read;
let the caller of buf_page_get_gen() decide.
btr_pcur_t::restore_position(): Propagate failure status to the caller
by returning CORRUPTED.
opt_search_plan_for_table(): Simplify the code.
row_purge_del_mark(), row_purge_upd_exist_or_extern_func(),
row_undo_ins_remove_sec_rec(), row_undo_mod_upd_del_sec(),
row_undo_mod_del_mark_sec(): Avoid mem_heap_create()/mem_heap_free()
when no secondary indexes exist.
row_undo_mod_upd_exist_sec(): Simplify the code.
row_upd_clust_step(), dict_load_table_one(): Return DB_TABLE_CORRUPT
if the clustered index (and therefore the table) is corrupted, similar
to what we do in row_insert_for_mysql().
fut_get_ptr(): Replace with buf_page_get_gen() calls.
buf_page_get_gen(): Return nullptr and *err=DB_CORRUPTION
if the page is marked as freed. For other modes than
BUF_GET_POSSIBLY_FREED or BUF_PEEK_IF_IN_POOL this will
trigger a debug assertion failure. For BUF_GET_POSSIBLY_FREED,
we will return nullptr for freed pages, so that the callers
can be simplified. The purge of transaction history will be
a new user of BUF_GET_POSSIBLY_FREED, to avoid crashes on
corrupted data.
buf_page_get_low(): Never crash on a corrupted page, but simply
return nullptr.
fseg_page_is_allocated(): Replaces fseg_page_is_free().
fts_drop_common_tables(): Return an error if the transaction
was rolled back.
fil_space_t::set_corrupted(): Report a tablespace as corrupted if
it was not reported already.
fil_space_t::io(): Invoke fil_space_t::set_corrupted() to report
out-of-bounds page access or other errors.
Clean up mtr_t::page_lock()
buf_page_get_low(): Validate the page identifier (to check for
recently read corrupted pages) after acquiring the page latch.
buf_page_t::read_complete(): Flag uninitialized (all-zero) pages
with DB_FAIL. Return DB_PAGE_CORRUPTED on page number mismatch.
mtr_t::defer_drop_ahi(): Renamed from mtr_defer_drop_ahi().
recv_sys_t::free_corrupted_page(): Only set_corrupt_fs()
if any log records exist for the page. We do not mind if read-ahead
produces corrupted (or all-zero) pages that were not actually needed
during recovery.
recv_recover_page(): Return whether the operation succeeded.
recv_sys_t::recover_low(): Simplify the logic. Check for recovery error.
Thanks to Matthias Leich for testing this extensively and to the
authors of https://rr-project.org for making it easy to diagnose
and fix any failures that were found during the testing.
3 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-13542: Crashing on corrupted page is unhelpful
The approach to handling corruption that was chosen by Oracle in
commit 177d8b0c125b841c0650d27d735e3b87509dc286
is not really useful. Not only did it actually fail to prevent InnoDB
from crashing, but it is making things worse by blocking attempts to
rescue data from or rebuild a partially readable table.
We will try to prevent crashes in a different way: by propagating
errors up the call stack. We will never mark the clustered index
persistently corrupted, so that data recovery may be attempted by
reading from the table, or by rebuilding the table.
This should also fix MDEV-13680 (crash on btr_page_alloc() failure);
it was extensively tested with innodb_file_per_table=0 and a
non-autoextend system tablespace.
We should now avoid crashes in many cases, such as when a page
cannot be read or allocated, or an inconsistency is detected when
attempting to update multiple pages. We will not crash on double-free,
such as on the recovery of DDL in system tablespace in case something
was corrupted.
Crashes on corrupted data are still possible. The fault injection mechanism
that is introduced in the subsequent commit may help catch more of them.
buf_page_import_corrupt_failure: Remove the fault injection, and instead
corrupt some pages using Perl code in the tests.
btr_cur_pessimistic_insert(): Always reserve extents (except for the
change buffer), in order to prevent a subsequent allocation failure.
btr_pcur_open_at_rnd_pos(): Merged to the only caller ibuf_merge_pages().
btr_assert_not_corrupted(), btr_corruption_report(): Remove.
Similar checks are already part of btr_block_get().
FSEG_MAGIC_N_BYTES: Replaces FSEG_MAGIC_N_VALUE.
dict_hdr_get(), trx_rsegf_get_new(), trx_undo_page_get(),
trx_undo_page_get_s_latched(): Replaced with error-checking calls.
trx_rseg_t::get(mtr_t*): Replaces trx_rsegf_get().
trx_rseg_header_create(): Let the caller update the TRX_SYS page if needed.
trx_sys_create_sys_pages(): Merged with trx_sysf_create().
dict_check_tablespaces_and_store_max_id(): Do not access
DICT_HDR_MAX_SPACE_ID, because it was already recovered in dict_boot().
Merge dict_check_sys_tables() with this function.
dir_pathname(): Replaces os_file_make_new_pathname().
row_undo_ins_remove_sec(): Do not modify the undo page by adding
a terminating NUL byte to the record.
btr_decryption_failed(): Report decryption failures
dict_set_corrupted_by_space(), dict_set_encrypted_by_space(),
dict_set_corrupted_index_cache_only(): Remove.
dict_set_corrupted(): Remove the constant parameter dict_locked=false.
Never flag the clustered index corrupted in SYS_INDEXES, because
that would deny further access to the table. It might be possible to
repair the table by executing ALTER TABLE or OPTIMIZE TABLE, in case
no B-tree leaf page is corrupted.
dict_table_skip_corrupt_index(), dict_table_next_uncorrupted_index(),
row_purge_skip_uncommitted_virtual_index(): Remove, and refactor
the callers to read dict_index_t::type only once.
dict_table_is_corrupted(): Remove.
dict_index_t::is_btree(): Determine if the index is a valid B-tree.
BUF_GET_NO_LATCH, BUF_EVICT_IF_IN_POOL: Remove.
UNIV_BTR_DEBUG: Remove. Any inconsistency will no longer trigger
assertion failures, but error codes being returned.
buf_corrupt_page_release(): Replaced with a direct call to
buf_pool.corrupted_evict().
fil_invalid_page_access_msg(): Never crash on an invalid read;
let the caller of buf_page_get_gen() decide.
btr_pcur_t::restore_position(): Propagate failure status to the caller
by returning CORRUPTED.
opt_search_plan_for_table(): Simplify the code.
row_purge_del_mark(), row_purge_upd_exist_or_extern_func(),
row_undo_ins_remove_sec_rec(), row_undo_mod_upd_del_sec(),
row_undo_mod_del_mark_sec(): Avoid mem_heap_create()/mem_heap_free()
when no secondary indexes exist.
row_undo_mod_upd_exist_sec(): Simplify the code.
row_upd_clust_step(), dict_load_table_one(): Return DB_TABLE_CORRUPT
if the clustered index (and therefore the table) is corrupted, similar
to what we do in row_insert_for_mysql().
fut_get_ptr(): Replace with buf_page_get_gen() calls.
buf_page_get_gen(): Return nullptr and *err=DB_CORRUPTION
if the page is marked as freed. For other modes than
BUF_GET_POSSIBLY_FREED or BUF_PEEK_IF_IN_POOL this will
trigger a debug assertion failure. For BUF_GET_POSSIBLY_FREED,
we will return nullptr for freed pages, so that the callers
can be simplified. The purge of transaction history will be
a new user of BUF_GET_POSSIBLY_FREED, to avoid crashes on
corrupted data.
buf_page_get_low(): Never crash on a corrupted page, but simply
return nullptr.
fseg_page_is_allocated(): Replaces fseg_page_is_free().
fts_drop_common_tables(): Return an error if the transaction
was rolled back.
fil_space_t::set_corrupted(): Report a tablespace as corrupted if
it was not reported already.
fil_space_t::io(): Invoke fil_space_t::set_corrupted() to report
out-of-bounds page access or other errors.
Clean up mtr_t::page_lock()
buf_page_get_low(): Validate the page identifier (to check for
recently read corrupted pages) after acquiring the page latch.
buf_page_t::read_complete(): Flag uninitialized (all-zero) pages
with DB_FAIL. Return DB_PAGE_CORRUPTED on page number mismatch.
mtr_t::defer_drop_ahi(): Renamed from mtr_defer_drop_ahi().
recv_sys_t::free_corrupted_page(): Only set_corrupt_fs()
if any log records exist for the page. We do not mind if read-ahead
produces corrupted (or all-zero) pages that were not actually needed
during recovery.
recv_recover_page(): Return whether the operation succeeded.
recv_sys_t::recover_low(): Simplify the logic. Check for recovery error.
Thanks to Matthias Leich for testing this extensively and to the
authors of https://rr-project.org for making it easy to diagnose
and fix any failures that were found during the testing.
3 years ago MDEV-13542: Crashing on corrupted page is unhelpful
The approach to handling corruption that was chosen by Oracle in
commit 177d8b0c125b841c0650d27d735e3b87509dc286
is not really useful. Not only did it actually fail to prevent InnoDB
from crashing, but it is making things worse by blocking attempts to
rescue data from or rebuild a partially readable table.
We will try to prevent crashes in a different way: by propagating
errors up the call stack. We will never mark the clustered index
persistently corrupted, so that data recovery may be attempted by
reading from the table, or by rebuilding the table.
This should also fix MDEV-13680 (crash on btr_page_alloc() failure);
it was extensively tested with innodb_file_per_table=0 and a
non-autoextend system tablespace.
We should now avoid crashes in many cases, such as when a page
cannot be read or allocated, or an inconsistency is detected when
attempting to update multiple pages. We will not crash on double-free,
such as on the recovery of DDL in system tablespace in case something
was corrupted.
Crashes on corrupted data are still possible. The fault injection mechanism
that is introduced in the subsequent commit may help catch more of them.
buf_page_import_corrupt_failure: Remove the fault injection, and instead
corrupt some pages using Perl code in the tests.
btr_cur_pessimistic_insert(): Always reserve extents (except for the
change buffer), in order to prevent a subsequent allocation failure.
btr_pcur_open_at_rnd_pos(): Merged to the only caller ibuf_merge_pages().
btr_assert_not_corrupted(), btr_corruption_report(): Remove.
Similar checks are already part of btr_block_get().
FSEG_MAGIC_N_BYTES: Replaces FSEG_MAGIC_N_VALUE.
dict_hdr_get(), trx_rsegf_get_new(), trx_undo_page_get(),
trx_undo_page_get_s_latched(): Replaced with error-checking calls.
trx_rseg_t::get(mtr_t*): Replaces trx_rsegf_get().
trx_rseg_header_create(): Let the caller update the TRX_SYS page if needed.
trx_sys_create_sys_pages(): Merged with trx_sysf_create().
dict_check_tablespaces_and_store_max_id(): Do not access
DICT_HDR_MAX_SPACE_ID, because it was already recovered in dict_boot().
Merge dict_check_sys_tables() with this function.
dir_pathname(): Replaces os_file_make_new_pathname().
row_undo_ins_remove_sec(): Do not modify the undo page by adding
a terminating NUL byte to the record.
btr_decryption_failed(): Report decryption failures
dict_set_corrupted_by_space(), dict_set_encrypted_by_space(),
dict_set_corrupted_index_cache_only(): Remove.
dict_set_corrupted(): Remove the constant parameter dict_locked=false.
Never flag the clustered index corrupted in SYS_INDEXES, because
that would deny further access to the table. It might be possible to
repair the table by executing ALTER TABLE or OPTIMIZE TABLE, in case
no B-tree leaf page is corrupted.
dict_table_skip_corrupt_index(), dict_table_next_uncorrupted_index(),
row_purge_skip_uncommitted_virtual_index(): Remove, and refactor
the callers to read dict_index_t::type only once.
dict_table_is_corrupted(): Remove.
dict_index_t::is_btree(): Determine if the index is a valid B-tree.
BUF_GET_NO_LATCH, BUF_EVICT_IF_IN_POOL: Remove.
UNIV_BTR_DEBUG: Remove. Any inconsistency will no longer trigger
assertion failures, but error codes being returned.
buf_corrupt_page_release(): Replaced with a direct call to
buf_pool.corrupted_evict().
fil_invalid_page_access_msg(): Never crash on an invalid read;
let the caller of buf_page_get_gen() decide.
btr_pcur_t::restore_position(): Propagate failure status to the caller
by returning CORRUPTED.
opt_search_plan_for_table(): Simplify the code.
row_purge_del_mark(), row_purge_upd_exist_or_extern_func(),
row_undo_ins_remove_sec_rec(), row_undo_mod_upd_del_sec(),
row_undo_mod_del_mark_sec(): Avoid mem_heap_create()/mem_heap_free()
when no secondary indexes exist.
row_undo_mod_upd_exist_sec(): Simplify the code.
row_upd_clust_step(), dict_load_table_one(): Return DB_TABLE_CORRUPT
if the clustered index (and therefore the table) is corrupted, similar
to what we do in row_insert_for_mysql().
fut_get_ptr(): Replace with buf_page_get_gen() calls.
buf_page_get_gen(): Return nullptr and *err=DB_CORRUPTION
if the page is marked as freed. For other modes than
BUF_GET_POSSIBLY_FREED or BUF_PEEK_IF_IN_POOL this will
trigger a debug assertion failure. For BUF_GET_POSSIBLY_FREED,
we will return nullptr for freed pages, so that the callers
can be simplified. The purge of transaction history will be
a new user of BUF_GET_POSSIBLY_FREED, to avoid crashes on
corrupted data.
buf_page_get_low(): Never crash on a corrupted page, but simply
return nullptr.
fseg_page_is_allocated(): Replaces fseg_page_is_free().
fts_drop_common_tables(): Return an error if the transaction
was rolled back.
fil_space_t::set_corrupted(): Report a tablespace as corrupted if
it was not reported already.
fil_space_t::io(): Invoke fil_space_t::set_corrupted() to report
out-of-bounds page access or other errors.
Clean up mtr_t::page_lock()
buf_page_get_low(): Validate the page identifier (to check for
recently read corrupted pages) after acquiring the page latch.
buf_page_t::read_complete(): Flag uninitialized (all-zero) pages
with DB_FAIL. Return DB_PAGE_CORRUPTED on page number mismatch.
mtr_t::defer_drop_ahi(): Renamed from mtr_defer_drop_ahi().
recv_sys_t::free_corrupted_page(): Only set_corrupt_fs()
if any log records exist for the page. We do not mind if read-ahead
produces corrupted (or all-zero) pages that were not actually needed
during recovery.
recv_recover_page(): Return whether the operation succeeded.
recv_sys_t::recover_low(): Simplify the logic. Check for recovery error.
Thanks to Matthias Leich for testing this extensively and to the
authors of https://rr-project.org for making it easy to diagnose
and fix any failures that were found during the testing.
3 years ago MDEV-13542: Crashing on corrupted page is unhelpful
The approach to handling corruption that was chosen by Oracle in
commit 177d8b0c125b841c0650d27d735e3b87509dc286
is not really useful. Not only did it actually fail to prevent InnoDB
from crashing, but it is making things worse by blocking attempts to
rescue data from or rebuild a partially readable table.
We will try to prevent crashes in a different way: by propagating
errors up the call stack. We will never mark the clustered index
persistently corrupted, so that data recovery may be attempted by
reading from the table, or by rebuilding the table.
This should also fix MDEV-13680 (crash on btr_page_alloc() failure);
it was extensively tested with innodb_file_per_table=0 and a
non-autoextend system tablespace.
We should now avoid crashes in many cases, such as when a page
cannot be read or allocated, or an inconsistency is detected when
attempting to update multiple pages. We will not crash on double-free,
such as on the recovery of DDL in system tablespace in case something
was corrupted.
Crashes on corrupted data are still possible. The fault injection mechanism
that is introduced in the subsequent commit may help catch more of them.
buf_page_import_corrupt_failure: Remove the fault injection, and instead
corrupt some pages using Perl code in the tests.
btr_cur_pessimistic_insert(): Always reserve extents (except for the
change buffer), in order to prevent a subsequent allocation failure.
btr_pcur_open_at_rnd_pos(): Merged to the only caller ibuf_merge_pages().
btr_assert_not_corrupted(), btr_corruption_report(): Remove.
Similar checks are already part of btr_block_get().
FSEG_MAGIC_N_BYTES: Replaces FSEG_MAGIC_N_VALUE.
dict_hdr_get(), trx_rsegf_get_new(), trx_undo_page_get(),
trx_undo_page_get_s_latched(): Replaced with error-checking calls.
trx_rseg_t::get(mtr_t*): Replaces trx_rsegf_get().
trx_rseg_header_create(): Let the caller update the TRX_SYS page if needed.
trx_sys_create_sys_pages(): Merged with trx_sysf_create().
dict_check_tablespaces_and_store_max_id(): Do not access
DICT_HDR_MAX_SPACE_ID, because it was already recovered in dict_boot().
Merge dict_check_sys_tables() with this function.
dir_pathname(): Replaces os_file_make_new_pathname().
row_undo_ins_remove_sec(): Do not modify the undo page by adding
a terminating NUL byte to the record.
btr_decryption_failed(): Report decryption failures
dict_set_corrupted_by_space(), dict_set_encrypted_by_space(),
dict_set_corrupted_index_cache_only(): Remove.
dict_set_corrupted(): Remove the constant parameter dict_locked=false.
Never flag the clustered index corrupted in SYS_INDEXES, because
that would deny further access to the table. It might be possible to
repair the table by executing ALTER TABLE or OPTIMIZE TABLE, in case
no B-tree leaf page is corrupted.
dict_table_skip_corrupt_index(), dict_table_next_uncorrupted_index(),
row_purge_skip_uncommitted_virtual_index(): Remove, and refactor
the callers to read dict_index_t::type only once.
dict_table_is_corrupted(): Remove.
dict_index_t::is_btree(): Determine if the index is a valid B-tree.
BUF_GET_NO_LATCH, BUF_EVICT_IF_IN_POOL: Remove.
UNIV_BTR_DEBUG: Remove. Any inconsistency will no longer trigger
assertion failures, but error codes being returned.
buf_corrupt_page_release(): Replaced with a direct call to
buf_pool.corrupted_evict().
fil_invalid_page_access_msg(): Never crash on an invalid read;
let the caller of buf_page_get_gen() decide.
btr_pcur_t::restore_position(): Propagate failure status to the caller
by returning CORRUPTED.
opt_search_plan_for_table(): Simplify the code.
row_purge_del_mark(), row_purge_upd_exist_or_extern_func(),
row_undo_ins_remove_sec_rec(), row_undo_mod_upd_del_sec(),
row_undo_mod_del_mark_sec(): Avoid mem_heap_create()/mem_heap_free()
when no secondary indexes exist.
row_undo_mod_upd_exist_sec(): Simplify the code.
row_upd_clust_step(), dict_load_table_one(): Return DB_TABLE_CORRUPT
if the clustered index (and therefore the table) is corrupted, similar
to what we do in row_insert_for_mysql().
fut_get_ptr(): Replace with buf_page_get_gen() calls.
buf_page_get_gen(): Return nullptr and *err=DB_CORRUPTION
if the page is marked as freed. For other modes than
BUF_GET_POSSIBLY_FREED or BUF_PEEK_IF_IN_POOL this will
trigger a debug assertion failure. For BUF_GET_POSSIBLY_FREED,
we will return nullptr for freed pages, so that the callers
can be simplified. The purge of transaction history will be
a new user of BUF_GET_POSSIBLY_FREED, to avoid crashes on
corrupted data.
buf_page_get_low(): Never crash on a corrupted page, but simply
return nullptr.
fseg_page_is_allocated(): Replaces fseg_page_is_free().
fts_drop_common_tables(): Return an error if the transaction
was rolled back.
fil_space_t::set_corrupted(): Report a tablespace as corrupted if
it was not reported already.
fil_space_t::io(): Invoke fil_space_t::set_corrupted() to report
out-of-bounds page access or other errors.
Clean up mtr_t::page_lock()
buf_page_get_low(): Validate the page identifier (to check for
recently read corrupted pages) after acquiring the page latch.
buf_page_t::read_complete(): Flag uninitialized (all-zero) pages
with DB_FAIL. Return DB_PAGE_CORRUPTED on page number mismatch.
mtr_t::defer_drop_ahi(): Renamed from mtr_defer_drop_ahi().
recv_sys_t::free_corrupted_page(): Only set_corrupt_fs()
if any log records exist for the page. We do not mind if read-ahead
produces corrupted (or all-zero) pages that were not actually needed
during recovery.
recv_recover_page(): Return whether the operation succeeded.
recv_sys_t::recover_low(): Simplify the logic. Check for recovery error.
Thanks to Matthias Leich for testing this extensively and to the
authors of https://rr-project.org for making it easy to diagnose
and fix any failures that were found during the testing.
3 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-13542: Crashing on corrupted page is unhelpful
The approach to handling corruption that was chosen by Oracle in
commit 177d8b0c125b841c0650d27d735e3b87509dc286
is not really useful. Not only did it actually fail to prevent InnoDB
from crashing, but it is making things worse by blocking attempts to
rescue data from or rebuild a partially readable table.
We will try to prevent crashes in a different way: by propagating
errors up the call stack. We will never mark the clustered index
persistently corrupted, so that data recovery may be attempted by
reading from the table, or by rebuilding the table.
This should also fix MDEV-13680 (crash on btr_page_alloc() failure);
it was extensively tested with innodb_file_per_table=0 and a
non-autoextend system tablespace.
We should now avoid crashes in many cases, such as when a page
cannot be read or allocated, or an inconsistency is detected when
attempting to update multiple pages. We will not crash on double-free,
such as on the recovery of DDL in system tablespace in case something
was corrupted.
Crashes on corrupted data are still possible. The fault injection mechanism
that is introduced in the subsequent commit may help catch more of them.
buf_page_import_corrupt_failure: Remove the fault injection, and instead
corrupt some pages using Perl code in the tests.
btr_cur_pessimistic_insert(): Always reserve extents (except for the
change buffer), in order to prevent a subsequent allocation failure.
btr_pcur_open_at_rnd_pos(): Merged to the only caller ibuf_merge_pages().
btr_assert_not_corrupted(), btr_corruption_report(): Remove.
Similar checks are already part of btr_block_get().
FSEG_MAGIC_N_BYTES: Replaces FSEG_MAGIC_N_VALUE.
dict_hdr_get(), trx_rsegf_get_new(), trx_undo_page_get(),
trx_undo_page_get_s_latched(): Replaced with error-checking calls.
trx_rseg_t::get(mtr_t*): Replaces trx_rsegf_get().
trx_rseg_header_create(): Let the caller update the TRX_SYS page if needed.
trx_sys_create_sys_pages(): Merged with trx_sysf_create().
dict_check_tablespaces_and_store_max_id(): Do not access
DICT_HDR_MAX_SPACE_ID, because it was already recovered in dict_boot().
Merge dict_check_sys_tables() with this function.
dir_pathname(): Replaces os_file_make_new_pathname().
row_undo_ins_remove_sec(): Do not modify the undo page by adding
a terminating NUL byte to the record.
btr_decryption_failed(): Report decryption failures
dict_set_corrupted_by_space(), dict_set_encrypted_by_space(),
dict_set_corrupted_index_cache_only(): Remove.
dict_set_corrupted(): Remove the constant parameter dict_locked=false.
Never flag the clustered index corrupted in SYS_INDEXES, because
that would deny further access to the table. It might be possible to
repair the table by executing ALTER TABLE or OPTIMIZE TABLE, in case
no B-tree leaf page is corrupted.
dict_table_skip_corrupt_index(), dict_table_next_uncorrupted_index(),
row_purge_skip_uncommitted_virtual_index(): Remove, and refactor
the callers to read dict_index_t::type only once.
dict_table_is_corrupted(): Remove.
dict_index_t::is_btree(): Determine if the index is a valid B-tree.
BUF_GET_NO_LATCH, BUF_EVICT_IF_IN_POOL: Remove.
UNIV_BTR_DEBUG: Remove. Any inconsistency will no longer trigger
assertion failures, but error codes being returned.
buf_corrupt_page_release(): Replaced with a direct call to
buf_pool.corrupted_evict().
fil_invalid_page_access_msg(): Never crash on an invalid read;
let the caller of buf_page_get_gen() decide.
btr_pcur_t::restore_position(): Propagate failure status to the caller
by returning CORRUPTED.
opt_search_plan_for_table(): Simplify the code.
row_purge_del_mark(), row_purge_upd_exist_or_extern_func(),
row_undo_ins_remove_sec_rec(), row_undo_mod_upd_del_sec(),
row_undo_mod_del_mark_sec(): Avoid mem_heap_create()/mem_heap_free()
when no secondary indexes exist.
row_undo_mod_upd_exist_sec(): Simplify the code.
row_upd_clust_step(), dict_load_table_one(): Return DB_TABLE_CORRUPT
if the clustered index (and therefore the table) is corrupted, similar
to what we do in row_insert_for_mysql().
fut_get_ptr(): Replace with buf_page_get_gen() calls.
buf_page_get_gen(): Return nullptr and *err=DB_CORRUPTION
if the page is marked as freed. For other modes than
BUF_GET_POSSIBLY_FREED or BUF_PEEK_IF_IN_POOL this will
trigger a debug assertion failure. For BUF_GET_POSSIBLY_FREED,
we will return nullptr for freed pages, so that the callers
can be simplified. The purge of transaction history will be
a new user of BUF_GET_POSSIBLY_FREED, to avoid crashes on
corrupted data.
buf_page_get_low(): Never crash on a corrupted page, but simply
return nullptr.
fseg_page_is_allocated(): Replaces fseg_page_is_free().
fts_drop_common_tables(): Return an error if the transaction
was rolled back.
fil_space_t::set_corrupted(): Report a tablespace as corrupted if
it was not reported already.
fil_space_t::io(): Invoke fil_space_t::set_corrupted() to report
out-of-bounds page access or other errors.
Clean up mtr_t::page_lock()
buf_page_get_low(): Validate the page identifier (to check for
recently read corrupted pages) after acquiring the page latch.
buf_page_t::read_complete(): Flag uninitialized (all-zero) pages
with DB_FAIL. Return DB_PAGE_CORRUPTED on page number mismatch.
mtr_t::defer_drop_ahi(): Renamed from mtr_defer_drop_ahi().
recv_sys_t::free_corrupted_page(): Only set_corrupt_fs()
if any log records exist for the page. We do not mind if read-ahead
produces corrupted (or all-zero) pages that were not actually needed
during recovery.
recv_recover_page(): Return whether the operation succeeded.
recv_sys_t::recover_low(): Simplify the logic. Check for recovery error.
Thanks to Matthias Leich for testing this extensively and to the
authors of https://rr-project.org for making it easy to diagnose
and fix any failures that were found during the testing.
3 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-23399: Performance regression with write workloads
The buffer pool refactoring in MDEV-15053 and MDEV-22871 shifted
the performance bottleneck to the page flushing.
The configuration parameters will be changed as follows:
innodb_lru_flush_size=32 (new: how many pages to flush on LRU eviction)
innodb_lru_scan_depth=1536 (old: 1024)
innodb_max_dirty_pages_pct=90 (old: 75)
innodb_max_dirty_pages_pct_lwm=75 (old: 0)
Note: The parameter innodb_lru_scan_depth will only affect LRU
eviction of buffer pool pages when a new page is being allocated. The
page cleaner thread will no longer evict any pages. It used to
guarantee that some pages will remain free in the buffer pool. Now, we
perform that eviction 'on demand' in buf_LRU_get_free_block().
The parameter innodb_lru_scan_depth(srv_LRU_scan_depth) is used as follows:
* When the buffer pool is being shrunk in buf_pool_t::withdraw_blocks()
* As a buf_pool.free limit in buf_LRU_list_batch() for terminating
the flushing that is initiated e.g., by buf_LRU_get_free_block()
The parameter also used to serve as an initial limit for unzip_LRU
eviction (evicting uncompressed page frames while retaining
ROW_FORMAT=COMPRESSED pages), but now we will use a hard-coded limit
of 100 or unlimited for invoking buf_LRU_scan_and_free_block().
The status variables will be changed as follows:
innodb_buffer_pool_pages_flushed: This includes also the count of
innodb_buffer_pool_pages_LRU_flushed and should work reliably,
updated one by one in buf_flush_page() to give more real-time
statistics. The function buf_flush_stats(), which we are removing,
was not called in every code path. For both counters, we will use
regular variables that are incremented in a critical section of
buf_pool.mutex. Note that show_innodb_vars() directly links to the
variables, and reads of the counters will *not* be protected by
buf_pool.mutex, so you cannot get a consistent snapshot of both variables.
The following INFORMATION_SCHEMA.INNODB_METRICS counters will be
removed, because the page cleaner no longer deals with writing or
evicting least recently used pages, and because the single-page writes
have been removed:
* buffer_LRU_batch_flush_avg_time_slot
* buffer_LRU_batch_flush_avg_time_thread
* buffer_LRU_batch_flush_avg_time_est
* buffer_LRU_batch_flush_avg_pass
* buffer_LRU_single_flush_scanned
* buffer_LRU_single_flush_num_scan
* buffer_LRU_single_flush_scanned_per_call
When moving to a single buffer pool instance in MDEV-15058, we missed
some opportunity to simplify the buf_flush_page_cleaner thread. It was
unnecessarily using a mutex and some complex data structures, even
though we always have a single page cleaner thread.
Furthermore, the buf_flush_page_cleaner thread had separate 'recovery'
and 'shutdown' modes where it was waiting to be triggered by some
other thread, adding unnecessary latency and potential for hangs in
relatively rarely executed startup or shutdown code.
The page cleaner was also running two kinds of batches in an
interleaved fashion: "LRU flush" (writing out some least recently used
pages and evicting them on write completion) and the normal batches
that aim to increase the MIN(oldest_modification) in the buffer pool,
to help the log checkpoint advance.
The buf_pool.flush_list flushing was being blocked by
buf_block_t::lock for no good reason. Furthermore, if the FIL_PAGE_LSN
of a page is ahead of log_sys.get_flushed_lsn(), that is, what has
been persistently written to the redo log, we would trigger a log
flush and then resume the page flushing. This would unnecessarily
limit the performance of the page cleaner thread and trigger the
infamous messages "InnoDB: page_cleaner: 1000ms intended loop took 4450ms.
The settings might not be optimal" that were suppressed in
commit d1ab89037a518fcffbc50c24e4bd94e4ec33aed0 unless log_warnings>2.
Our revised algorithm will make log_sys.get_flushed_lsn() advance at
the start of buf_flush_lists(), and then execute a 'best effort' to
write out all pages. The flush batches will skip pages that were modified
since the log was written, or are are currently exclusively locked.
The MDEV-13670 message "page_cleaner: 1000ms intended loop took" message
will be removed, because by design, the buf_flush_page_cleaner() should
not be blocked during a batch for extended periods of time.
We will remove the single-page flushing altogether. Related to this,
the debug parameter innodb_doublewrite_batch_size will be removed,
because all of the doublewrite buffer will be used for flushing
batches. If a page needs to be evicted from the buffer pool and all
100 least recently used pages in the buffer pool have unflushed
changes, buf_LRU_get_free_block() will execute buf_flush_lists() to
write out and evict innodb_lru_flush_size pages. At most one thread
will execute buf_flush_lists() in buf_LRU_get_free_block(); other
threads will wait for that LRU flushing batch to finish.
To improve concurrency, we will replace the InnoDB ib_mutex_t and
os_event_t native mutexes and condition variables in this area of code.
Most notably, this means that the buffer pool mutex (buf_pool.mutex)
is no longer instrumented via any InnoDB interfaces. It will continue
to be instrumented via PERFORMANCE_SCHEMA.
For now, both buf_pool.flush_list_mutex and buf_pool.mutex will be
declared with MY_MUTEX_INIT_FAST (PTHREAD_MUTEX_ADAPTIVE_NP). The critical
sections of buf_pool.flush_list_mutex should be shorter than those for
buf_pool.mutex, because in the worst case, they cover a linear scan of
buf_pool.flush_list, while the worst case of a critical section of
buf_pool.mutex covers a linear scan of the potentially much longer
buf_pool.LRU list.
mysql_mutex_is_owner(), safe_mutex_is_owner(): New predicate, usable
with SAFE_MUTEX. Some InnoDB debug assertions need this predicate
instead of mysql_mutex_assert_owner() or mysql_mutex_assert_not_owner().
buf_pool_t::n_flush_LRU, buf_pool_t::n_flush_list:
Replaces buf_pool_t::init_flush[] and buf_pool_t::n_flush[].
The number of active flush operations.
buf_pool_t::mutex, buf_pool_t::flush_list_mutex: Use mysql_mutex_t
instead of ib_mutex_t, to have native mutexes with PERFORMANCE_SCHEMA
and SAFE_MUTEX instrumentation.
buf_pool_t::done_flush_LRU: Condition variable for !n_flush_LRU.
buf_pool_t::done_flush_list: Condition variable for !n_flush_list.
buf_pool_t::do_flush_list: Condition variable to wake up the
buf_flush_page_cleaner when a log checkpoint needs to be written
or the server is being shut down. Replaces buf_flush_event.
We will keep using timed waits (the page cleaner thread will wake
_at least_ once per second), because the calculations for
innodb_adaptive_flushing depend on fixed time intervals.
buf_dblwr: Allocate statically, and move all code to member functions.
Use a native mutex and condition variable. Remove code to deal with
single-page flushing.
buf_dblwr_check_block(): Make the check debug-only. We were spending
a significant amount of execution time in page_simple_validate_new().
flush_counters_t::unzip_LRU_evicted: Remove.
IORequest: Make more members const. FIXME: m_fil_node should be removed.
buf_flush_sync_lsn: Protect by std::atomic, not page_cleaner.mutex
(which we are removing).
page_cleaner_slot_t, page_cleaner_t: Remove many redundant members.
pc_request_flush_slot(): Replaces pc_request() and pc_flush_slot().
recv_writer_thread: Remove. Recovery works just fine without it, if we
simply invoke buf_flush_sync() at the end of each batch in
recv_sys_t::apply().
recv_recovery_from_checkpoint_finish(): Remove. We can simply call
recv_sys.debug_free() directly.
srv_started_redo: Replaces srv_start_state.
SRV_SHUTDOWN_FLUSH_PHASE: Remove. logs_empty_and_mark_files_at_shutdown()
can communicate with the normal page cleaner loop via the new function
flush_buffer_pool().
buf_flush_remove(): Assert that the calling thread is holding
buf_pool.flush_list_mutex. This removes unnecessary mutex operations
from buf_flush_remove_pages() and buf_flush_dirty_pages(),
which replace buf_LRU_flush_or_remove_pages().
buf_flush_lists(): Renamed from buf_flush_batch(), with simplified
interface. Return the number of flushed pages. Clarified comments and
renamed min_n to max_n. Identify LRU batch by lsn=0. Merge all the functions
buf_flush_start(), buf_flush_batch(), buf_flush_end() directly to this
function, which was their only caller, and remove 2 unnecessary
buf_pool.mutex release/re-acquisition that we used to perform around
the buf_flush_batch() call. At the start, if not all log has been
durably written, wait for a background task to do it, or start a new
task to do it. This allows the log write to run concurrently with our
page flushing batch. Any pages that were skipped due to too recent
FIL_PAGE_LSN or due to them being latched by a writer should be flushed
during the next batch, unless there are further modifications to those
pages. It is possible that a page that we must flush due to small
oldest_modification also carries a recent FIL_PAGE_LSN or is being
constantly modified. In the worst case, all writers would then end up
waiting in log_free_check() to allow the flushing and the checkpoint
to complete.
buf_do_flush_list_batch(): Clarify comments, and rename min_n to max_n.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_flush_space(): Auxiliary function to look up a tablespace for
page flushing.
buf_flush_page(): Defer the computation of space->full_crc32(). Never
call log_write_up_to(), but instead skip persistent pages whose latest
modification (FIL_PAGE_LSN) is newer than the redo log. Also skip
pages on which we cannot acquire a shared latch without waiting.
buf_flush_try_neighbors(): Do not bother checking buf_fix_count
because buf_flush_page() will no longer wait for the page latch.
Take the tablespace as a parameter, and only execute this function
when innodb_flush_neighbors>0. Avoid repeated calls of page_id_t::fold().
buf_flush_relocate_on_flush_list(): Declare as cold, and push down
a condition from the callers.
buf_flush_check_neighbor(): Take id.fold() as a parameter.
buf_flush_sync(): Ensure that the buf_pool.flush_list is empty,
because the flushing batch will skip pages whose modifications have
not yet been written to the log or were latched for modification.
buf_free_from_unzip_LRU_list_batch(): Remove redundant local variables.
buf_flush_LRU_list_batch(): Let the caller buf_do_LRU_batch() initialize
the counters, and report n->evicted.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_do_LRU_batch(): Return the number of pages flushed.
buf_LRU_free_page(): Only release and re-acquire buf_pool.mutex if
adaptive hash index entries are pointing to the block.
buf_LRU_get_free_block(): Do not wake up the page cleaner, because it
will no longer perform any useful work for us, and we do not want it
to compete for I/O while buf_flush_lists(innodb_lru_flush_size, 0)
writes out and evicts at most innodb_lru_flush_size pages. (The
function buf_do_LRU_batch() may complete after writing fewer pages if
more than innodb_lru_scan_depth pages end up in buf_pool.free list.)
Eliminate some mutex release-acquire cycles, and wait for the LRU
flush batch to complete before rescanning.
buf_LRU_check_size_of_non_data_objects(): Simplify the code.
buf_page_write_complete(): Remove the parameter evict, and always
evict pages that were part of an LRU flush.
buf_page_create(): Take a pre-allocated page as a parameter.
buf_pool_t::free_block(): Free a pre-allocated block.
recv_sys_t::recover_low(), recv_sys_t::apply(): Preallocate the block
while not holding recv_sys.mutex. During page allocation, we may
initiate a page flush, which in turn may initiate a log flush, which
would require acquiring log_sys.mutex, which should always be acquired
before recv_sys.mutex in order to avoid deadlocks. Therefore, we must
not be holding recv_sys.mutex while allocating a buffer pool block.
BtrBulk::logFreeCheck(): Skip a redundant condition.
row_undo_step(): Do not invoke srv_inc_activity_count() for every row
that is being rolled back. It should suffice to invoke the function in
trx_flush_log_if_needed() during trx_t::commit_in_memory() when the
rollback completes.
sync_check_enable(): Remove. We will enable innodb_sync_debug from the
very beginning.
Reviewed by: Vladislav Vaintroub
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-22456 Dropping the adaptive hash index may cause DDL to lock up InnoDB
If the InnoDB buffer pool contains many pages for a table or index
that is being dropped or rebuilt, and if many of such pages are
pointed to by the adaptive hash index, dropping the adaptive hash index
may consume a lot of time.
The time-consuming operation of dropping the adaptive hash index entries
is being executed while the InnoDB data dictionary cache dict_sys is
exclusively locked.
It is not actually necessary to drop all adaptive hash index entries
at the time a table or index is being dropped or rebuilt. We can let
the LRU replacement policy of the buffer pool take care of this gradually.
For this to work, we must detach the dict_table_t and dict_index_t
objects from the main dict_sys cache, and once the last
adaptive hash index entry for the detached table is removed
(when the garbage page is evicted from the buffer pool) we can free
the dict_table_t and dict_index_t object.
Related to this, in MDEV-16283, we made ALTER TABLE...DISCARD TABLESPACE
skip both the buffer pool eviction and the drop of the adaptive hash index.
We shifted the burden to ALTER TABLE...IMPORT TABLESPACE or DROP TABLE.
We can remove the eviction from DROP TABLE. We must retain the eviction
in the ALTER TABLE...IMPORT TABLESPACE code path, so that in case the
discarded table is being re-imported with the same tablespace identifier,
the fresh data from the imported tablespace will replace any stale pages
in the buffer pool.
rpl.rpl_failed_drop_tbl_binlog: Remove the test. DROP TABLE can
no longer be interrupted inside InnoDB.
fseg_free_page(), fseg_free_step(), fseg_free_step_not_header(),
fseg_free_page_low(), fseg_free_extent(): Remove the parameter
that specifies whether the adaptive hash index should be dropped.
btr_search_lazy_free(): Lazily free an index when the last
reference to it is dropped from the adaptive hash index.
buf_pool_clear_hash_index(): Declare static, and move to the
same compilation unit with the bulk of the adaptive hash index
code.
dict_index_t::clone(), dict_index_t::clone_if_needed():
Clone an index that is being rebuilt while adaptive hash index
entries exist. The original index will be inserted into
dict_table_t::freed_indexes and dict_index_t::set_freed()
will be called.
dict_index_t::set_freed(), dict_index_t::freed(): Note that
or check whether the index has been freed. We will use the
impossible page number 1 to denote this condition.
dict_index_t::n_ahi_pages(): Replaces btr_search_info_get_ref_count().
dict_index_t::detach_columns(): Move the assignment n_fields=0
to ha_innobase_inplace_ctx::clear_added_indexes().
We must have access to the columns when freeing the
adaptive hash index. Note: dict_table_t::v_cols[] will remain
valid. If virtual columns are dropped or added, the table
definition will be reloaded in ha_innobase::commit_inplace_alter_table().
buf_page_mtr_lock(): Drop a stale adaptive hash index if needed.
We will also reduce the number of btr_get_search_latch() calls
and enclose some more code inside #ifdef BTR_CUR_HASH_ADAPT
in order to benefit cmake -DWITH_INNODB_AHI=OFF.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-13542: Crashing on corrupted page is unhelpful
The approach to handling corruption that was chosen by Oracle in
commit 177d8b0c125b841c0650d27d735e3b87509dc286
is not really useful. Not only did it actually fail to prevent InnoDB
from crashing, but it is making things worse by blocking attempts to
rescue data from or rebuild a partially readable table.
We will try to prevent crashes in a different way: by propagating
errors up the call stack. We will never mark the clustered index
persistently corrupted, so that data recovery may be attempted by
reading from the table, or by rebuilding the table.
This should also fix MDEV-13680 (crash on btr_page_alloc() failure);
it was extensively tested with innodb_file_per_table=0 and a
non-autoextend system tablespace.
We should now avoid crashes in many cases, such as when a page
cannot be read or allocated, or an inconsistency is detected when
attempting to update multiple pages. We will not crash on double-free,
such as on the recovery of DDL in system tablespace in case something
was corrupted.
Crashes on corrupted data are still possible. The fault injection mechanism
that is introduced in the subsequent commit may help catch more of them.
buf_page_import_corrupt_failure: Remove the fault injection, and instead
corrupt some pages using Perl code in the tests.
btr_cur_pessimistic_insert(): Always reserve extents (except for the
change buffer), in order to prevent a subsequent allocation failure.
btr_pcur_open_at_rnd_pos(): Merged to the only caller ibuf_merge_pages().
btr_assert_not_corrupted(), btr_corruption_report(): Remove.
Similar checks are already part of btr_block_get().
FSEG_MAGIC_N_BYTES: Replaces FSEG_MAGIC_N_VALUE.
dict_hdr_get(), trx_rsegf_get_new(), trx_undo_page_get(),
trx_undo_page_get_s_latched(): Replaced with error-checking calls.
trx_rseg_t::get(mtr_t*): Replaces trx_rsegf_get().
trx_rseg_header_create(): Let the caller update the TRX_SYS page if needed.
trx_sys_create_sys_pages(): Merged with trx_sysf_create().
dict_check_tablespaces_and_store_max_id(): Do not access
DICT_HDR_MAX_SPACE_ID, because it was already recovered in dict_boot().
Merge dict_check_sys_tables() with this function.
dir_pathname(): Replaces os_file_make_new_pathname().
row_undo_ins_remove_sec(): Do not modify the undo page by adding
a terminating NUL byte to the record.
btr_decryption_failed(): Report decryption failures
dict_set_corrupted_by_space(), dict_set_encrypted_by_space(),
dict_set_corrupted_index_cache_only(): Remove.
dict_set_corrupted(): Remove the constant parameter dict_locked=false.
Never flag the clustered index corrupted in SYS_INDEXES, because
that would deny further access to the table. It might be possible to
repair the table by executing ALTER TABLE or OPTIMIZE TABLE, in case
no B-tree leaf page is corrupted.
dict_table_skip_corrupt_index(), dict_table_next_uncorrupted_index(),
row_purge_skip_uncommitted_virtual_index(): Remove, and refactor
the callers to read dict_index_t::type only once.
dict_table_is_corrupted(): Remove.
dict_index_t::is_btree(): Determine if the index is a valid B-tree.
BUF_GET_NO_LATCH, BUF_EVICT_IF_IN_POOL: Remove.
UNIV_BTR_DEBUG: Remove. Any inconsistency will no longer trigger
assertion failures, but error codes being returned.
buf_corrupt_page_release(): Replaced with a direct call to
buf_pool.corrupted_evict().
fil_invalid_page_access_msg(): Never crash on an invalid read;
let the caller of buf_page_get_gen() decide.
btr_pcur_t::restore_position(): Propagate failure status to the caller
by returning CORRUPTED.
opt_search_plan_for_table(): Simplify the code.
row_purge_del_mark(), row_purge_upd_exist_or_extern_func(),
row_undo_ins_remove_sec_rec(), row_undo_mod_upd_del_sec(),
row_undo_mod_del_mark_sec(): Avoid mem_heap_create()/mem_heap_free()
when no secondary indexes exist.
row_undo_mod_upd_exist_sec(): Simplify the code.
row_upd_clust_step(), dict_load_table_one(): Return DB_TABLE_CORRUPT
if the clustered index (and therefore the table) is corrupted, similar
to what we do in row_insert_for_mysql().
fut_get_ptr(): Replace with buf_page_get_gen() calls.
buf_page_get_gen(): Return nullptr and *err=DB_CORRUPTION
if the page is marked as freed. For other modes than
BUF_GET_POSSIBLY_FREED or BUF_PEEK_IF_IN_POOL this will
trigger a debug assertion failure. For BUF_GET_POSSIBLY_FREED,
we will return nullptr for freed pages, so that the callers
can be simplified. The purge of transaction history will be
a new user of BUF_GET_POSSIBLY_FREED, to avoid crashes on
corrupted data.
buf_page_get_low(): Never crash on a corrupted page, but simply
return nullptr.
fseg_page_is_allocated(): Replaces fseg_page_is_free().
fts_drop_common_tables(): Return an error if the transaction
was rolled back.
fil_space_t::set_corrupted(): Report a tablespace as corrupted if
it was not reported already.
fil_space_t::io(): Invoke fil_space_t::set_corrupted() to report
out-of-bounds page access or other errors.
Clean up mtr_t::page_lock()
buf_page_get_low(): Validate the page identifier (to check for
recently read corrupted pages) after acquiring the page latch.
buf_page_t::read_complete(): Flag uninitialized (all-zero) pages
with DB_FAIL. Return DB_PAGE_CORRUPTED on page number mismatch.
mtr_t::defer_drop_ahi(): Renamed from mtr_defer_drop_ahi().
recv_sys_t::free_corrupted_page(): Only set_corrupt_fs()
if any log records exist for the page. We do not mind if read-ahead
produces corrupted (or all-zero) pages that were not actually needed
during recovery.
recv_recover_page(): Return whether the operation succeeded.
recv_sys_t::recover_low(): Simplify the logic. Check for recovery error.
Thanks to Matthias Leich for testing this extensively and to the
authors of https://rr-project.org for making it easy to diagnose
and fix any failures that were found during the testing.
3 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-13542: Crashing on corrupted page is unhelpful
The approach to handling corruption that was chosen by Oracle in
commit 177d8b0c125b841c0650d27d735e3b87509dc286
is not really useful. Not only did it actually fail to prevent InnoDB
from crashing, but it is making things worse by blocking attempts to
rescue data from or rebuild a partially readable table.
We will try to prevent crashes in a different way: by propagating
errors up the call stack. We will never mark the clustered index
persistently corrupted, so that data recovery may be attempted by
reading from the table, or by rebuilding the table.
This should also fix MDEV-13680 (crash on btr_page_alloc() failure);
it was extensively tested with innodb_file_per_table=0 and a
non-autoextend system tablespace.
We should now avoid crashes in many cases, such as when a page
cannot be read or allocated, or an inconsistency is detected when
attempting to update multiple pages. We will not crash on double-free,
such as on the recovery of DDL in system tablespace in case something
was corrupted.
Crashes on corrupted data are still possible. The fault injection mechanism
that is introduced in the subsequent commit may help catch more of them.
buf_page_import_corrupt_failure: Remove the fault injection, and instead
corrupt some pages using Perl code in the tests.
btr_cur_pessimistic_insert(): Always reserve extents (except for the
change buffer), in order to prevent a subsequent allocation failure.
btr_pcur_open_at_rnd_pos(): Merged to the only caller ibuf_merge_pages().
btr_assert_not_corrupted(), btr_corruption_report(): Remove.
Similar checks are already part of btr_block_get().
FSEG_MAGIC_N_BYTES: Replaces FSEG_MAGIC_N_VALUE.
dict_hdr_get(), trx_rsegf_get_new(), trx_undo_page_get(),
trx_undo_page_get_s_latched(): Replaced with error-checking calls.
trx_rseg_t::get(mtr_t*): Replaces trx_rsegf_get().
trx_rseg_header_create(): Let the caller update the TRX_SYS page if needed.
trx_sys_create_sys_pages(): Merged with trx_sysf_create().
dict_check_tablespaces_and_store_max_id(): Do not access
DICT_HDR_MAX_SPACE_ID, because it was already recovered in dict_boot().
Merge dict_check_sys_tables() with this function.
dir_pathname(): Replaces os_file_make_new_pathname().
row_undo_ins_remove_sec(): Do not modify the undo page by adding
a terminating NUL byte to the record.
btr_decryption_failed(): Report decryption failures
dict_set_corrupted_by_space(), dict_set_encrypted_by_space(),
dict_set_corrupted_index_cache_only(): Remove.
dict_set_corrupted(): Remove the constant parameter dict_locked=false.
Never flag the clustered index corrupted in SYS_INDEXES, because
that would deny further access to the table. It might be possible to
repair the table by executing ALTER TABLE or OPTIMIZE TABLE, in case
no B-tree leaf page is corrupted.
dict_table_skip_corrupt_index(), dict_table_next_uncorrupted_index(),
row_purge_skip_uncommitted_virtual_index(): Remove, and refactor
the callers to read dict_index_t::type only once.
dict_table_is_corrupted(): Remove.
dict_index_t::is_btree(): Determine if the index is a valid B-tree.
BUF_GET_NO_LATCH, BUF_EVICT_IF_IN_POOL: Remove.
UNIV_BTR_DEBUG: Remove. Any inconsistency will no longer trigger
assertion failures, but error codes being returned.
buf_corrupt_page_release(): Replaced with a direct call to
buf_pool.corrupted_evict().
fil_invalid_page_access_msg(): Never crash on an invalid read;
let the caller of buf_page_get_gen() decide.
btr_pcur_t::restore_position(): Propagate failure status to the caller
by returning CORRUPTED.
opt_search_plan_for_table(): Simplify the code.
row_purge_del_mark(), row_purge_upd_exist_or_extern_func(),
row_undo_ins_remove_sec_rec(), row_undo_mod_upd_del_sec(),
row_undo_mod_del_mark_sec(): Avoid mem_heap_create()/mem_heap_free()
when no secondary indexes exist.
row_undo_mod_upd_exist_sec(): Simplify the code.
row_upd_clust_step(), dict_load_table_one(): Return DB_TABLE_CORRUPT
if the clustered index (and therefore the table) is corrupted, similar
to what we do in row_insert_for_mysql().
fut_get_ptr(): Replace with buf_page_get_gen() calls.
buf_page_get_gen(): Return nullptr and *err=DB_CORRUPTION
if the page is marked as freed. For other modes than
BUF_GET_POSSIBLY_FREED or BUF_PEEK_IF_IN_POOL this will
trigger a debug assertion failure. For BUF_GET_POSSIBLY_FREED,
we will return nullptr for freed pages, so that the callers
can be simplified. The purge of transaction history will be
a new user of BUF_GET_POSSIBLY_FREED, to avoid crashes on
corrupted data.
buf_page_get_low(): Never crash on a corrupted page, but simply
return nullptr.
fseg_page_is_allocated(): Replaces fseg_page_is_free().
fts_drop_common_tables(): Return an error if the transaction
was rolled back.
fil_space_t::set_corrupted(): Report a tablespace as corrupted if
it was not reported already.
fil_space_t::io(): Invoke fil_space_t::set_corrupted() to report
out-of-bounds page access or other errors.
Clean up mtr_t::page_lock()
buf_page_get_low(): Validate the page identifier (to check for
recently read corrupted pages) after acquiring the page latch.
buf_page_t::read_complete(): Flag uninitialized (all-zero) pages
with DB_FAIL. Return DB_PAGE_CORRUPTED on page number mismatch.
mtr_t::defer_drop_ahi(): Renamed from mtr_defer_drop_ahi().
recv_sys_t::free_corrupted_page(): Only set_corrupt_fs()
if any log records exist for the page. We do not mind if read-ahead
produces corrupted (or all-zero) pages that were not actually needed
during recovery.
recv_recover_page(): Return whether the operation succeeded.
recv_sys_t::recover_low(): Simplify the logic. Check for recovery error.
Thanks to Matthias Leich for testing this extensively and to the
authors of https://rr-project.org for making it easy to diagnose
and fix any failures that were found during the testing.
3 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-22456 Dropping the adaptive hash index may cause DDL to lock up InnoDB
If the InnoDB buffer pool contains many pages for a table or index
that is being dropped or rebuilt, and if many of such pages are
pointed to by the adaptive hash index, dropping the adaptive hash index
may consume a lot of time.
The time-consuming operation of dropping the adaptive hash index entries
is being executed while the InnoDB data dictionary cache dict_sys is
exclusively locked.
It is not actually necessary to drop all adaptive hash index entries
at the time a table or index is being dropped or rebuilt. We can let
the LRU replacement policy of the buffer pool take care of this gradually.
For this to work, we must detach the dict_table_t and dict_index_t
objects from the main dict_sys cache, and once the last
adaptive hash index entry for the detached table is removed
(when the garbage page is evicted from the buffer pool) we can free
the dict_table_t and dict_index_t object.
Related to this, in MDEV-16283, we made ALTER TABLE...DISCARD TABLESPACE
skip both the buffer pool eviction and the drop of the adaptive hash index.
We shifted the burden to ALTER TABLE...IMPORT TABLESPACE or DROP TABLE.
We can remove the eviction from DROP TABLE. We must retain the eviction
in the ALTER TABLE...IMPORT TABLESPACE code path, so that in case the
discarded table is being re-imported with the same tablespace identifier,
the fresh data from the imported tablespace will replace any stale pages
in the buffer pool.
rpl.rpl_failed_drop_tbl_binlog: Remove the test. DROP TABLE can
no longer be interrupted inside InnoDB.
fseg_free_page(), fseg_free_step(), fseg_free_step_not_header(),
fseg_free_page_low(), fseg_free_extent(): Remove the parameter
that specifies whether the adaptive hash index should be dropped.
btr_search_lazy_free(): Lazily free an index when the last
reference to it is dropped from the adaptive hash index.
buf_pool_clear_hash_index(): Declare static, and move to the
same compilation unit with the bulk of the adaptive hash index
code.
dict_index_t::clone(), dict_index_t::clone_if_needed():
Clone an index that is being rebuilt while adaptive hash index
entries exist. The original index will be inserted into
dict_table_t::freed_indexes and dict_index_t::set_freed()
will be called.
dict_index_t::set_freed(), dict_index_t::freed(): Note that
or check whether the index has been freed. We will use the
impossible page number 1 to denote this condition.
dict_index_t::n_ahi_pages(): Replaces btr_search_info_get_ref_count().
dict_index_t::detach_columns(): Move the assignment n_fields=0
to ha_innobase_inplace_ctx::clear_added_indexes().
We must have access to the columns when freeing the
adaptive hash index. Note: dict_table_t::v_cols[] will remain
valid. If virtual columns are dropped or added, the table
definition will be reloaded in ha_innobase::commit_inplace_alter_table().
buf_page_mtr_lock(): Drop a stale adaptive hash index if needed.
We will also reduce the number of btr_get_search_latch() calls
and enclose some more code inside #ifdef BTR_CUR_HASH_ADAPT
in order to benefit cmake -DWITH_INNODB_AHI=OFF.
5 years ago MDEV-22456 Dropping the adaptive hash index may cause DDL to lock up InnoDB
If the InnoDB buffer pool contains many pages for a table or index
that is being dropped or rebuilt, and if many of such pages are
pointed to by the adaptive hash index, dropping the adaptive hash index
may consume a lot of time.
The time-consuming operation of dropping the adaptive hash index entries
is being executed while the InnoDB data dictionary cache dict_sys is
exclusively locked.
It is not actually necessary to drop all adaptive hash index entries
at the time a table or index is being dropped or rebuilt. We can let
the LRU replacement policy of the buffer pool take care of this gradually.
For this to work, we must detach the dict_table_t and dict_index_t
objects from the main dict_sys cache, and once the last
adaptive hash index entry for the detached table is removed
(when the garbage page is evicted from the buffer pool) we can free
the dict_table_t and dict_index_t object.
Related to this, in MDEV-16283, we made ALTER TABLE...DISCARD TABLESPACE
skip both the buffer pool eviction and the drop of the adaptive hash index.
We shifted the burden to ALTER TABLE...IMPORT TABLESPACE or DROP TABLE.
We can remove the eviction from DROP TABLE. We must retain the eviction
in the ALTER TABLE...IMPORT TABLESPACE code path, so that in case the
discarded table is being re-imported with the same tablespace identifier,
the fresh data from the imported tablespace will replace any stale pages
in the buffer pool.
rpl.rpl_failed_drop_tbl_binlog: Remove the test. DROP TABLE can
no longer be interrupted inside InnoDB.
fseg_free_page(), fseg_free_step(), fseg_free_step_not_header(),
fseg_free_page_low(), fseg_free_extent(): Remove the parameter
that specifies whether the adaptive hash index should be dropped.
btr_search_lazy_free(): Lazily free an index when the last
reference to it is dropped from the adaptive hash index.
buf_pool_clear_hash_index(): Declare static, and move to the
same compilation unit with the bulk of the adaptive hash index
code.
dict_index_t::clone(), dict_index_t::clone_if_needed():
Clone an index that is being rebuilt while adaptive hash index
entries exist. The original index will be inserted into
dict_table_t::freed_indexes and dict_index_t::set_freed()
will be called.
dict_index_t::set_freed(), dict_index_t::freed(): Note that
or check whether the index has been freed. We will use the
impossible page number 1 to denote this condition.
dict_index_t::n_ahi_pages(): Replaces btr_search_info_get_ref_count().
dict_index_t::detach_columns(): Move the assignment n_fields=0
to ha_innobase_inplace_ctx::clear_added_indexes().
We must have access to the columns when freeing the
adaptive hash index. Note: dict_table_t::v_cols[] will remain
valid. If virtual columns are dropped or added, the table
definition will be reloaded in ha_innobase::commit_inplace_alter_table().
buf_page_mtr_lock(): Drop a stale adaptive hash index if needed.
We will also reduce the number of btr_get_search_latch() calls
and enclose some more code inside #ifdef BTR_CUR_HASH_ADAPT
in order to benefit cmake -DWITH_INNODB_AHI=OFF.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-26827 Make page flushing even faster
For more convenient monitoring of something that could greatly affect
the volume of page writes, we add the status variable
Innodb_buffer_pool_pages_split that was previously only available
via information_schema.innodb_metrics as "innodb_page_splits".
This was suggested by Axel Schwenke.
buf_flush_page_count: Replaced with buf_pool.stat.n_pages_written.
We protect buf_pool.stat (except n_page_gets) with buf_pool.mutex
and remove unnecessary export_vars indirection.
buf_pool.flush_list_bytes: Moved from buf_pool.stat.flush_list_bytes.
Protected by buf_pool.flush_list_mutex.
buf_pool_t::page_cleaner_status: Replaces buf_pool_t::n_flush_LRU_,
buf_pool_t::n_flush_list_, and buf_pool_t::page_cleaner_is_idle.
Protected by buf_pool.flush_list_mutex. We will exclusively broadcast
buf_pool.done_flush_list by the buf_flush_page_cleaner thread,
and only wait for it when communicating with buf_flush_page_cleaner.
There is no need to keep a count of pending writes by the
buf_pool.flush_list processing. A single flag suffices for that.
Waits for page write completion can be performed by
simply waiting on block->page.lock, or by invoking
buf_dblwr.wait_for_page_writes().
buf_LRU_block_free_non_file_page(): Broadcast buf_pool.done_free and
set buf_pool.try_LRU_scan when freeing a page. This would be
executed also as part of buf_page_write_complete().
buf_page_write_complete(): Do not broadcast buf_pool.done_flush_list,
and do not acquire buf_pool.mutex unless buf_pool.LRU eviction is needed.
Let buf_dblwr count all writes to persistent pages and broadcast a
condition variable when no outstanding writes remain.
buf_flush_page_cleaner(): Prioritize LRU flushing and eviction right after
"furious flushing" (lsn_limit). Simplify the conditions and reduce the
hold time of buf_pool.flush_list_mutex. Refuse to shut down
or sleep if buf_pool.ran_out(), that is, LRU eviction is needed.
buf_pool_t::page_cleaner_wakeup(): Add the optional parameter for_LRU.
buf_LRU_get_free_block(): Protect buf_lru_free_blocks_error_printed
with buf_pool.mutex. Invoke buf_pool.page_cleaner_wakeup(true) to
to ensure that buf_flush_page_cleaner() will process the LRU flush
request.
buf_do_LRU_batch(), buf_flush_list(), buf_flush_list_space():
Update buf_pool.stat.n_pages_written when submitting writes
(while holding buf_pool.mutex), not when completing them.
buf_page_t::flush(), buf_flush_discard_page(): Require that
the page U-latch be acquired upfront, and remove
buf_page_t::ready_for_flush().
buf_pool_t::delete_from_flush_list(): Remove the parameter "bool clear".
buf_flush_page(): Count pending page writes via buf_dblwr.
buf_flush_try_neighbors(): Take the block of page_id as a parameter.
If the tablespace is dropped before our page has been written out,
release the page U-latch.
buf_pool_invalidate(): Let the caller ensure that there are no
outstanding writes.
buf_flush_wait_batch_end(false),
buf_flush_wait_batch_end_acquiring_mutex(false):
Replaced with buf_dblwr.wait_for_page_writes().
buf_flush_wait_LRU_batch_end(): Replaces buf_flush_wait_batch_end(true).
buf_flush_list(): Remove some broadcast of buf_pool.done_flush_list.
buf_flush_buffer_pool(): Invoke also buf_dblwr.wait_for_page_writes().
buf_pool_t::io_pending(), buf_pool_t::n_flush_list(): Remove.
Outstanding writes are reflected by buf_dblwr.pending_writes().
buf_dblwr_t::init(): New function, to initialize the mutex and
the condition variables, but not the backing store.
buf_dblwr_t::is_created(): Replaces buf_dblwr_t::is_initialised().
buf_dblwr_t::pending_writes(), buf_dblwr_t::writes_pending:
Keeps track of writes of persistent data pages.
buf_flush_LRU(): Allow calls while LRU flushing may be in progress
in another thread.
Tested by Matthias Leich (correctness) and Axel Schwenke (performance)
3 years ago MDEV-23399: Performance regression with write workloads
The buffer pool refactoring in MDEV-15053 and MDEV-22871 shifted
the performance bottleneck to the page flushing.
The configuration parameters will be changed as follows:
innodb_lru_flush_size=32 (new: how many pages to flush on LRU eviction)
innodb_lru_scan_depth=1536 (old: 1024)
innodb_max_dirty_pages_pct=90 (old: 75)
innodb_max_dirty_pages_pct_lwm=75 (old: 0)
Note: The parameter innodb_lru_scan_depth will only affect LRU
eviction of buffer pool pages when a new page is being allocated. The
page cleaner thread will no longer evict any pages. It used to
guarantee that some pages will remain free in the buffer pool. Now, we
perform that eviction 'on demand' in buf_LRU_get_free_block().
The parameter innodb_lru_scan_depth(srv_LRU_scan_depth) is used as follows:
* When the buffer pool is being shrunk in buf_pool_t::withdraw_blocks()
* As a buf_pool.free limit in buf_LRU_list_batch() for terminating
the flushing that is initiated e.g., by buf_LRU_get_free_block()
The parameter also used to serve as an initial limit for unzip_LRU
eviction (evicting uncompressed page frames while retaining
ROW_FORMAT=COMPRESSED pages), but now we will use a hard-coded limit
of 100 or unlimited for invoking buf_LRU_scan_and_free_block().
The status variables will be changed as follows:
innodb_buffer_pool_pages_flushed: This includes also the count of
innodb_buffer_pool_pages_LRU_flushed and should work reliably,
updated one by one in buf_flush_page() to give more real-time
statistics. The function buf_flush_stats(), which we are removing,
was not called in every code path. For both counters, we will use
regular variables that are incremented in a critical section of
buf_pool.mutex. Note that show_innodb_vars() directly links to the
variables, and reads of the counters will *not* be protected by
buf_pool.mutex, so you cannot get a consistent snapshot of both variables.
The following INFORMATION_SCHEMA.INNODB_METRICS counters will be
removed, because the page cleaner no longer deals with writing or
evicting least recently used pages, and because the single-page writes
have been removed:
* buffer_LRU_batch_flush_avg_time_slot
* buffer_LRU_batch_flush_avg_time_thread
* buffer_LRU_batch_flush_avg_time_est
* buffer_LRU_batch_flush_avg_pass
* buffer_LRU_single_flush_scanned
* buffer_LRU_single_flush_num_scan
* buffer_LRU_single_flush_scanned_per_call
When moving to a single buffer pool instance in MDEV-15058, we missed
some opportunity to simplify the buf_flush_page_cleaner thread. It was
unnecessarily using a mutex and some complex data structures, even
though we always have a single page cleaner thread.
Furthermore, the buf_flush_page_cleaner thread had separate 'recovery'
and 'shutdown' modes where it was waiting to be triggered by some
other thread, adding unnecessary latency and potential for hangs in
relatively rarely executed startup or shutdown code.
The page cleaner was also running two kinds of batches in an
interleaved fashion: "LRU flush" (writing out some least recently used
pages and evicting them on write completion) and the normal batches
that aim to increase the MIN(oldest_modification) in the buffer pool,
to help the log checkpoint advance.
The buf_pool.flush_list flushing was being blocked by
buf_block_t::lock for no good reason. Furthermore, if the FIL_PAGE_LSN
of a page is ahead of log_sys.get_flushed_lsn(), that is, what has
been persistently written to the redo log, we would trigger a log
flush and then resume the page flushing. This would unnecessarily
limit the performance of the page cleaner thread and trigger the
infamous messages "InnoDB: page_cleaner: 1000ms intended loop took 4450ms.
The settings might not be optimal" that were suppressed in
commit d1ab89037a518fcffbc50c24e4bd94e4ec33aed0 unless log_warnings>2.
Our revised algorithm will make log_sys.get_flushed_lsn() advance at
the start of buf_flush_lists(), and then execute a 'best effort' to
write out all pages. The flush batches will skip pages that were modified
since the log was written, or are are currently exclusively locked.
The MDEV-13670 message "page_cleaner: 1000ms intended loop took" message
will be removed, because by design, the buf_flush_page_cleaner() should
not be blocked during a batch for extended periods of time.
We will remove the single-page flushing altogether. Related to this,
the debug parameter innodb_doublewrite_batch_size will be removed,
because all of the doublewrite buffer will be used for flushing
batches. If a page needs to be evicted from the buffer pool and all
100 least recently used pages in the buffer pool have unflushed
changes, buf_LRU_get_free_block() will execute buf_flush_lists() to
write out and evict innodb_lru_flush_size pages. At most one thread
will execute buf_flush_lists() in buf_LRU_get_free_block(); other
threads will wait for that LRU flushing batch to finish.
To improve concurrency, we will replace the InnoDB ib_mutex_t and
os_event_t native mutexes and condition variables in this area of code.
Most notably, this means that the buffer pool mutex (buf_pool.mutex)
is no longer instrumented via any InnoDB interfaces. It will continue
to be instrumented via PERFORMANCE_SCHEMA.
For now, both buf_pool.flush_list_mutex and buf_pool.mutex will be
declared with MY_MUTEX_INIT_FAST (PTHREAD_MUTEX_ADAPTIVE_NP). The critical
sections of buf_pool.flush_list_mutex should be shorter than those for
buf_pool.mutex, because in the worst case, they cover a linear scan of
buf_pool.flush_list, while the worst case of a critical section of
buf_pool.mutex covers a linear scan of the potentially much longer
buf_pool.LRU list.
mysql_mutex_is_owner(), safe_mutex_is_owner(): New predicate, usable
with SAFE_MUTEX. Some InnoDB debug assertions need this predicate
instead of mysql_mutex_assert_owner() or mysql_mutex_assert_not_owner().
buf_pool_t::n_flush_LRU, buf_pool_t::n_flush_list:
Replaces buf_pool_t::init_flush[] and buf_pool_t::n_flush[].
The number of active flush operations.
buf_pool_t::mutex, buf_pool_t::flush_list_mutex: Use mysql_mutex_t
instead of ib_mutex_t, to have native mutexes with PERFORMANCE_SCHEMA
and SAFE_MUTEX instrumentation.
buf_pool_t::done_flush_LRU: Condition variable for !n_flush_LRU.
buf_pool_t::done_flush_list: Condition variable for !n_flush_list.
buf_pool_t::do_flush_list: Condition variable to wake up the
buf_flush_page_cleaner when a log checkpoint needs to be written
or the server is being shut down. Replaces buf_flush_event.
We will keep using timed waits (the page cleaner thread will wake
_at least_ once per second), because the calculations for
innodb_adaptive_flushing depend on fixed time intervals.
buf_dblwr: Allocate statically, and move all code to member functions.
Use a native mutex and condition variable. Remove code to deal with
single-page flushing.
buf_dblwr_check_block(): Make the check debug-only. We were spending
a significant amount of execution time in page_simple_validate_new().
flush_counters_t::unzip_LRU_evicted: Remove.
IORequest: Make more members const. FIXME: m_fil_node should be removed.
buf_flush_sync_lsn: Protect by std::atomic, not page_cleaner.mutex
(which we are removing).
page_cleaner_slot_t, page_cleaner_t: Remove many redundant members.
pc_request_flush_slot(): Replaces pc_request() and pc_flush_slot().
recv_writer_thread: Remove. Recovery works just fine without it, if we
simply invoke buf_flush_sync() at the end of each batch in
recv_sys_t::apply().
recv_recovery_from_checkpoint_finish(): Remove. We can simply call
recv_sys.debug_free() directly.
srv_started_redo: Replaces srv_start_state.
SRV_SHUTDOWN_FLUSH_PHASE: Remove. logs_empty_and_mark_files_at_shutdown()
can communicate with the normal page cleaner loop via the new function
flush_buffer_pool().
buf_flush_remove(): Assert that the calling thread is holding
buf_pool.flush_list_mutex. This removes unnecessary mutex operations
from buf_flush_remove_pages() and buf_flush_dirty_pages(),
which replace buf_LRU_flush_or_remove_pages().
buf_flush_lists(): Renamed from buf_flush_batch(), with simplified
interface. Return the number of flushed pages. Clarified comments and
renamed min_n to max_n. Identify LRU batch by lsn=0. Merge all the functions
buf_flush_start(), buf_flush_batch(), buf_flush_end() directly to this
function, which was their only caller, and remove 2 unnecessary
buf_pool.mutex release/re-acquisition that we used to perform around
the buf_flush_batch() call. At the start, if not all log has been
durably written, wait for a background task to do it, or start a new
task to do it. This allows the log write to run concurrently with our
page flushing batch. Any pages that were skipped due to too recent
FIL_PAGE_LSN or due to them being latched by a writer should be flushed
during the next batch, unless there are further modifications to those
pages. It is possible that a page that we must flush due to small
oldest_modification also carries a recent FIL_PAGE_LSN or is being
constantly modified. In the worst case, all writers would then end up
waiting in log_free_check() to allow the flushing and the checkpoint
to complete.
buf_do_flush_list_batch(): Clarify comments, and rename min_n to max_n.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_flush_space(): Auxiliary function to look up a tablespace for
page flushing.
buf_flush_page(): Defer the computation of space->full_crc32(). Never
call log_write_up_to(), but instead skip persistent pages whose latest
modification (FIL_PAGE_LSN) is newer than the redo log. Also skip
pages on which we cannot acquire a shared latch without waiting.
buf_flush_try_neighbors(): Do not bother checking buf_fix_count
because buf_flush_page() will no longer wait for the page latch.
Take the tablespace as a parameter, and only execute this function
when innodb_flush_neighbors>0. Avoid repeated calls of page_id_t::fold().
buf_flush_relocate_on_flush_list(): Declare as cold, and push down
a condition from the callers.
buf_flush_check_neighbor(): Take id.fold() as a parameter.
buf_flush_sync(): Ensure that the buf_pool.flush_list is empty,
because the flushing batch will skip pages whose modifications have
not yet been written to the log or were latched for modification.
buf_free_from_unzip_LRU_list_batch(): Remove redundant local variables.
buf_flush_LRU_list_batch(): Let the caller buf_do_LRU_batch() initialize
the counters, and report n->evicted.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_do_LRU_batch(): Return the number of pages flushed.
buf_LRU_free_page(): Only release and re-acquire buf_pool.mutex if
adaptive hash index entries are pointing to the block.
buf_LRU_get_free_block(): Do not wake up the page cleaner, because it
will no longer perform any useful work for us, and we do not want it
to compete for I/O while buf_flush_lists(innodb_lru_flush_size, 0)
writes out and evicts at most innodb_lru_flush_size pages. (The
function buf_do_LRU_batch() may complete after writing fewer pages if
more than innodb_lru_scan_depth pages end up in buf_pool.free list.)
Eliminate some mutex release-acquire cycles, and wait for the LRU
flush batch to complete before rescanning.
buf_LRU_check_size_of_non_data_objects(): Simplify the code.
buf_page_write_complete(): Remove the parameter evict, and always
evict pages that were part of an LRU flush.
buf_page_create(): Take a pre-allocated page as a parameter.
buf_pool_t::free_block(): Free a pre-allocated block.
recv_sys_t::recover_low(), recv_sys_t::apply(): Preallocate the block
while not holding recv_sys.mutex. During page allocation, we may
initiate a page flush, which in turn may initiate a log flush, which
would require acquiring log_sys.mutex, which should always be acquired
before recv_sys.mutex in order to avoid deadlocks. Therefore, we must
not be holding recv_sys.mutex while allocating a buffer pool block.
BtrBulk::logFreeCheck(): Skip a redundant condition.
row_undo_step(): Do not invoke srv_inc_activity_count() for every row
that is being rolled back. It should suffice to invoke the function in
trx_flush_log_if_needed() during trx_t::commit_in_memory() when the
rollback completes.
sync_check_enable(): Remove. We will enable innodb_sync_debug from the
very beginning.
Reviewed by: Vladislav Vaintroub
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-11369 Instant ADD COLUMN for InnoDB
For InnoDB tables, adding, dropping and reordering columns has
required a rebuild of the table and all its indexes. Since MySQL 5.6
(and MariaDB 10.0) this has been supported online (LOCK=NONE), allowing
concurrent modification of the tables.
This work revises the InnoDB ROW_FORMAT=REDUNDANT, ROW_FORMAT=COMPACT
and ROW_FORMAT=DYNAMIC so that columns can be appended instantaneously,
with only minor changes performed to the table structure. The counter
innodb_instant_alter_column in INFORMATION_SCHEMA.GLOBAL_STATUS
is incremented whenever a table rebuild operation is converted into
an instant ADD COLUMN operation.
ROW_FORMAT=COMPRESSED tables will not support instant ADD COLUMN.
Some usability limitations will be addressed in subsequent work:
MDEV-13134 Introduce ALTER TABLE attributes ALGORITHM=NOCOPY
and ALGORITHM=INSTANT
MDEV-14016 Allow instant ADD COLUMN, ADD INDEX, LOCK=NONE
The format of the clustered index (PRIMARY KEY) is changed as follows:
(1) The FIL_PAGE_TYPE of the root page will be FIL_PAGE_TYPE_INSTANT,
and a new field PAGE_INSTANT will contain the original number of fields
in the clustered index ('core' fields).
If instant ADD COLUMN has not been used or the table becomes empty,
or the very first instant ADD COLUMN operation is rolled back,
the fields PAGE_INSTANT and FIL_PAGE_TYPE will be reset
to 0 and FIL_PAGE_INDEX.
(2) A special 'default row' record is inserted into the leftmost leaf,
between the page infimum and the first user record. This record is
distinguished by the REC_INFO_MIN_REC_FLAG, and it is otherwise in the
same format as records that contain values for the instantly added
columns. This 'default row' always has the same number of fields as
the clustered index according to the table definition. The values of
'core' fields are to be ignored. For other fields, the 'default row'
will contain the default values as they were during the ALTER TABLE
statement. (If the column default values are changed later, those
values will only be stored in the .frm file. The 'default row' will
contain the original evaluated values, which must be the same for
every row.) The 'default row' must be completely hidden from
higher-level access routines. Assertions have been added to ensure
that no 'default row' is ever present in the adaptive hash index
or in locked records. The 'default row' is never delete-marked.
(3) In clustered index leaf page records, the number of fields must
reside between the number of 'core' fields (dict_index_t::n_core_fields
introduced in this work) and dict_index_t::n_fields. If the number
of fields is less than dict_index_t::n_fields, the missing fields
are replaced with the column value of the 'default row'.
Note: The number of fields in the record may shrink if some of the
last instantly added columns are updated to the value that is
in the 'default row'. The function btr_cur_trim() implements this
'compression' on update and rollback; dtuple::trim() implements it
on insert.
(4) In ROW_FORMAT=COMPACT and ROW_FORMAT=DYNAMIC records, the new
status value REC_STATUS_COLUMNS_ADDED will indicate the presence of
a new record header that will encode n_fields-n_core_fields-1 in
1 or 2 bytes. (In ROW_FORMAT=REDUNDANT records, the record header
always explicitly encodes the number of fields.)
We introduce the undo log record type TRX_UNDO_INSERT_DEFAULT for
covering the insert of the 'default row' record when instant ADD COLUMN
is used for the first time. Subsequent instant ADD COLUMN can use
TRX_UNDO_UPD_EXIST_REC.
This is joint work with Vin Chen (陈福荣) from Tencent. The design
that was discussed in April 2017 would not have allowed import or
export of data files, because instead of the 'default row' it would
have introduced a data dictionary table. The test
rpl.rpl_alter_instant is exactly as contributed in pull request #408.
The test innodb.instant_alter is based on a contributed test.
The redo log record format changes for ROW_FORMAT=DYNAMIC and
ROW_FORMAT=COMPACT are as contributed. (With this change present,
crash recovery from MariaDB 10.3.1 will fail in spectacular ways!)
Also the semantics of higher-level redo log records that modify the
PAGE_INSTANT field is changed. The redo log format version identifier
was already changed to LOG_HEADER_FORMAT_CURRENT=103 in MariaDB 10.3.1.
Everything else has been rewritten by me. Thanks to Elena Stepanova,
the code has been tested extensively.
When rolling back an instant ADD COLUMN operation, we must empty the
PAGE_FREE list after deleting or shortening the 'default row' record,
by calling either btr_page_empty() or btr_page_reorganize(). We must
know the size of each entry in the PAGE_FREE list. If rollback left a
freed copy of the 'default row' in the PAGE_FREE list, we would be
unable to determine its size (if it is in ROW_FORMAT=COMPACT or
ROW_FORMAT=DYNAMIC) because it would contain more fields than the
rolled-back definition of the clustered index.
UNIV_SQL_DEFAULT: A new special constant that designates an instantly
added column that is not present in the clustered index record.
len_is_stored(): Check if a length is an actual length. There are
two magic length values: UNIV_SQL_DEFAULT, UNIV_SQL_NULL.
dict_col_t::def_val: The 'default row' value of the column. If the
column is not added instantly, def_val.len will be UNIV_SQL_DEFAULT.
dict_col_t: Add the accessors is_virtual(), is_nullable(), is_instant(),
instant_value().
dict_col_t::remove_instant(): Remove the 'instant ADD' status of
a column.
dict_col_t::name(const dict_table_t& table): Replaces
dict_table_get_col_name().
dict_index_t::n_core_fields: The original number of fields.
For secondary indexes and if instant ADD COLUMN has not been used,
this will be equal to dict_index_t::n_fields.
dict_index_t::n_core_null_bytes: Number of bytes needed to
represent the null flags; usually equal to UT_BITS_IN_BYTES(n_nullable).
dict_index_t::NO_CORE_NULL_BYTES: Magic value signalling that
n_core_null_bytes was not initialized yet from the clustered index
root page.
dict_index_t: Add the accessors is_instant(), is_clust(),
get_n_nullable(), instant_field_value().
dict_index_t::instant_add_field(): Adjust clustered index metadata
for instant ADD COLUMN.
dict_index_t::remove_instant(): Remove the 'instant ADD' status
of a clustered index when the table becomes empty, or the very first
instant ADD COLUMN operation is rolled back.
dict_table_t: Add the accessors is_instant(), is_temporary(),
supports_instant().
dict_table_t::instant_add_column(): Adjust metadata for
instant ADD COLUMN.
dict_table_t::rollback_instant(): Adjust metadata on the rollback
of instant ADD COLUMN.
prepare_inplace_alter_table_dict(): First create the ctx->new_table,
and only then decide if the table really needs to be rebuilt.
We must split the creation of table or index metadata from the
creation of the dictionary table records and the creation of
the data. In this way, we can transform a table-rebuilding operation
into an instant ADD COLUMN operation. Dictionary objects will only
be added to cache when table rebuilding or index creation is needed.
The ctx->instant_table will never be added to cache.
dict_table_t::add_to_cache(): Modified and renamed from
dict_table_add_to_cache(). Do not modify the table metadata.
Let the callers invoke dict_table_add_system_columns() and if needed,
set can_be_evicted.
dict_create_sys_tables_tuple(), dict_create_table_step(): Omit the
system columns (which will now exist in the dict_table_t object
already at this point).
dict_create_table_step(): Expect the callers to invoke
dict_table_add_system_columns().
pars_create_table(): Before creating the table creation execution
graph, invoke dict_table_add_system_columns().
row_create_table_for_mysql(): Expect all callers to invoke
dict_table_add_system_columns().
create_index_dict(): Replaces row_merge_create_index_graph().
innodb_update_n_cols(): Renamed from innobase_update_n_virtual().
Call my_error() if an error occurs.
btr_cur_instant_init(), btr_cur_instant_init_low(),
btr_cur_instant_root_init():
Load additional metadata from the clustered index and set
dict_index_t::n_core_null_bytes. This is invoked
when table metadata is first loaded into the data dictionary.
dict_boot(): Initialize n_core_null_bytes for the four hard-coded
dictionary tables.
dict_create_index_step(): Initialize n_core_null_bytes. This is
executed as part of CREATE TABLE.
dict_index_build_internal_clust(): Initialize n_core_null_bytes to
NO_CORE_NULL_BYTES if table->supports_instant().
row_create_index_for_mysql(): Initialize n_core_null_bytes for
CREATE TEMPORARY TABLE.
commit_cache_norebuild(): Call the code to rename or enlarge columns
in the cache only if instant ADD COLUMN is not being used.
(Instant ADD COLUMN would copy all column metadata from
instant_table to old_table, including the names and lengths.)
PAGE_INSTANT: A new 13-bit field for storing dict_index_t::n_core_fields.
This is repurposing the 16-bit field PAGE_DIRECTION, of which only the
least significant 3 bits were used. The original byte containing
PAGE_DIRECTION will be accessible via the new constant PAGE_DIRECTION_B.
page_get_instant(), page_set_instant(): Accessors for the PAGE_INSTANT.
page_ptr_get_direction(), page_get_direction(),
page_ptr_set_direction(): Accessors for PAGE_DIRECTION.
page_direction_reset(): Reset PAGE_DIRECTION, PAGE_N_DIRECTION.
page_direction_increment(): Increment PAGE_N_DIRECTION
and set PAGE_DIRECTION.
rec_get_offsets(): Use the 'leaf' parameter for non-debug purposes,
and assume that heap_no is always set.
Initialize all dict_index_t::n_fields for ROW_FORMAT=REDUNDANT records,
even if the record contains fewer fields.
rec_offs_make_valid(): Add the parameter 'leaf'.
rec_copy_prefix_to_dtuple(): Assert that the tuple is only built
on the core fields. Instant ADD COLUMN only applies to the
clustered index, and we should never build a search key that has
more than the PRIMARY KEY and possibly DB_TRX_ID,DB_ROLL_PTR.
All these columns are always present.
dict_index_build_data_tuple(): Remove assertions that would be
duplicated in rec_copy_prefix_to_dtuple().
rec_init_offsets(): Support ROW_FORMAT=REDUNDANT records whose
number of fields is between n_core_fields and n_fields.
cmp_rec_rec_with_match(): Implement the comparison between two
MIN_REC_FLAG records.
trx_t::in_rollback: Make the field available in non-debug builds.
trx_start_for_ddl_low(): Remove dangerous error-tolerance.
A dictionary transaction must be flagged as such before it has generated
any undo log records. This is because trx_undo_assign_undo() will mark
the transaction as a dictionary transaction in the undo log header
right before the very first undo log record is being written.
btr_index_rec_validate(): Account for instant ADD COLUMN
row_undo_ins_remove_clust_rec(): On the rollback of an insert into
SYS_COLUMNS, revert instant ADD COLUMN in the cache by removing the
last column from the table and the clustered index.
row_search_on_row_ref(), row_undo_mod_parse_undo_rec(), row_undo_mod(),
trx_undo_update_rec_get_update(): Handle the 'default row'
as a special case.
dtuple_t::trim(index): Omit a redundant suffix of an index tuple right
before insert or update. After instant ADD COLUMN, if the last fields
of a clustered index tuple match the 'default row', there is no
need to store them. While trimming the entry, we must hold a page latch,
so that the table cannot be emptied and the 'default row' be deleted.
btr_cur_optimistic_update(), btr_cur_pessimistic_update(),
row_upd_clust_rec_by_insert(), row_ins_clust_index_entry_low():
Invoke dtuple_t::trim() if needed.
row_ins_clust_index_entry(): Restore dtuple_t::n_fields after calling
row_ins_clust_index_entry_low().
rec_get_converted_size(), rec_get_converted_size_comp(): Allow the number
of fields to be between n_core_fields and n_fields. Do not support
infimum,supremum. They are never supposed to be stored in dtuple_t,
because page creation nowadays uses a lower-level method for initializing
them.
rec_convert_dtuple_to_rec_comp(): Assign the status bits based on the
number of fields.
btr_cur_trim(): In an update, trim the index entry as needed. For the
'default row', handle rollback specially. For user records, omit
fields that match the 'default row'.
btr_cur_optimistic_delete_func(), btr_cur_pessimistic_delete():
Skip locking and adaptive hash index for the 'default row'.
row_log_table_apply_convert_mrec(): Replace 'default row' values if needed.
In the temporary file that is applied by row_log_table_apply(),
we must identify whether the records contain the extra header for
instantly added columns. For now, we will allocate an additional byte
for this for ROW_T_INSERT and ROW_T_UPDATE records when the source table
has been subject to instant ADD COLUMN. The ROW_T_DELETE records are
fine, as they will be converted and will only contain 'core' columns
(PRIMARY KEY and some system columns) that are converted from dtuple_t.
rec_get_converted_size_temp(), rec_init_offsets_temp(),
rec_convert_dtuple_to_temp(): Add the parameter 'status'.
REC_INFO_DEFAULT_ROW = REC_INFO_MIN_REC_FLAG | REC_STATUS_COLUMNS_ADDED:
An info_bits constant for distinguishing the 'default row' record.
rec_comp_status_t: An enum of the status bit values.
rec_leaf_format: An enum that replaces the bool parameter of
rec_init_offsets_comp_ordinary().
8 years ago MDEV-11369 Instant ADD COLUMN for InnoDB
For InnoDB tables, adding, dropping and reordering columns has
required a rebuild of the table and all its indexes. Since MySQL 5.6
(and MariaDB 10.0) this has been supported online (LOCK=NONE), allowing
concurrent modification of the tables.
This work revises the InnoDB ROW_FORMAT=REDUNDANT, ROW_FORMAT=COMPACT
and ROW_FORMAT=DYNAMIC so that columns can be appended instantaneously,
with only minor changes performed to the table structure. The counter
innodb_instant_alter_column in INFORMATION_SCHEMA.GLOBAL_STATUS
is incremented whenever a table rebuild operation is converted into
an instant ADD COLUMN operation.
ROW_FORMAT=COMPRESSED tables will not support instant ADD COLUMN.
Some usability limitations will be addressed in subsequent work:
MDEV-13134 Introduce ALTER TABLE attributes ALGORITHM=NOCOPY
and ALGORITHM=INSTANT
MDEV-14016 Allow instant ADD COLUMN, ADD INDEX, LOCK=NONE
The format of the clustered index (PRIMARY KEY) is changed as follows:
(1) The FIL_PAGE_TYPE of the root page will be FIL_PAGE_TYPE_INSTANT,
and a new field PAGE_INSTANT will contain the original number of fields
in the clustered index ('core' fields).
If instant ADD COLUMN has not been used or the table becomes empty,
or the very first instant ADD COLUMN operation is rolled back,
the fields PAGE_INSTANT and FIL_PAGE_TYPE will be reset
to 0 and FIL_PAGE_INDEX.
(2) A special 'default row' record is inserted into the leftmost leaf,
between the page infimum and the first user record. This record is
distinguished by the REC_INFO_MIN_REC_FLAG, and it is otherwise in the
same format as records that contain values for the instantly added
columns. This 'default row' always has the same number of fields as
the clustered index according to the table definition. The values of
'core' fields are to be ignored. For other fields, the 'default row'
will contain the default values as they were during the ALTER TABLE
statement. (If the column default values are changed later, those
values will only be stored in the .frm file. The 'default row' will
contain the original evaluated values, which must be the same for
every row.) The 'default row' must be completely hidden from
higher-level access routines. Assertions have been added to ensure
that no 'default row' is ever present in the adaptive hash index
or in locked records. The 'default row' is never delete-marked.
(3) In clustered index leaf page records, the number of fields must
reside between the number of 'core' fields (dict_index_t::n_core_fields
introduced in this work) and dict_index_t::n_fields. If the number
of fields is less than dict_index_t::n_fields, the missing fields
are replaced with the column value of the 'default row'.
Note: The number of fields in the record may shrink if some of the
last instantly added columns are updated to the value that is
in the 'default row'. The function btr_cur_trim() implements this
'compression' on update and rollback; dtuple::trim() implements it
on insert.
(4) In ROW_FORMAT=COMPACT and ROW_FORMAT=DYNAMIC records, the new
status value REC_STATUS_COLUMNS_ADDED will indicate the presence of
a new record header that will encode n_fields-n_core_fields-1 in
1 or 2 bytes. (In ROW_FORMAT=REDUNDANT records, the record header
always explicitly encodes the number of fields.)
We introduce the undo log record type TRX_UNDO_INSERT_DEFAULT for
covering the insert of the 'default row' record when instant ADD COLUMN
is used for the first time. Subsequent instant ADD COLUMN can use
TRX_UNDO_UPD_EXIST_REC.
This is joint work with Vin Chen (陈福荣) from Tencent. The design
that was discussed in April 2017 would not have allowed import or
export of data files, because instead of the 'default row' it would
have introduced a data dictionary table. The test
rpl.rpl_alter_instant is exactly as contributed in pull request #408.
The test innodb.instant_alter is based on a contributed test.
The redo log record format changes for ROW_FORMAT=DYNAMIC and
ROW_FORMAT=COMPACT are as contributed. (With this change present,
crash recovery from MariaDB 10.3.1 will fail in spectacular ways!)
Also the semantics of higher-level redo log records that modify the
PAGE_INSTANT field is changed. The redo log format version identifier
was already changed to LOG_HEADER_FORMAT_CURRENT=103 in MariaDB 10.3.1.
Everything else has been rewritten by me. Thanks to Elena Stepanova,
the code has been tested extensively.
When rolling back an instant ADD COLUMN operation, we must empty the
PAGE_FREE list after deleting or shortening the 'default row' record,
by calling either btr_page_empty() or btr_page_reorganize(). We must
know the size of each entry in the PAGE_FREE list. If rollback left a
freed copy of the 'default row' in the PAGE_FREE list, we would be
unable to determine its size (if it is in ROW_FORMAT=COMPACT or
ROW_FORMAT=DYNAMIC) because it would contain more fields than the
rolled-back definition of the clustered index.
UNIV_SQL_DEFAULT: A new special constant that designates an instantly
added column that is not present in the clustered index record.
len_is_stored(): Check if a length is an actual length. There are
two magic length values: UNIV_SQL_DEFAULT, UNIV_SQL_NULL.
dict_col_t::def_val: The 'default row' value of the column. If the
column is not added instantly, def_val.len will be UNIV_SQL_DEFAULT.
dict_col_t: Add the accessors is_virtual(), is_nullable(), is_instant(),
instant_value().
dict_col_t::remove_instant(): Remove the 'instant ADD' status of
a column.
dict_col_t::name(const dict_table_t& table): Replaces
dict_table_get_col_name().
dict_index_t::n_core_fields: The original number of fields.
For secondary indexes and if instant ADD COLUMN has not been used,
this will be equal to dict_index_t::n_fields.
dict_index_t::n_core_null_bytes: Number of bytes needed to
represent the null flags; usually equal to UT_BITS_IN_BYTES(n_nullable).
dict_index_t::NO_CORE_NULL_BYTES: Magic value signalling that
n_core_null_bytes was not initialized yet from the clustered index
root page.
dict_index_t: Add the accessors is_instant(), is_clust(),
get_n_nullable(), instant_field_value().
dict_index_t::instant_add_field(): Adjust clustered index metadata
for instant ADD COLUMN.
dict_index_t::remove_instant(): Remove the 'instant ADD' status
of a clustered index when the table becomes empty, or the very first
instant ADD COLUMN operation is rolled back.
dict_table_t: Add the accessors is_instant(), is_temporary(),
supports_instant().
dict_table_t::instant_add_column(): Adjust metadata for
instant ADD COLUMN.
dict_table_t::rollback_instant(): Adjust metadata on the rollback
of instant ADD COLUMN.
prepare_inplace_alter_table_dict(): First create the ctx->new_table,
and only then decide if the table really needs to be rebuilt.
We must split the creation of table or index metadata from the
creation of the dictionary table records and the creation of
the data. In this way, we can transform a table-rebuilding operation
into an instant ADD COLUMN operation. Dictionary objects will only
be added to cache when table rebuilding or index creation is needed.
The ctx->instant_table will never be added to cache.
dict_table_t::add_to_cache(): Modified and renamed from
dict_table_add_to_cache(). Do not modify the table metadata.
Let the callers invoke dict_table_add_system_columns() and if needed,
set can_be_evicted.
dict_create_sys_tables_tuple(), dict_create_table_step(): Omit the
system columns (which will now exist in the dict_table_t object
already at this point).
dict_create_table_step(): Expect the callers to invoke
dict_table_add_system_columns().
pars_create_table(): Before creating the table creation execution
graph, invoke dict_table_add_system_columns().
row_create_table_for_mysql(): Expect all callers to invoke
dict_table_add_system_columns().
create_index_dict(): Replaces row_merge_create_index_graph().
innodb_update_n_cols(): Renamed from innobase_update_n_virtual().
Call my_error() if an error occurs.
btr_cur_instant_init(), btr_cur_instant_init_low(),
btr_cur_instant_root_init():
Load additional metadata from the clustered index and set
dict_index_t::n_core_null_bytes. This is invoked
when table metadata is first loaded into the data dictionary.
dict_boot(): Initialize n_core_null_bytes for the four hard-coded
dictionary tables.
dict_create_index_step(): Initialize n_core_null_bytes. This is
executed as part of CREATE TABLE.
dict_index_build_internal_clust(): Initialize n_core_null_bytes to
NO_CORE_NULL_BYTES if table->supports_instant().
row_create_index_for_mysql(): Initialize n_core_null_bytes for
CREATE TEMPORARY TABLE.
commit_cache_norebuild(): Call the code to rename or enlarge columns
in the cache only if instant ADD COLUMN is not being used.
(Instant ADD COLUMN would copy all column metadata from
instant_table to old_table, including the names and lengths.)
PAGE_INSTANT: A new 13-bit field for storing dict_index_t::n_core_fields.
This is repurposing the 16-bit field PAGE_DIRECTION, of which only the
least significant 3 bits were used. The original byte containing
PAGE_DIRECTION will be accessible via the new constant PAGE_DIRECTION_B.
page_get_instant(), page_set_instant(): Accessors for the PAGE_INSTANT.
page_ptr_get_direction(), page_get_direction(),
page_ptr_set_direction(): Accessors for PAGE_DIRECTION.
page_direction_reset(): Reset PAGE_DIRECTION, PAGE_N_DIRECTION.
page_direction_increment(): Increment PAGE_N_DIRECTION
and set PAGE_DIRECTION.
rec_get_offsets(): Use the 'leaf' parameter for non-debug purposes,
and assume that heap_no is always set.
Initialize all dict_index_t::n_fields for ROW_FORMAT=REDUNDANT records,
even if the record contains fewer fields.
rec_offs_make_valid(): Add the parameter 'leaf'.
rec_copy_prefix_to_dtuple(): Assert that the tuple is only built
on the core fields. Instant ADD COLUMN only applies to the
clustered index, and we should never build a search key that has
more than the PRIMARY KEY and possibly DB_TRX_ID,DB_ROLL_PTR.
All these columns are always present.
dict_index_build_data_tuple(): Remove assertions that would be
duplicated in rec_copy_prefix_to_dtuple().
rec_init_offsets(): Support ROW_FORMAT=REDUNDANT records whose
number of fields is between n_core_fields and n_fields.
cmp_rec_rec_with_match(): Implement the comparison between two
MIN_REC_FLAG records.
trx_t::in_rollback: Make the field available in non-debug builds.
trx_start_for_ddl_low(): Remove dangerous error-tolerance.
A dictionary transaction must be flagged as such before it has generated
any undo log records. This is because trx_undo_assign_undo() will mark
the transaction as a dictionary transaction in the undo log header
right before the very first undo log record is being written.
btr_index_rec_validate(): Account for instant ADD COLUMN
row_undo_ins_remove_clust_rec(): On the rollback of an insert into
SYS_COLUMNS, revert instant ADD COLUMN in the cache by removing the
last column from the table and the clustered index.
row_search_on_row_ref(), row_undo_mod_parse_undo_rec(), row_undo_mod(),
trx_undo_update_rec_get_update(): Handle the 'default row'
as a special case.
dtuple_t::trim(index): Omit a redundant suffix of an index tuple right
before insert or update. After instant ADD COLUMN, if the last fields
of a clustered index tuple match the 'default row', there is no
need to store them. While trimming the entry, we must hold a page latch,
so that the table cannot be emptied and the 'default row' be deleted.
btr_cur_optimistic_update(), btr_cur_pessimistic_update(),
row_upd_clust_rec_by_insert(), row_ins_clust_index_entry_low():
Invoke dtuple_t::trim() if needed.
row_ins_clust_index_entry(): Restore dtuple_t::n_fields after calling
row_ins_clust_index_entry_low().
rec_get_converted_size(), rec_get_converted_size_comp(): Allow the number
of fields to be between n_core_fields and n_fields. Do not support
infimum,supremum. They are never supposed to be stored in dtuple_t,
because page creation nowadays uses a lower-level method for initializing
them.
rec_convert_dtuple_to_rec_comp(): Assign the status bits based on the
number of fields.
btr_cur_trim(): In an update, trim the index entry as needed. For the
'default row', handle rollback specially. For user records, omit
fields that match the 'default row'.
btr_cur_optimistic_delete_func(), btr_cur_pessimistic_delete():
Skip locking and adaptive hash index for the 'default row'.
row_log_table_apply_convert_mrec(): Replace 'default row' values if needed.
In the temporary file that is applied by row_log_table_apply(),
we must identify whether the records contain the extra header for
instantly added columns. For now, we will allocate an additional byte
for this for ROW_T_INSERT and ROW_T_UPDATE records when the source table
has been subject to instant ADD COLUMN. The ROW_T_DELETE records are
fine, as they will be converted and will only contain 'core' columns
(PRIMARY KEY and some system columns) that are converted from dtuple_t.
rec_get_converted_size_temp(), rec_init_offsets_temp(),
rec_convert_dtuple_to_temp(): Add the parameter 'status'.
REC_INFO_DEFAULT_ROW = REC_INFO_MIN_REC_FLAG | REC_STATUS_COLUMNS_ADDED:
An info_bits constant for distinguishing the 'default row' record.
rec_comp_status_t: An enum of the status bit values.
rec_leaf_format: An enum that replaces the bool parameter of
rec_init_offsets_comp_ordinary().
8 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-12026: Implement innodb_checksum_algorithm=full_crc32
MariaDB data-at-rest encryption (innodb_encrypt_tables)
had repurposed the same unused data field that was repurposed
in MySQL 5.7 (and MariaDB 10.2) for the Split Sequence Number (SSN)
field of SPATIAL INDEX. Because of this, MariaDB was unable to
support encryption on SPATIAL INDEX pages.
Furthermore, InnoDB page checksums skipped some bytes, and there
are multiple variations and checksum algorithms. By default,
InnoDB accepts all variations of all algorithms that ever existed.
This unnecessarily weakens the page checksums.
We hereby introduce two more innodb_checksum_algorithm variants
(full_crc32, strict_full_crc32) that are special in a way:
When either setting is active, newly created data files will
carry a flag (fil_space_t::full_crc32()) that indicates that
all pages of the file will use a full CRC-32C checksum over the
entire page contents (excluding the bytes where the checksum
is stored, at the very end of the page). Such files will always
use that checksum, no matter what the parameter
innodb_checksum_algorithm is assigned to.
For old files, the old checksum algorithms will continue to be
used. The value strict_full_crc32 will be equivalent to strict_crc32
and the value full_crc32 will be equivalent to crc32.
ROW_FORMAT=COMPRESSED tables will only use the old format.
These tables do not support new features, such as larger
innodb_page_size or instant ADD/DROP COLUMN. They may be
deprecated in the future. We do not want an unnecessary
file format change for them.
The new full_crc32() format also cleans up the MariaDB tablespace
flags. We will reserve flags to store the page_compressed
compression algorithm, and to store the compressed payload length,
so that checksum can be computed over the compressed (and
possibly encrypted) stream and can be validated without
decrypting or decompressing the page.
In the full_crc32 format, there no longer are separate before-encryption
and after-encryption checksums for pages. The single checksum is
computed on the page contents that is written to the file.
We do not make the new algorithm the default for two reasons.
First, MariaDB 10.4.2 was a beta release, and the default values
of parameters should not change after beta. Second, we did not
yet implement the full_crc32 format for page_compressed pages.
This will be fixed in MDEV-18644.
This is joint work with Marko Mäkelä.
7 years ago MDEV-18644: Support full_crc32 for page_compressed
This is a follow-up task to MDEV-12026, which introduced
innodb_checksum_algorithm=full_crc32 and a simpler page format.
MDEV-12026 did not enable full_crc32 for page_compressed tables,
which we will be doing now.
This is joint work with Thirunarayanan Balathandayuthapani.
For innodb_checksum_algorithm=full_crc32 we change the
page_compressed format as follows:
FIL_PAGE_TYPE: The most significant bit will be set to indicate
page_compressed format. The least significant bits will contain
the compressed page size, rounded up to a multiple of 256 bytes.
The checksum will be stored in the last 4 bytes of the page
(whether it is the full page or a page_compressed page whose
size is determined by FIL_PAGE_TYPE), covering all preceding
bytes of the page. If encryption is used, then the page will
be encrypted between compression and computing the checksum.
For page_compressed, FIL_PAGE_LSN will not be repeated at
the end of the page.
FSP_SPACE_FLAGS (already implemented as part of MDEV-12026):
We will store the innodb_compression_algorithm that may be used
to compress pages. Previously, the choice of algorithm was written
to each compressed data page separately, and one would be unable
to know in advance which compression algorithm(s) are used.
fil_space_t::full_crc32_page_compressed_len(): Determine if the
page_compressed algorithm of the tablespace needs to know the
exact length of the compressed data. If yes, we will reserve and
write an extra byte for this right before the checksum.
buf_page_is_compressed(): Determine if a page uses page_compressed
(in any innodb_checksum_algorithm).
fil_page_decompress(): Pass also fil_space_t::flags so that the
format can be determined.
buf_page_is_zeroes(): Check if a page is full of zero bytes.
buf_page_full_crc32_is_corrupted(): Renamed from
buf_encrypted_full_crc32_page_is_corrupted(). For full_crc32,
we always simply validate the checksum to the page contents,
while the physical page size is explicitly specified by an
unencrypted part of the page header.
buf_page_full_crc32_size(): Determine the size of a full_crc32 page.
buf_dblwr_check_page_lsn(): Make this a debug-only function, because
it involves potentially costly lookups of fil_space_t.
create_table_info_t::check_table_options(),
ha_innobase::check_if_supported_inplace_alter(): Do allow the creation
of SPATIAL INDEX with full_crc32 also when page_compressed is used.
commit_cache_norebuild(): Preserve the compression algorithm when
updating the page_compression_level.
dict_tf_to_fsp_flags(): Set the flags for page compression algorithm.
FIXME: Maybe there should be a table option page_compression_algorithm
and a session variable to back it?
7 years ago MDEV-12026: Implement innodb_checksum_algorithm=full_crc32
MariaDB data-at-rest encryption (innodb_encrypt_tables)
had repurposed the same unused data field that was repurposed
in MySQL 5.7 (and MariaDB 10.2) for the Split Sequence Number (SSN)
field of SPATIAL INDEX. Because of this, MariaDB was unable to
support encryption on SPATIAL INDEX pages.
Furthermore, InnoDB page checksums skipped some bytes, and there
are multiple variations and checksum algorithms. By default,
InnoDB accepts all variations of all algorithms that ever existed.
This unnecessarily weakens the page checksums.
We hereby introduce two more innodb_checksum_algorithm variants
(full_crc32, strict_full_crc32) that are special in a way:
When either setting is active, newly created data files will
carry a flag (fil_space_t::full_crc32()) that indicates that
all pages of the file will use a full CRC-32C checksum over the
entire page contents (excluding the bytes where the checksum
is stored, at the very end of the page). Such files will always
use that checksum, no matter what the parameter
innodb_checksum_algorithm is assigned to.
For old files, the old checksum algorithms will continue to be
used. The value strict_full_crc32 will be equivalent to strict_crc32
and the value full_crc32 will be equivalent to crc32.
ROW_FORMAT=COMPRESSED tables will only use the old format.
These tables do not support new features, such as larger
innodb_page_size or instant ADD/DROP COLUMN. They may be
deprecated in the future. We do not want an unnecessary
file format change for them.
The new full_crc32() format also cleans up the MariaDB tablespace
flags. We will reserve flags to store the page_compressed
compression algorithm, and to store the compressed payload length,
so that checksum can be computed over the compressed (and
possibly encrypted) stream and can be validated without
decrypting or decompressing the page.
In the full_crc32 format, there no longer are separate before-encryption
and after-encryption checksums for pages. The single checksum is
computed on the page contents that is written to the file.
We do not make the new algorithm the default for two reasons.
First, MariaDB 10.4.2 was a beta release, and the default values
of parameters should not change after beta. Second, we did not
yet implement the full_crc32 format for page_compressed pages.
This will be fixed in MDEV-18644.
This is joint work with Marko Mäkelä.
7 years ago MDEV-12026: Implement innodb_checksum_algorithm=full_crc32
MariaDB data-at-rest encryption (innodb_encrypt_tables)
had repurposed the same unused data field that was repurposed
in MySQL 5.7 (and MariaDB 10.2) for the Split Sequence Number (SSN)
field of SPATIAL INDEX. Because of this, MariaDB was unable to
support encryption on SPATIAL INDEX pages.
Furthermore, InnoDB page checksums skipped some bytes, and there
are multiple variations and checksum algorithms. By default,
InnoDB accepts all variations of all algorithms that ever existed.
This unnecessarily weakens the page checksums.
We hereby introduce two more innodb_checksum_algorithm variants
(full_crc32, strict_full_crc32) that are special in a way:
When either setting is active, newly created data files will
carry a flag (fil_space_t::full_crc32()) that indicates that
all pages of the file will use a full CRC-32C checksum over the
entire page contents (excluding the bytes where the checksum
is stored, at the very end of the page). Such files will always
use that checksum, no matter what the parameter
innodb_checksum_algorithm is assigned to.
For old files, the old checksum algorithms will continue to be
used. The value strict_full_crc32 will be equivalent to strict_crc32
and the value full_crc32 will be equivalent to crc32.
ROW_FORMAT=COMPRESSED tables will only use the old format.
These tables do not support new features, such as larger
innodb_page_size or instant ADD/DROP COLUMN. They may be
deprecated in the future. We do not want an unnecessary
file format change for them.
The new full_crc32() format also cleans up the MariaDB tablespace
flags. We will reserve flags to store the page_compressed
compression algorithm, and to store the compressed payload length,
so that checksum can be computed over the compressed (and
possibly encrypted) stream and can be validated without
decrypting or decompressing the page.
In the full_crc32 format, there no longer are separate before-encryption
and after-encryption checksums for pages. The single checksum is
computed on the page contents that is written to the file.
We do not make the new algorithm the default for two reasons.
First, MariaDB 10.4.2 was a beta release, and the default values
of parameters should not change after beta. Second, we did not
yet implement the full_crc32 format for page_compressed pages.
This will be fixed in MDEV-18644.
This is joint work with Marko Mäkelä.
7 years ago MDEV-12026: Implement innodb_checksum_algorithm=full_crc32
MariaDB data-at-rest encryption (innodb_encrypt_tables)
had repurposed the same unused data field that was repurposed
in MySQL 5.7 (and MariaDB 10.2) for the Split Sequence Number (SSN)
field of SPATIAL INDEX. Because of this, MariaDB was unable to
support encryption on SPATIAL INDEX pages.
Furthermore, InnoDB page checksums skipped some bytes, and there
are multiple variations and checksum algorithms. By default,
InnoDB accepts all variations of all algorithms that ever existed.
This unnecessarily weakens the page checksums.
We hereby introduce two more innodb_checksum_algorithm variants
(full_crc32, strict_full_crc32) that are special in a way:
When either setting is active, newly created data files will
carry a flag (fil_space_t::full_crc32()) that indicates that
all pages of the file will use a full CRC-32C checksum over the
entire page contents (excluding the bytes where the checksum
is stored, at the very end of the page). Such files will always
use that checksum, no matter what the parameter
innodb_checksum_algorithm is assigned to.
For old files, the old checksum algorithms will continue to be
used. The value strict_full_crc32 will be equivalent to strict_crc32
and the value full_crc32 will be equivalent to crc32.
ROW_FORMAT=COMPRESSED tables will only use the old format.
These tables do not support new features, such as larger
innodb_page_size or instant ADD/DROP COLUMN. They may be
deprecated in the future. We do not want an unnecessary
file format change for them.
The new full_crc32() format also cleans up the MariaDB tablespace
flags. We will reserve flags to store the page_compressed
compression algorithm, and to store the compressed payload length,
so that checksum can be computed over the compressed (and
possibly encrypted) stream and can be validated without
decrypting or decompressing the page.
In the full_crc32 format, there no longer are separate before-encryption
and after-encryption checksums for pages. The single checksum is
computed on the page contents that is written to the file.
We do not make the new algorithm the default for two reasons.
First, MariaDB 10.4.2 was a beta release, and the default values
of parameters should not change after beta. Second, we did not
yet implement the full_crc32 format for page_compressed pages.
This will be fixed in MDEV-18644.
This is joint work with Marko Mäkelä.
7 years ago MDEV-12026: Implement innodb_checksum_algorithm=full_crc32
MariaDB data-at-rest encryption (innodb_encrypt_tables)
had repurposed the same unused data field that was repurposed
in MySQL 5.7 (and MariaDB 10.2) for the Split Sequence Number (SSN)
field of SPATIAL INDEX. Because of this, MariaDB was unable to
support encryption on SPATIAL INDEX pages.
Furthermore, InnoDB page checksums skipped some bytes, and there
are multiple variations and checksum algorithms. By default,
InnoDB accepts all variations of all algorithms that ever existed.
This unnecessarily weakens the page checksums.
We hereby introduce two more innodb_checksum_algorithm variants
(full_crc32, strict_full_crc32) that are special in a way:
When either setting is active, newly created data files will
carry a flag (fil_space_t::full_crc32()) that indicates that
all pages of the file will use a full CRC-32C checksum over the
entire page contents (excluding the bytes where the checksum
is stored, at the very end of the page). Such files will always
use that checksum, no matter what the parameter
innodb_checksum_algorithm is assigned to.
For old files, the old checksum algorithms will continue to be
used. The value strict_full_crc32 will be equivalent to strict_crc32
and the value full_crc32 will be equivalent to crc32.
ROW_FORMAT=COMPRESSED tables will only use the old format.
These tables do not support new features, such as larger
innodb_page_size or instant ADD/DROP COLUMN. They may be
deprecated in the future. We do not want an unnecessary
file format change for them.
The new full_crc32() format also cleans up the MariaDB tablespace
flags. We will reserve flags to store the page_compressed
compression algorithm, and to store the compressed payload length,
so that checksum can be computed over the compressed (and
possibly encrypted) stream and can be validated without
decrypting or decompressing the page.
In the full_crc32 format, there no longer are separate before-encryption
and after-encryption checksums for pages. The single checksum is
computed on the page contents that is written to the file.
We do not make the new algorithm the default for two reasons.
First, MariaDB 10.4.2 was a beta release, and the default values
of parameters should not change after beta. Second, we did not
yet implement the full_crc32 format for page_compressed pages.
This will be fixed in MDEV-18644.
This is joint work with Marko Mäkelä.
7 years ago MDEV-12026: Implement innodb_checksum_algorithm=full_crc32
MariaDB data-at-rest encryption (innodb_encrypt_tables)
had repurposed the same unused data field that was repurposed
in MySQL 5.7 (and MariaDB 10.2) for the Split Sequence Number (SSN)
field of SPATIAL INDEX. Because of this, MariaDB was unable to
support encryption on SPATIAL INDEX pages.
Furthermore, InnoDB page checksums skipped some bytes, and there
are multiple variations and checksum algorithms. By default,
InnoDB accepts all variations of all algorithms that ever existed.
This unnecessarily weakens the page checksums.
We hereby introduce two more innodb_checksum_algorithm variants
(full_crc32, strict_full_crc32) that are special in a way:
When either setting is active, newly created data files will
carry a flag (fil_space_t::full_crc32()) that indicates that
all pages of the file will use a full CRC-32C checksum over the
entire page contents (excluding the bytes where the checksum
is stored, at the very end of the page). Such files will always
use that checksum, no matter what the parameter
innodb_checksum_algorithm is assigned to.
For old files, the old checksum algorithms will continue to be
used. The value strict_full_crc32 will be equivalent to strict_crc32
and the value full_crc32 will be equivalent to crc32.
ROW_FORMAT=COMPRESSED tables will only use the old format.
These tables do not support new features, such as larger
innodb_page_size or instant ADD/DROP COLUMN. They may be
deprecated in the future. We do not want an unnecessary
file format change for them.
The new full_crc32() format also cleans up the MariaDB tablespace
flags. We will reserve flags to store the page_compressed
compression algorithm, and to store the compressed payload length,
so that checksum can be computed over the compressed (and
possibly encrypted) stream and can be validated without
decrypting or decompressing the page.
In the full_crc32 format, there no longer are separate before-encryption
and after-encryption checksums for pages. The single checksum is
computed on the page contents that is written to the file.
We do not make the new algorithm the default for two reasons.
First, MariaDB 10.4.2 was a beta release, and the default values
of parameters should not change after beta. Second, we did not
yet implement the full_crc32 format for page_compressed pages.
This will be fixed in MDEV-18644.
This is joint work with Marko Mäkelä.
7 years ago MDEV-12602 InnoDB: Failing assertion: space->n_pending_ops == 0
This fixes a regression caused by MDEV-12428.
When we introduced a variant of fil_space_acquire() that could
increment space->n_pending_ops after space->stop_new_ops was set,
the logic of fil_check_pending_operations() was broken.
fil_space_t::n_pending_ios: A new field to track read or write
access from the buffer pool routines immediately before a block
write or after a block read in the file system.
fil_space_acquire_for_io(), fil_space_release_for_io(): Similar
to fil_space_acquire_silent() and fil_space_release(), but
modify fil_space_t::n_pending_ios instead of fil_space_t::n_pending_ops.
Adjust a number of places accordingly, and remove some redundant
tablespace lookups.
The following parts of this fix differ from the 10.2 version of this fix:
buf_page_get_corrupt(): Add a tablespace parameter.
In 10.2, we already had a two-phase process of freeing fil_space objects
(first, fil_space_detach(), then release fil_system->mutex, and finally
free the fil_space and fil_node objects).
fil_space_free_and_mutex_exit(): Renamed from fil_space_free().
Detach the tablespace from the fil_system cache, release the
fil_system->mutex, and then wait for space->n_pending_ios to reach 0,
to avoid accessing freed data in a concurrent thread.
During the wait, future calls to fil_space_acquire_for_io() will
not find this tablespace, and the count can only be decremented to 0,
at which point it is safe to free the objects.
fil_node_free_part1(), fil_node_free_part2(): Refactored from
fil_node_free().
9 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-12602 InnoDB: Failing assertion: space->n_pending_ops == 0
This fixes a regression caused by MDEV-12428.
When we introduced a variant of fil_space_acquire() that could
increment space->n_pending_ops after space->stop_new_ops was set,
the logic of fil_check_pending_operations() was broken.
fil_space_t::n_pending_ios: A new field to track read or write
access from the buffer pool routines immediately before a block
write or after a block read in the file system.
fil_space_acquire_for_io(), fil_space_release_for_io(): Similar
to fil_space_acquire_silent() and fil_space_release(), but
modify fil_space_t::n_pending_ios instead of fil_space_t::n_pending_ops.
Adjust a number of places accordingly, and remove some redundant
tablespace lookups.
The following parts of this fix differ from the 10.2 version of this fix:
buf_page_get_corrupt(): Add a tablespace parameter.
In 10.2, we already had a two-phase process of freeing fil_space objects
(first, fil_space_detach(), then release fil_system->mutex, and finally
free the fil_space and fil_node objects).
fil_space_free_and_mutex_exit(): Renamed from fil_space_free().
Detach the tablespace from the fil_system cache, release the
fil_system->mutex, and then wait for space->n_pending_ios to reach 0,
to avoid accessing freed data in a concurrent thread.
During the wait, future calls to fil_space_acquire_for_io() will
not find this tablespace, and the count can only be decremented to 0,
at which point it is safe to free the objects.
fil_node_free_part1(), fil_node_free_part2(): Refactored from
fil_node_free().
9 years ago MDEV-34830: LSN in the future is not being treated as serious corruption
The invariant of write-ahead logging is that before any change to a
page is written to the data file, the corresponding log record must
must first have been durably written.
In crash recovery, there were some sloppy checks for this. Let us
implement accurate checks and flag an inconsistency as a hard error,
so that we can avoid further corruption of a corrupted database.
For data extraction from the corrupted database, innodb_force_recovery
can be used.
Before recovery is reading any data pages or invoking
buf_dblwr_t::recover() to recover torn pages from the
doublewrite buffer, InnoDB will have parsed the log until the
final LSN and updated log_sys.lsn to that. So, we can rely on
log_sys.lsn at all times. The doublewrite buffer recovery has been
refactored in such a way that the recv_sys.dblwr.pages may be consulted
while discovering files and their page sizes, but nothing will be
written back to data files before buf_dblwr_t::recover() is invoked.
recv_max_page_lsn, recv_lsn_checks_on: Remove.
recv_sys_t::validate_checkpoint(): Validate the write-ahead-logging
condition at the end of the recovery.
recv_dblwr_t::validate_page(): Keep track of the maximum LSN
(if we are checking a non-doublewrite copy of a page) but
do not complain LSN being in the future. The doublewrite buffer
is a special case, because it will be read early during recovery.
Besides, starting with commit 762bcb81b5bf9bbde61fed59afb26417f4ce1e86
the dblwr=true copies of pages may legitimately be "too new".
recv_dblwr_t::find_page(): Find a valid page with the smallest
FIL_PAGE_LSN that is in the valid range for recovery.
recv_dblwr_t::restore_first_page(): Replaced by find_page().
Only buf_dblwr_t::recover() will write to data files.
buf_dblwr_t::recover(): Simplify the message output. Do attempt
doublewrite recovery on user page read error. Ignore doublewrite
pages whose FIL_PAGE_LSN is outside the usable bounds. Previously,
we could wrongly recover a too new page from the doublewrite buffer.
It is unlikely that this could have lead to an actual error.
Write back all recovered pages from the doublewrite buffer here,
including for the first page of any tablespace.
buf_page_is_corrupted(): Distinguish the return values
CORRUPTED_FUTURE_LSN and CORRUPTED_OTHER.
buf_page_check_corrupt(): Return the error code DB_CORRUPTION
in case the LSN is in the future.
Datafile::read_first_page_flags(): Split from read_first_page().
Take a copy of the first page as a parameter.
recv_sys_t::free_corrupted_page(): Take the file as a parameter
and return whether a message was displayed. This avoids some duplicated
and incomplete error messages.
buf_page_t::read_complete(): Remove some redundant output and always
display the name of the corrupted file. Never return DB_FAIL;
use it only in internal error handling.
IORequest::read_complete(): Assume that buf_page_t::read_complete()
will have reported any error.
fil_space_t::set_corrupted(): Return whether this is the first time
the tablespace had been flagged as corrupted.
Datafile::validate_first_page(), fil_node_open_file_low(),
fil_node_open_file(), fil_space_t::read_page0(),
fil_node_t::read_page0(): Add a parameter for a copy of the
first page, and a parameter to indicate whether the FIL_PAGE_LSN
check should be suppressed. Before buf_dblwr_t::recover() is
invoked, we cannot validate the FIL_PAGE_LSN, but we can trust the
FSP_SPACE_FLAGS and the tablespace ID that may be present in a
potentially too new copy of a page.
Reviewed by: Debarun Banerjee
12 months ago MDEV-12602 InnoDB: Failing assertion: space->n_pending_ops == 0
This fixes a regression caused by MDEV-12428.
When we introduced a variant of fil_space_acquire() that could
increment space->n_pending_ops after space->stop_new_ops was set,
the logic of fil_check_pending_operations() was broken.
fil_space_t::n_pending_ios: A new field to track read or write
access from the buffer pool routines immediately before a block
write or after a block read in the file system.
fil_space_acquire_for_io(), fil_space_release_for_io(): Similar
to fil_space_acquire_silent() and fil_space_release(), but
modify fil_space_t::n_pending_ios instead of fil_space_t::n_pending_ops.
Adjust a number of places accordingly, and remove some redundant
tablespace lookups.
The following parts of this fix differ from the 10.2 version of this fix:
buf_page_get_corrupt(): Add a tablespace parameter.
In 10.2, we already had a two-phase process of freeing fil_space objects
(first, fil_space_detach(), then release fil_system->mutex, and finally
free the fil_space and fil_node objects).
fil_space_free_and_mutex_exit(): Renamed from fil_space_free().
Detach the tablespace from the fil_system cache, release the
fil_system->mutex, and then wait for space->n_pending_ios to reach 0,
to avoid accessing freed data in a concurrent thread.
During the wait, future calls to fil_space_acquire_for_io() will
not find this tablespace, and the count can only be decremented to 0,
at which point it is safe to free the objects.
fil_node_free_part1(), fil_node_free_part2(): Refactored from
fil_node_free().
9 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-12253: Buffer pool blocks are accessed after they have been freed
Problem was that bpage was referenced after it was already freed
from LRU. Fixed by adding a new variable encrypted that is
passed down to buf_page_check_corrupt() and used in
buf_page_get_gen() to stop processing page read.
This patch should also address following test failures and
bugs:
MDEV-12419: IMPORT should not look up tablespace in
PageConverter::validate(). This is now removed.
MDEV-10099: encryption.innodb_onlinealter_encryption fails
sporadically in buildbot
MDEV-11420: encryption.innodb_encryption-page-compression
failed in buildbot
MDEV-11222: encryption.encrypt_and_grep failed in buildbot on P8
Removed dict_table_t::is_encrypted and dict_table_t::ibd_file_missing
and replaced these with dict_table_t::file_unreadable. Table
ibd file is missing if fil_get_space(space_id) returns NULL
and encrypted if not. Removed dict_table_t::is_corrupted field.
Ported FilSpace class from 10.2 and using that on buf_page_check_corrupt(),
buf_page_decrypt_after_read(), buf_page_encrypt_before_write(),
buf_dblwr_process(), buf_read_page(), dict_stats_save_defrag_stats().
Added test cases when enrypted page could be read while doing
redo log crash recovery. Also added test case for row compressed
blobs.
btr_cur_open_at_index_side_func(),
btr_cur_open_at_rnd_pos_func(): Avoid referencing block that is
NULL.
buf_page_get_zip(): Issue error if page read fails.
buf_page_get_gen(): Use dberr_t for error detection and
do not reference bpage after we hare freed it.
buf_mark_space_corrupt(): remove bpage from LRU also when
it is encrypted.
buf_page_check_corrupt(): @return DB_SUCCESS if page has
been read and is not corrupted,
DB_PAGE_CORRUPTED if page based on checksum check is corrupted,
DB_DECRYPTION_FAILED if page post encryption checksum matches but
after decryption normal page checksum does not match. In read
case only DB_SUCCESS is possible.
buf_page_io_complete(): use dberr_t for error handling.
buf_flush_write_block_low(),
buf_read_ahead_random(),
buf_read_page_async(),
buf_read_ahead_linear(),
buf_read_ibuf_merge_pages(),
buf_read_recv_pages(),
fil_aio_wait():
Issue error if page read fails.
btr_pcur_move_to_next_page(): Do not reference page if it is
NULL.
Introduced dict_table_t::is_readable() and dict_index_t::is_readable()
that will return true if tablespace exists and pages read from
tablespace are not corrupted or page decryption failed.
Removed buf_page_t::key_version. After page decryption the
key version is not removed from page frame. For unencrypted
pages, old key_version is removed at buf_page_encrypt_before_write()
dict_stats_update_transient_for_index(),
dict_stats_update_transient()
Do not continue if table decryption failed or table
is corrupted.
dict0stats.cc: Introduced a dict_stats_report_error function
to avoid code duplication.
fil_parse_write_crypt_data():
Check that key read from redo log entry is found from
encryption plugin and if it is not, refuse to start.
PageConverter::validate(): Removed access to fil_space_t as
tablespace is not available during import.
Fixed error code on innodb.innodb test.
Merged test cased innodb-bad-key-change5 and innodb-bad-key-shutdown
to innodb-bad-key-change2. Removed innodb-bad-key-change5 test.
Decreased unnecessary complexity on some long lasting tests.
Removed fil_inc_pending_ops(), fil_decr_pending_ops(),
fil_get_first_space(), fil_get_next_space(),
fil_get_first_space_safe(), fil_get_next_space_safe()
functions.
fil_space_verify_crypt_checksum(): Fixed bug found using ASAN
where FIL_PAGE_END_LSN_OLD_CHECKSUM field was incorrectly
accessed from row compressed tables. Fixed out of page frame
bug for row compressed tables in
fil_space_verify_crypt_checksum() found using ASAN. Incorrect
function was called for compressed table.
Added new tests for discard, rename table and drop (we should allow them
even when page decryption fails). Alter table rename is not allowed.
Added test for restart with innodb-force-recovery=1 when page read on
redo-recovery cant be decrypted. Added test for corrupted table where
both page data and FIL_PAGE_FILE_FLUSH_LSN_OR_KEY_VERSION is corrupted.
Adjusted the test case innodb_bug14147491 so that it does not anymore
expect crash. Instead table is just mostly not usable.
fil0fil.h: fil_space_acquire_low is not visible function
and fil_space_acquire and fil_space_acquire_silent are
inline functions. FilSpace class uses fil_space_acquire_low
directly.
recv_apply_hashed_log_recs() does not return anything.
9 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-11759: Encryption code in MariaDB 10.1/10.2 causes
compatibility problems
Pages that are encrypted contain post encryption checksum on
different location that normal checksum fields. Therefore,
we should before decryption check this checksum to avoid
unencrypting corrupted pages. After decryption we can use
traditional checksum check to detect if page is corrupted
or unencryption was done using incorrect key.
Pages that are page compressed do not contain any checksum,
here we need to fist unencrypt, decompress and finally
use tradional checksum check to detect page corruption
or that we used incorrect key in unencryption.
buf0buf.cc: buf_page_is_corrupted() mofified so that
compressed pages are skipped.
buf0buf.h, buf_block_init(), buf_page_init_low():
removed unnecessary page_encrypted, page_compressed,
stored_checksum, valculated_checksum fields from
buf_page_t
buf_page_get_gen(): use new buf_page_check_corrupt() function
to detect corrupted pages.
buf_page_check_corrupt(): If page was not yet decrypted
check if post encryption checksum still matches.
If page is not anymore encrypted, use buf_page_is_corrupted()
traditional checksum method.
If page is detected as corrupted and it is not encrypted
we print corruption message to error log.
If page is still encrypted or it was encrypted and now
corrupted, we will print message that page is
encrypted to error log.
buf_page_io_complete(): use new buf_page_check_corrupt()
function to detect corrupted pages.
buf_page_decrypt_after_read(): Verify post encryption
checksum before tring to decrypt.
fil0crypt.cc: fil_encrypt_buf() verify post encryption
checksum and ind fil_space_decrypt() return true
if we really decrypted the page.
fil_space_verify_crypt_checksum(): rewrite to use
the method used when calculating post encryption
checksum. We also check if post encryption checksum
matches that traditional checksum check does not
match.
fil0fil.ic: Add missed page type encrypted and page
compressed to fil_get_page_type_name()
Note that this change does not yet fix innochecksum tool,
that will be done in separate MDEV.
Fix test failures caused by buf page corruption injection.
9 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-12253: Buffer pool blocks are accessed after they have been freed
Problem was that bpage was referenced after it was already freed
from LRU. Fixed by adding a new variable encrypted that is
passed down to buf_page_check_corrupt() and used in
buf_page_get_gen() to stop processing page read.
This patch should also address following test failures and
bugs:
MDEV-12419: IMPORT should not look up tablespace in
PageConverter::validate(). This is now removed.
MDEV-10099: encryption.innodb_onlinealter_encryption fails
sporadically in buildbot
MDEV-11420: encryption.innodb_encryption-page-compression
failed in buildbot
MDEV-11222: encryption.encrypt_and_grep failed in buildbot on P8
Removed dict_table_t::is_encrypted and dict_table_t::ibd_file_missing
and replaced these with dict_table_t::file_unreadable. Table
ibd file is missing if fil_get_space(space_id) returns NULL
and encrypted if not. Removed dict_table_t::is_corrupted field.
Ported FilSpace class from 10.2 and using that on buf_page_check_corrupt(),
buf_page_decrypt_after_read(), buf_page_encrypt_before_write(),
buf_dblwr_process(), buf_read_page(), dict_stats_save_defrag_stats().
Added test cases when enrypted page could be read while doing
redo log crash recovery. Also added test case for row compressed
blobs.
btr_cur_open_at_index_side_func(),
btr_cur_open_at_rnd_pos_func(): Avoid referencing block that is
NULL.
buf_page_get_zip(): Issue error if page read fails.
buf_page_get_gen(): Use dberr_t for error detection and
do not reference bpage after we hare freed it.
buf_mark_space_corrupt(): remove bpage from LRU also when
it is encrypted.
buf_page_check_corrupt(): @return DB_SUCCESS if page has
been read and is not corrupted,
DB_PAGE_CORRUPTED if page based on checksum check is corrupted,
DB_DECRYPTION_FAILED if page post encryption checksum matches but
after decryption normal page checksum does not match. In read
case only DB_SUCCESS is possible.
buf_page_io_complete(): use dberr_t for error handling.
buf_flush_write_block_low(),
buf_read_ahead_random(),
buf_read_page_async(),
buf_read_ahead_linear(),
buf_read_ibuf_merge_pages(),
buf_read_recv_pages(),
fil_aio_wait():
Issue error if page read fails.
btr_pcur_move_to_next_page(): Do not reference page if it is
NULL.
Introduced dict_table_t::is_readable() and dict_index_t::is_readable()
that will return true if tablespace exists and pages read from
tablespace are not corrupted or page decryption failed.
Removed buf_page_t::key_version. After page decryption the
key version is not removed from page frame. For unencrypted
pages, old key_version is removed at buf_page_encrypt_before_write()
dict_stats_update_transient_for_index(),
dict_stats_update_transient()
Do not continue if table decryption failed or table
is corrupted.
dict0stats.cc: Introduced a dict_stats_report_error function
to avoid code duplication.
fil_parse_write_crypt_data():
Check that key read from redo log entry is found from
encryption plugin and if it is not, refuse to start.
PageConverter::validate(): Removed access to fil_space_t as
tablespace is not available during import.
Fixed error code on innodb.innodb test.
Merged test cased innodb-bad-key-change5 and innodb-bad-key-shutdown
to innodb-bad-key-change2. Removed innodb-bad-key-change5 test.
Decreased unnecessary complexity on some long lasting tests.
Removed fil_inc_pending_ops(), fil_decr_pending_ops(),
fil_get_first_space(), fil_get_next_space(),
fil_get_first_space_safe(), fil_get_next_space_safe()
functions.
fil_space_verify_crypt_checksum(): Fixed bug found using ASAN
where FIL_PAGE_END_LSN_OLD_CHECKSUM field was incorrectly
accessed from row compressed tables. Fixed out of page frame
bug for row compressed tables in
fil_space_verify_crypt_checksum() found using ASAN. Incorrect
function was called for compressed table.
Added new tests for discard, rename table and drop (we should allow them
even when page decryption fails). Alter table rename is not allowed.
Added test for restart with innodb-force-recovery=1 when page read on
redo-recovery cant be decrypted. Added test for corrupted table where
both page data and FIL_PAGE_FILE_FLUSH_LSN_OR_KEY_VERSION is corrupted.
Adjusted the test case innodb_bug14147491 so that it does not anymore
expect crash. Instead table is just mostly not usable.
fil0fil.h: fil_space_acquire_low is not visible function
and fil_space_acquire and fil_space_acquire_silent are
inline functions. FilSpace class uses fil_space_acquire_low
directly.
recv_apply_hashed_log_recs() does not return anything.
9 years ago MDEV-11759: Encryption code in MariaDB 10.1/10.2 causes
compatibility problems
Pages that are encrypted contain post encryption checksum on
different location that normal checksum fields. Therefore,
we should before decryption check this checksum to avoid
unencrypting corrupted pages. After decryption we can use
traditional checksum check to detect if page is corrupted
or unencryption was done using incorrect key.
Pages that are page compressed do not contain any checksum,
here we need to fist unencrypt, decompress and finally
use tradional checksum check to detect page corruption
or that we used incorrect key in unencryption.
buf0buf.cc: buf_page_is_corrupted() mofified so that
compressed pages are skipped.
buf0buf.h, buf_block_init(), buf_page_init_low():
removed unnecessary page_encrypted, page_compressed,
stored_checksum, valculated_checksum fields from
buf_page_t
buf_page_get_gen(): use new buf_page_check_corrupt() function
to detect corrupted pages.
buf_page_check_corrupt(): If page was not yet decrypted
check if post encryption checksum still matches.
If page is not anymore encrypted, use buf_page_is_corrupted()
traditional checksum method.
If page is detected as corrupted and it is not encrypted
we print corruption message to error log.
If page is still encrypted or it was encrypted and now
corrupted, we will print message that page is
encrypted to error log.
buf_page_io_complete(): use new buf_page_check_corrupt()
function to detect corrupted pages.
buf_page_decrypt_after_read(): Verify post encryption
checksum before tring to decrypt.
fil0crypt.cc: fil_encrypt_buf() verify post encryption
checksum and ind fil_space_decrypt() return true
if we really decrypted the page.
fil_space_verify_crypt_checksum(): rewrite to use
the method used when calculating post encryption
checksum. We also check if post encryption checksum
matches that traditional checksum check does not
match.
fil0fil.ic: Add missed page type encrypted and page
compressed to fil_get_page_type_name()
Note that this change does not yet fix innochecksum tool,
that will be done in separate MDEV.
Fix test failures caused by buf page corruption injection.
9 years ago MDEV-34830: LSN in the future is not being treated as serious corruption
The invariant of write-ahead logging is that before any change to a
page is written to the data file, the corresponding log record must
must first have been durably written.
In crash recovery, there were some sloppy checks for this. Let us
implement accurate checks and flag an inconsistency as a hard error,
so that we can avoid further corruption of a corrupted database.
For data extraction from the corrupted database, innodb_force_recovery
can be used.
Before recovery is reading any data pages or invoking
buf_dblwr_t::recover() to recover torn pages from the
doublewrite buffer, InnoDB will have parsed the log until the
final LSN and updated log_sys.lsn to that. So, we can rely on
log_sys.lsn at all times. The doublewrite buffer recovery has been
refactored in such a way that the recv_sys.dblwr.pages may be consulted
while discovering files and their page sizes, but nothing will be
written back to data files before buf_dblwr_t::recover() is invoked.
recv_max_page_lsn, recv_lsn_checks_on: Remove.
recv_sys_t::validate_checkpoint(): Validate the write-ahead-logging
condition at the end of the recovery.
recv_dblwr_t::validate_page(): Keep track of the maximum LSN
(if we are checking a non-doublewrite copy of a page) but
do not complain LSN being in the future. The doublewrite buffer
is a special case, because it will be read early during recovery.
Besides, starting with commit 762bcb81b5bf9bbde61fed59afb26417f4ce1e86
the dblwr=true copies of pages may legitimately be "too new".
recv_dblwr_t::find_page(): Find a valid page with the smallest
FIL_PAGE_LSN that is in the valid range for recovery.
recv_dblwr_t::restore_first_page(): Replaced by find_page().
Only buf_dblwr_t::recover() will write to data files.
buf_dblwr_t::recover(): Simplify the message output. Do attempt
doublewrite recovery on user page read error. Ignore doublewrite
pages whose FIL_PAGE_LSN is outside the usable bounds. Previously,
we could wrongly recover a too new page from the doublewrite buffer.
It is unlikely that this could have lead to an actual error.
Write back all recovered pages from the doublewrite buffer here,
including for the first page of any tablespace.
buf_page_is_corrupted(): Distinguish the return values
CORRUPTED_FUTURE_LSN and CORRUPTED_OTHER.
buf_page_check_corrupt(): Return the error code DB_CORRUPTION
in case the LSN is in the future.
Datafile::read_first_page_flags(): Split from read_first_page().
Take a copy of the first page as a parameter.
recv_sys_t::free_corrupted_page(): Take the file as a parameter
and return whether a message was displayed. This avoids some duplicated
and incomplete error messages.
buf_page_t::read_complete(): Remove some redundant output and always
display the name of the corrupted file. Never return DB_FAIL;
use it only in internal error handling.
IORequest::read_complete(): Assume that buf_page_t::read_complete()
will have reported any error.
fil_space_t::set_corrupted(): Return whether this is the first time
the tablespace had been flagged as corrupted.
Datafile::validate_first_page(), fil_node_open_file_low(),
fil_node_open_file(), fil_space_t::read_page0(),
fil_node_t::read_page0(): Add a parameter for a copy of the
first page, and a parameter to indicate whether the FIL_PAGE_LSN
check should be suppressed. Before buf_dblwr_t::recover() is
invoked, we cannot validate the FIL_PAGE_LSN, but we can trust the
FSP_SPACE_FLAGS and the tablespace ID that may be present in a
potentially too new copy of a page.
Reviewed by: Debarun Banerjee
12 months ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-12253: Buffer pool blocks are accessed after they have been freed
Problem was that bpage was referenced after it was already freed
from LRU. Fixed by adding a new variable encrypted that is
passed down to buf_page_check_corrupt() and used in
buf_page_get_gen() to stop processing page read.
This patch should also address following test failures and
bugs:
MDEV-12419: IMPORT should not look up tablespace in
PageConverter::validate(). This is now removed.
MDEV-10099: encryption.innodb_onlinealter_encryption fails
sporadically in buildbot
MDEV-11420: encryption.innodb_encryption-page-compression
failed in buildbot
MDEV-11222: encryption.encrypt_and_grep failed in buildbot on P8
Removed dict_table_t::is_encrypted and dict_table_t::ibd_file_missing
and replaced these with dict_table_t::file_unreadable. Table
ibd file is missing if fil_get_space(space_id) returns NULL
and encrypted if not. Removed dict_table_t::is_corrupted field.
Ported FilSpace class from 10.2 and using that on buf_page_check_corrupt(),
buf_page_decrypt_after_read(), buf_page_encrypt_before_write(),
buf_dblwr_process(), buf_read_page(), dict_stats_save_defrag_stats().
Added test cases when enrypted page could be read while doing
redo log crash recovery. Also added test case for row compressed
blobs.
btr_cur_open_at_index_side_func(),
btr_cur_open_at_rnd_pos_func(): Avoid referencing block that is
NULL.
buf_page_get_zip(): Issue error if page read fails.
buf_page_get_gen(): Use dberr_t for error detection and
do not reference bpage after we hare freed it.
buf_mark_space_corrupt(): remove bpage from LRU also when
it is encrypted.
buf_page_check_corrupt(): @return DB_SUCCESS if page has
been read and is not corrupted,
DB_PAGE_CORRUPTED if page based on checksum check is corrupted,
DB_DECRYPTION_FAILED if page post encryption checksum matches but
after decryption normal page checksum does not match. In read
case only DB_SUCCESS is possible.
buf_page_io_complete(): use dberr_t for error handling.
buf_flush_write_block_low(),
buf_read_ahead_random(),
buf_read_page_async(),
buf_read_ahead_linear(),
buf_read_ibuf_merge_pages(),
buf_read_recv_pages(),
fil_aio_wait():
Issue error if page read fails.
btr_pcur_move_to_next_page(): Do not reference page if it is
NULL.
Introduced dict_table_t::is_readable() and dict_index_t::is_readable()
that will return true if tablespace exists and pages read from
tablespace are not corrupted or page decryption failed.
Removed buf_page_t::key_version. After page decryption the
key version is not removed from page frame. For unencrypted
pages, old key_version is removed at buf_page_encrypt_before_write()
dict_stats_update_transient_for_index(),
dict_stats_update_transient()
Do not continue if table decryption failed or table
is corrupted.
dict0stats.cc: Introduced a dict_stats_report_error function
to avoid code duplication.
fil_parse_write_crypt_data():
Check that key read from redo log entry is found from
encryption plugin and if it is not, refuse to start.
PageConverter::validate(): Removed access to fil_space_t as
tablespace is not available during import.
Fixed error code on innodb.innodb test.
Merged test cased innodb-bad-key-change5 and innodb-bad-key-shutdown
to innodb-bad-key-change2. Removed innodb-bad-key-change5 test.
Decreased unnecessary complexity on some long lasting tests.
Removed fil_inc_pending_ops(), fil_decr_pending_ops(),
fil_get_first_space(), fil_get_next_space(),
fil_get_first_space_safe(), fil_get_next_space_safe()
functions.
fil_space_verify_crypt_checksum(): Fixed bug found using ASAN
where FIL_PAGE_END_LSN_OLD_CHECKSUM field was incorrectly
accessed from row compressed tables. Fixed out of page frame
bug for row compressed tables in
fil_space_verify_crypt_checksum() found using ASAN. Incorrect
function was called for compressed table.
Added new tests for discard, rename table and drop (we should allow them
even when page decryption fails). Alter table rename is not allowed.
Added test for restart with innodb-force-recovery=1 when page read on
redo-recovery cant be decrypted. Added test for corrupted table where
both page data and FIL_PAGE_FILE_FLUSH_LSN_OR_KEY_VERSION is corrupted.
Adjusted the test case innodb_bug14147491 so that it does not anymore
expect crash. Instead table is just mostly not usable.
fil0fil.h: fil_space_acquire_low is not visible function
and fil_space_acquire and fil_space_acquire_silent are
inline functions. FilSpace class uses fil_space_acquire_low
directly.
recv_apply_hashed_log_recs() does not return anything.
9 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-13542: Crashing on corrupted page is unhelpful
The approach to handling corruption that was chosen by Oracle in
commit 177d8b0c125b841c0650d27d735e3b87509dc286
is not really useful. Not only did it actually fail to prevent InnoDB
from crashing, but it is making things worse by blocking attempts to
rescue data from or rebuild a partially readable table.
We will try to prevent crashes in a different way: by propagating
errors up the call stack. We will never mark the clustered index
persistently corrupted, so that data recovery may be attempted by
reading from the table, or by rebuilding the table.
This should also fix MDEV-13680 (crash on btr_page_alloc() failure);
it was extensively tested with innodb_file_per_table=0 and a
non-autoextend system tablespace.
We should now avoid crashes in many cases, such as when a page
cannot be read or allocated, or an inconsistency is detected when
attempting to update multiple pages. We will not crash on double-free,
such as on the recovery of DDL in system tablespace in case something
was corrupted.
Crashes on corrupted data are still possible. The fault injection mechanism
that is introduced in the subsequent commit may help catch more of them.
buf_page_import_corrupt_failure: Remove the fault injection, and instead
corrupt some pages using Perl code in the tests.
btr_cur_pessimistic_insert(): Always reserve extents (except for the
change buffer), in order to prevent a subsequent allocation failure.
btr_pcur_open_at_rnd_pos(): Merged to the only caller ibuf_merge_pages().
btr_assert_not_corrupted(), btr_corruption_report(): Remove.
Similar checks are already part of btr_block_get().
FSEG_MAGIC_N_BYTES: Replaces FSEG_MAGIC_N_VALUE.
dict_hdr_get(), trx_rsegf_get_new(), trx_undo_page_get(),
trx_undo_page_get_s_latched(): Replaced with error-checking calls.
trx_rseg_t::get(mtr_t*): Replaces trx_rsegf_get().
trx_rseg_header_create(): Let the caller update the TRX_SYS page if needed.
trx_sys_create_sys_pages(): Merged with trx_sysf_create().
dict_check_tablespaces_and_store_max_id(): Do not access
DICT_HDR_MAX_SPACE_ID, because it was already recovered in dict_boot().
Merge dict_check_sys_tables() with this function.
dir_pathname(): Replaces os_file_make_new_pathname().
row_undo_ins_remove_sec(): Do not modify the undo page by adding
a terminating NUL byte to the record.
btr_decryption_failed(): Report decryption failures
dict_set_corrupted_by_space(), dict_set_encrypted_by_space(),
dict_set_corrupted_index_cache_only(): Remove.
dict_set_corrupted(): Remove the constant parameter dict_locked=false.
Never flag the clustered index corrupted in SYS_INDEXES, because
that would deny further access to the table. It might be possible to
repair the table by executing ALTER TABLE or OPTIMIZE TABLE, in case
no B-tree leaf page is corrupted.
dict_table_skip_corrupt_index(), dict_table_next_uncorrupted_index(),
row_purge_skip_uncommitted_virtual_index(): Remove, and refactor
the callers to read dict_index_t::type only once.
dict_table_is_corrupted(): Remove.
dict_index_t::is_btree(): Determine if the index is a valid B-tree.
BUF_GET_NO_LATCH, BUF_EVICT_IF_IN_POOL: Remove.
UNIV_BTR_DEBUG: Remove. Any inconsistency will no longer trigger
assertion failures, but error codes being returned.
buf_corrupt_page_release(): Replaced with a direct call to
buf_pool.corrupted_evict().
fil_invalid_page_access_msg(): Never crash on an invalid read;
let the caller of buf_page_get_gen() decide.
btr_pcur_t::restore_position(): Propagate failure status to the caller
by returning CORRUPTED.
opt_search_plan_for_table(): Simplify the code.
row_purge_del_mark(), row_purge_upd_exist_or_extern_func(),
row_undo_ins_remove_sec_rec(), row_undo_mod_upd_del_sec(),
row_undo_mod_del_mark_sec(): Avoid mem_heap_create()/mem_heap_free()
when no secondary indexes exist.
row_undo_mod_upd_exist_sec(): Simplify the code.
row_upd_clust_step(), dict_load_table_one(): Return DB_TABLE_CORRUPT
if the clustered index (and therefore the table) is corrupted, similar
to what we do in row_insert_for_mysql().
fut_get_ptr(): Replace with buf_page_get_gen() calls.
buf_page_get_gen(): Return nullptr and *err=DB_CORRUPTION
if the page is marked as freed. For other modes than
BUF_GET_POSSIBLY_FREED or BUF_PEEK_IF_IN_POOL this will
trigger a debug assertion failure. For BUF_GET_POSSIBLY_FREED,
we will return nullptr for freed pages, so that the callers
can be simplified. The purge of transaction history will be
a new user of BUF_GET_POSSIBLY_FREED, to avoid crashes on
corrupted data.
buf_page_get_low(): Never crash on a corrupted page, but simply
return nullptr.
fseg_page_is_allocated(): Replaces fseg_page_is_free().
fts_drop_common_tables(): Return an error if the transaction
was rolled back.
fil_space_t::set_corrupted(): Report a tablespace as corrupted if
it was not reported already.
fil_space_t::io(): Invoke fil_space_t::set_corrupted() to report
out-of-bounds page access or other errors.
Clean up mtr_t::page_lock()
buf_page_get_low(): Validate the page identifier (to check for
recently read corrupted pages) after acquiring the page latch.
buf_page_t::read_complete(): Flag uninitialized (all-zero) pages
with DB_FAIL. Return DB_PAGE_CORRUPTED on page number mismatch.
mtr_t::defer_drop_ahi(): Renamed from mtr_defer_drop_ahi().
recv_sys_t::free_corrupted_page(): Only set_corrupt_fs()
if any log records exist for the page. We do not mind if read-ahead
produces corrupted (or all-zero) pages that were not actually needed
during recovery.
recv_recover_page(): Return whether the operation succeeded.
recv_sys_t::recover_low(): Simplify the logic. Check for recovery error.
Thanks to Matthias Leich for testing this extensively and to the
authors of https://rr-project.org for making it easy to diagnose
and fix any failures that were found during the testing.
3 years ago MDEV-12253: Buffer pool blocks are accessed after they have been freed
Problem was that bpage was referenced after it was already freed
from LRU. Fixed by adding a new variable encrypted that is
passed down to buf_page_check_corrupt() and used in
buf_page_get_gen() to stop processing page read.
This patch should also address following test failures and
bugs:
MDEV-12419: IMPORT should not look up tablespace in
PageConverter::validate(). This is now removed.
MDEV-10099: encryption.innodb_onlinealter_encryption fails
sporadically in buildbot
MDEV-11420: encryption.innodb_encryption-page-compression
failed in buildbot
MDEV-11222: encryption.encrypt_and_grep failed in buildbot on P8
Removed dict_table_t::is_encrypted and dict_table_t::ibd_file_missing
and replaced these with dict_table_t::file_unreadable. Table
ibd file is missing if fil_get_space(space_id) returns NULL
and encrypted if not. Removed dict_table_t::is_corrupted field.
Ported FilSpace class from 10.2 and using that on buf_page_check_corrupt(),
buf_page_decrypt_after_read(), buf_page_encrypt_before_write(),
buf_dblwr_process(), buf_read_page(), dict_stats_save_defrag_stats().
Added test cases when enrypted page could be read while doing
redo log crash recovery. Also added test case for row compressed
blobs.
btr_cur_open_at_index_side_func(),
btr_cur_open_at_rnd_pos_func(): Avoid referencing block that is
NULL.
buf_page_get_zip(): Issue error if page read fails.
buf_page_get_gen(): Use dberr_t for error detection and
do not reference bpage after we hare freed it.
buf_mark_space_corrupt(): remove bpage from LRU also when
it is encrypted.
buf_page_check_corrupt(): @return DB_SUCCESS if page has
been read and is not corrupted,
DB_PAGE_CORRUPTED if page based on checksum check is corrupted,
DB_DECRYPTION_FAILED if page post encryption checksum matches but
after decryption normal page checksum does not match. In read
case only DB_SUCCESS is possible.
buf_page_io_complete(): use dberr_t for error handling.
buf_flush_write_block_low(),
buf_read_ahead_random(),
buf_read_page_async(),
buf_read_ahead_linear(),
buf_read_ibuf_merge_pages(),
buf_read_recv_pages(),
fil_aio_wait():
Issue error if page read fails.
btr_pcur_move_to_next_page(): Do not reference page if it is
NULL.
Introduced dict_table_t::is_readable() and dict_index_t::is_readable()
that will return true if tablespace exists and pages read from
tablespace are not corrupted or page decryption failed.
Removed buf_page_t::key_version. After page decryption the
key version is not removed from page frame. For unencrypted
pages, old key_version is removed at buf_page_encrypt_before_write()
dict_stats_update_transient_for_index(),
dict_stats_update_transient()
Do not continue if table decryption failed or table
is corrupted.
dict0stats.cc: Introduced a dict_stats_report_error function
to avoid code duplication.
fil_parse_write_crypt_data():
Check that key read from redo log entry is found from
encryption plugin and if it is not, refuse to start.
PageConverter::validate(): Removed access to fil_space_t as
tablespace is not available during import.
Fixed error code on innodb.innodb test.
Merged test cased innodb-bad-key-change5 and innodb-bad-key-shutdown
to innodb-bad-key-change2. Removed innodb-bad-key-change5 test.
Decreased unnecessary complexity on some long lasting tests.
Removed fil_inc_pending_ops(), fil_decr_pending_ops(),
fil_get_first_space(), fil_get_next_space(),
fil_get_first_space_safe(), fil_get_next_space_safe()
functions.
fil_space_verify_crypt_checksum(): Fixed bug found using ASAN
where FIL_PAGE_END_LSN_OLD_CHECKSUM field was incorrectly
accessed from row compressed tables. Fixed out of page frame
bug for row compressed tables in
fil_space_verify_crypt_checksum() found using ASAN. Incorrect
function was called for compressed table.
Added new tests for discard, rename table and drop (we should allow them
even when page decryption fails). Alter table rename is not allowed.
Added test for restart with innodb-force-recovery=1 when page read on
redo-recovery cant be decrypted. Added test for corrupted table where
both page data and FIL_PAGE_FILE_FLUSH_LSN_OR_KEY_VERSION is corrupted.
Adjusted the test case innodb_bug14147491 so that it does not anymore
expect crash. Instead table is just mostly not usable.
fil0fil.h: fil_space_acquire_low is not visible function
and fil_space_acquire and fil_space_acquire_silent are
inline functions. FilSpace class uses fil_space_acquire_low
directly.
recv_apply_hashed_log_recs() does not return anything.
9 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-34830: LSN in the future is not being treated as serious corruption
The invariant of write-ahead logging is that before any change to a
page is written to the data file, the corresponding log record must
must first have been durably written.
In crash recovery, there were some sloppy checks for this. Let us
implement accurate checks and flag an inconsistency as a hard error,
so that we can avoid further corruption of a corrupted database.
For data extraction from the corrupted database, innodb_force_recovery
can be used.
Before recovery is reading any data pages or invoking
buf_dblwr_t::recover() to recover torn pages from the
doublewrite buffer, InnoDB will have parsed the log until the
final LSN and updated log_sys.lsn to that. So, we can rely on
log_sys.lsn at all times. The doublewrite buffer recovery has been
refactored in such a way that the recv_sys.dblwr.pages may be consulted
while discovering files and their page sizes, but nothing will be
written back to data files before buf_dblwr_t::recover() is invoked.
recv_max_page_lsn, recv_lsn_checks_on: Remove.
recv_sys_t::validate_checkpoint(): Validate the write-ahead-logging
condition at the end of the recovery.
recv_dblwr_t::validate_page(): Keep track of the maximum LSN
(if we are checking a non-doublewrite copy of a page) but
do not complain LSN being in the future. The doublewrite buffer
is a special case, because it will be read early during recovery.
Besides, starting with commit 762bcb81b5bf9bbde61fed59afb26417f4ce1e86
the dblwr=true copies of pages may legitimately be "too new".
recv_dblwr_t::find_page(): Find a valid page with the smallest
FIL_PAGE_LSN that is in the valid range for recovery.
recv_dblwr_t::restore_first_page(): Replaced by find_page().
Only buf_dblwr_t::recover() will write to data files.
buf_dblwr_t::recover(): Simplify the message output. Do attempt
doublewrite recovery on user page read error. Ignore doublewrite
pages whose FIL_PAGE_LSN is outside the usable bounds. Previously,
we could wrongly recover a too new page from the doublewrite buffer.
It is unlikely that this could have lead to an actual error.
Write back all recovered pages from the doublewrite buffer here,
including for the first page of any tablespace.
buf_page_is_corrupted(): Distinguish the return values
CORRUPTED_FUTURE_LSN and CORRUPTED_OTHER.
buf_page_check_corrupt(): Return the error code DB_CORRUPTION
in case the LSN is in the future.
Datafile::read_first_page_flags(): Split from read_first_page().
Take a copy of the first page as a parameter.
recv_sys_t::free_corrupted_page(): Take the file as a parameter
and return whether a message was displayed. This avoids some duplicated
and incomplete error messages.
buf_page_t::read_complete(): Remove some redundant output and always
display the name of the corrupted file. Never return DB_FAIL;
use it only in internal error handling.
IORequest::read_complete(): Assume that buf_page_t::read_complete()
will have reported any error.
fil_space_t::set_corrupted(): Return whether this is the first time
the tablespace had been flagged as corrupted.
Datafile::validate_first_page(), fil_node_open_file_low(),
fil_node_open_file(), fil_space_t::read_page0(),
fil_node_t::read_page0(): Add a parameter for a copy of the
first page, and a parameter to indicate whether the FIL_PAGE_LSN
check should be suppressed. Before buf_dblwr_t::recover() is
invoked, we cannot validate the FIL_PAGE_LSN, but we can trust the
FSP_SPACE_FLAGS and the tablespace ID that may be present in a
potentially too new copy of a page.
Reviewed by: Debarun Banerjee
12 months ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-34830: LSN in the future is not being treated as serious corruption
The invariant of write-ahead logging is that before any change to a
page is written to the data file, the corresponding log record must
must first have been durably written.
In crash recovery, there were some sloppy checks for this. Let us
implement accurate checks and flag an inconsistency as a hard error,
so that we can avoid further corruption of a corrupted database.
For data extraction from the corrupted database, innodb_force_recovery
can be used.
Before recovery is reading any data pages or invoking
buf_dblwr_t::recover() to recover torn pages from the
doublewrite buffer, InnoDB will have parsed the log until the
final LSN and updated log_sys.lsn to that. So, we can rely on
log_sys.lsn at all times. The doublewrite buffer recovery has been
refactored in such a way that the recv_sys.dblwr.pages may be consulted
while discovering files and their page sizes, but nothing will be
written back to data files before buf_dblwr_t::recover() is invoked.
recv_max_page_lsn, recv_lsn_checks_on: Remove.
recv_sys_t::validate_checkpoint(): Validate the write-ahead-logging
condition at the end of the recovery.
recv_dblwr_t::validate_page(): Keep track of the maximum LSN
(if we are checking a non-doublewrite copy of a page) but
do not complain LSN being in the future. The doublewrite buffer
is a special case, because it will be read early during recovery.
Besides, starting with commit 762bcb81b5bf9bbde61fed59afb26417f4ce1e86
the dblwr=true copies of pages may legitimately be "too new".
recv_dblwr_t::find_page(): Find a valid page with the smallest
FIL_PAGE_LSN that is in the valid range for recovery.
recv_dblwr_t::restore_first_page(): Replaced by find_page().
Only buf_dblwr_t::recover() will write to data files.
buf_dblwr_t::recover(): Simplify the message output. Do attempt
doublewrite recovery on user page read error. Ignore doublewrite
pages whose FIL_PAGE_LSN is outside the usable bounds. Previously,
we could wrongly recover a too new page from the doublewrite buffer.
It is unlikely that this could have lead to an actual error.
Write back all recovered pages from the doublewrite buffer here,
including for the first page of any tablespace.
buf_page_is_corrupted(): Distinguish the return values
CORRUPTED_FUTURE_LSN and CORRUPTED_OTHER.
buf_page_check_corrupt(): Return the error code DB_CORRUPTION
in case the LSN is in the future.
Datafile::read_first_page_flags(): Split from read_first_page().
Take a copy of the first page as a parameter.
recv_sys_t::free_corrupted_page(): Take the file as a parameter
and return whether a message was displayed. This avoids some duplicated
and incomplete error messages.
buf_page_t::read_complete(): Remove some redundant output and always
display the name of the corrupted file. Never return DB_FAIL;
use it only in internal error handling.
IORequest::read_complete(): Assume that buf_page_t::read_complete()
will have reported any error.
fil_space_t::set_corrupted(): Return whether this is the first time
the tablespace had been flagged as corrupted.
Datafile::validate_first_page(), fil_node_open_file_low(),
fil_node_open_file(), fil_space_t::read_page0(),
fil_node_t::read_page0(): Add a parameter for a copy of the
first page, and a parameter to indicate whether the FIL_PAGE_LSN
check should be suppressed. Before buf_dblwr_t::recover() is
invoked, we cannot validate the FIL_PAGE_LSN, but we can trust the
FSP_SPACE_FLAGS and the tablespace ID that may be present in a
potentially too new copy of a page.
Reviewed by: Debarun Banerjee
12 months ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-13542: Crashing on corrupted page is unhelpful
The approach to handling corruption that was chosen by Oracle in
commit 177d8b0c125b841c0650d27d735e3b87509dc286
is not really useful. Not only did it actually fail to prevent InnoDB
from crashing, but it is making things worse by blocking attempts to
rescue data from or rebuild a partially readable table.
We will try to prevent crashes in a different way: by propagating
errors up the call stack. We will never mark the clustered index
persistently corrupted, so that data recovery may be attempted by
reading from the table, or by rebuilding the table.
This should also fix MDEV-13680 (crash on btr_page_alloc() failure);
it was extensively tested with innodb_file_per_table=0 and a
non-autoextend system tablespace.
We should now avoid crashes in many cases, such as when a page
cannot be read or allocated, or an inconsistency is detected when
attempting to update multiple pages. We will not crash on double-free,
such as on the recovery of DDL in system tablespace in case something
was corrupted.
Crashes on corrupted data are still possible. The fault injection mechanism
that is introduced in the subsequent commit may help catch more of them.
buf_page_import_corrupt_failure: Remove the fault injection, and instead
corrupt some pages using Perl code in the tests.
btr_cur_pessimistic_insert(): Always reserve extents (except for the
change buffer), in order to prevent a subsequent allocation failure.
btr_pcur_open_at_rnd_pos(): Merged to the only caller ibuf_merge_pages().
btr_assert_not_corrupted(), btr_corruption_report(): Remove.
Similar checks are already part of btr_block_get().
FSEG_MAGIC_N_BYTES: Replaces FSEG_MAGIC_N_VALUE.
dict_hdr_get(), trx_rsegf_get_new(), trx_undo_page_get(),
trx_undo_page_get_s_latched(): Replaced with error-checking calls.
trx_rseg_t::get(mtr_t*): Replaces trx_rsegf_get().
trx_rseg_header_create(): Let the caller update the TRX_SYS page if needed.
trx_sys_create_sys_pages(): Merged with trx_sysf_create().
dict_check_tablespaces_and_store_max_id(): Do not access
DICT_HDR_MAX_SPACE_ID, because it was already recovered in dict_boot().
Merge dict_check_sys_tables() with this function.
dir_pathname(): Replaces os_file_make_new_pathname().
row_undo_ins_remove_sec(): Do not modify the undo page by adding
a terminating NUL byte to the record.
btr_decryption_failed(): Report decryption failures
dict_set_corrupted_by_space(), dict_set_encrypted_by_space(),
dict_set_corrupted_index_cache_only(): Remove.
dict_set_corrupted(): Remove the constant parameter dict_locked=false.
Never flag the clustered index corrupted in SYS_INDEXES, because
that would deny further access to the table. It might be possible to
repair the table by executing ALTER TABLE or OPTIMIZE TABLE, in case
no B-tree leaf page is corrupted.
dict_table_skip_corrupt_index(), dict_table_next_uncorrupted_index(),
row_purge_skip_uncommitted_virtual_index(): Remove, and refactor
the callers to read dict_index_t::type only once.
dict_table_is_corrupted(): Remove.
dict_index_t::is_btree(): Determine if the index is a valid B-tree.
BUF_GET_NO_LATCH, BUF_EVICT_IF_IN_POOL: Remove.
UNIV_BTR_DEBUG: Remove. Any inconsistency will no longer trigger
assertion failures, but error codes being returned.
buf_corrupt_page_release(): Replaced with a direct call to
buf_pool.corrupted_evict().
fil_invalid_page_access_msg(): Never crash on an invalid read;
let the caller of buf_page_get_gen() decide.
btr_pcur_t::restore_position(): Propagate failure status to the caller
by returning CORRUPTED.
opt_search_plan_for_table(): Simplify the code.
row_purge_del_mark(), row_purge_upd_exist_or_extern_func(),
row_undo_ins_remove_sec_rec(), row_undo_mod_upd_del_sec(),
row_undo_mod_del_mark_sec(): Avoid mem_heap_create()/mem_heap_free()
when no secondary indexes exist.
row_undo_mod_upd_exist_sec(): Simplify the code.
row_upd_clust_step(), dict_load_table_one(): Return DB_TABLE_CORRUPT
if the clustered index (and therefore the table) is corrupted, similar
to what we do in row_insert_for_mysql().
fut_get_ptr(): Replace with buf_page_get_gen() calls.
buf_page_get_gen(): Return nullptr and *err=DB_CORRUPTION
if the page is marked as freed. For other modes than
BUF_GET_POSSIBLY_FREED or BUF_PEEK_IF_IN_POOL this will
trigger a debug assertion failure. For BUF_GET_POSSIBLY_FREED,
we will return nullptr for freed pages, so that the callers
can be simplified. The purge of transaction history will be
a new user of BUF_GET_POSSIBLY_FREED, to avoid crashes on
corrupted data.
buf_page_get_low(): Never crash on a corrupted page, but simply
return nullptr.
fseg_page_is_allocated(): Replaces fseg_page_is_free().
fts_drop_common_tables(): Return an error if the transaction
was rolled back.
fil_space_t::set_corrupted(): Report a tablespace as corrupted if
it was not reported already.
fil_space_t::io(): Invoke fil_space_t::set_corrupted() to report
out-of-bounds page access or other errors.
Clean up mtr_t::page_lock()
buf_page_get_low(): Validate the page identifier (to check for
recently read corrupted pages) after acquiring the page latch.
buf_page_t::read_complete(): Flag uninitialized (all-zero) pages
with DB_FAIL. Return DB_PAGE_CORRUPTED on page number mismatch.
mtr_t::defer_drop_ahi(): Renamed from mtr_defer_drop_ahi().
recv_sys_t::free_corrupted_page(): Only set_corrupt_fs()
if any log records exist for the page. We do not mind if read-ahead
produces corrupted (or all-zero) pages that were not actually needed
during recovery.
recv_recover_page(): Return whether the operation succeeded.
recv_sys_t::recover_low(): Simplify the logic. Check for recovery error.
Thanks to Matthias Leich for testing this extensively and to the
authors of https://rr-project.org for making it easy to diagnose
and fix any failures that were found during the testing.
3 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-13542: Crashing on corrupted page is unhelpful
The approach to handling corruption that was chosen by Oracle in
commit 177d8b0c125b841c0650d27d735e3b87509dc286
is not really useful. Not only did it actually fail to prevent InnoDB
from crashing, but it is making things worse by blocking attempts to
rescue data from or rebuild a partially readable table.
We will try to prevent crashes in a different way: by propagating
errors up the call stack. We will never mark the clustered index
persistently corrupted, so that data recovery may be attempted by
reading from the table, or by rebuilding the table.
This should also fix MDEV-13680 (crash on btr_page_alloc() failure);
it was extensively tested with innodb_file_per_table=0 and a
non-autoextend system tablespace.
We should now avoid crashes in many cases, such as when a page
cannot be read or allocated, or an inconsistency is detected when
attempting to update multiple pages. We will not crash on double-free,
such as on the recovery of DDL in system tablespace in case something
was corrupted.
Crashes on corrupted data are still possible. The fault injection mechanism
that is introduced in the subsequent commit may help catch more of them.
buf_page_import_corrupt_failure: Remove the fault injection, and instead
corrupt some pages using Perl code in the tests.
btr_cur_pessimistic_insert(): Always reserve extents (except for the
change buffer), in order to prevent a subsequent allocation failure.
btr_pcur_open_at_rnd_pos(): Merged to the only caller ibuf_merge_pages().
btr_assert_not_corrupted(), btr_corruption_report(): Remove.
Similar checks are already part of btr_block_get().
FSEG_MAGIC_N_BYTES: Replaces FSEG_MAGIC_N_VALUE.
dict_hdr_get(), trx_rsegf_get_new(), trx_undo_page_get(),
trx_undo_page_get_s_latched(): Replaced with error-checking calls.
trx_rseg_t::get(mtr_t*): Replaces trx_rsegf_get().
trx_rseg_header_create(): Let the caller update the TRX_SYS page if needed.
trx_sys_create_sys_pages(): Merged with trx_sysf_create().
dict_check_tablespaces_and_store_max_id(): Do not access
DICT_HDR_MAX_SPACE_ID, because it was already recovered in dict_boot().
Merge dict_check_sys_tables() with this function.
dir_pathname(): Replaces os_file_make_new_pathname().
row_undo_ins_remove_sec(): Do not modify the undo page by adding
a terminating NUL byte to the record.
btr_decryption_failed(): Report decryption failures
dict_set_corrupted_by_space(), dict_set_encrypted_by_space(),
dict_set_corrupted_index_cache_only(): Remove.
dict_set_corrupted(): Remove the constant parameter dict_locked=false.
Never flag the clustered index corrupted in SYS_INDEXES, because
that would deny further access to the table. It might be possible to
repair the table by executing ALTER TABLE or OPTIMIZE TABLE, in case
no B-tree leaf page is corrupted.
dict_table_skip_corrupt_index(), dict_table_next_uncorrupted_index(),
row_purge_skip_uncommitted_virtual_index(): Remove, and refactor
the callers to read dict_index_t::type only once.
dict_table_is_corrupted(): Remove.
dict_index_t::is_btree(): Determine if the index is a valid B-tree.
BUF_GET_NO_LATCH, BUF_EVICT_IF_IN_POOL: Remove.
UNIV_BTR_DEBUG: Remove. Any inconsistency will no longer trigger
assertion failures, but error codes being returned.
buf_corrupt_page_release(): Replaced with a direct call to
buf_pool.corrupted_evict().
fil_invalid_page_access_msg(): Never crash on an invalid read;
let the caller of buf_page_get_gen() decide.
btr_pcur_t::restore_position(): Propagate failure status to the caller
by returning CORRUPTED.
opt_search_plan_for_table(): Simplify the code.
row_purge_del_mark(), row_purge_upd_exist_or_extern_func(),
row_undo_ins_remove_sec_rec(), row_undo_mod_upd_del_sec(),
row_undo_mod_del_mark_sec(): Avoid mem_heap_create()/mem_heap_free()
when no secondary indexes exist.
row_undo_mod_upd_exist_sec(): Simplify the code.
row_upd_clust_step(), dict_load_table_one(): Return DB_TABLE_CORRUPT
if the clustered index (and therefore the table) is corrupted, similar
to what we do in row_insert_for_mysql().
fut_get_ptr(): Replace with buf_page_get_gen() calls.
buf_page_get_gen(): Return nullptr and *err=DB_CORRUPTION
if the page is marked as freed. For other modes than
BUF_GET_POSSIBLY_FREED or BUF_PEEK_IF_IN_POOL this will
trigger a debug assertion failure. For BUF_GET_POSSIBLY_FREED,
we will return nullptr for freed pages, so that the callers
can be simplified. The purge of transaction history will be
a new user of BUF_GET_POSSIBLY_FREED, to avoid crashes on
corrupted data.
buf_page_get_low(): Never crash on a corrupted page, but simply
return nullptr.
fseg_page_is_allocated(): Replaces fseg_page_is_free().
fts_drop_common_tables(): Return an error if the transaction
was rolled back.
fil_space_t::set_corrupted(): Report a tablespace as corrupted if
it was not reported already.
fil_space_t::io(): Invoke fil_space_t::set_corrupted() to report
out-of-bounds page access or other errors.
Clean up mtr_t::page_lock()
buf_page_get_low(): Validate the page identifier (to check for
recently read corrupted pages) after acquiring the page latch.
buf_page_t::read_complete(): Flag uninitialized (all-zero) pages
with DB_FAIL. Return DB_PAGE_CORRUPTED on page number mismatch.
mtr_t::defer_drop_ahi(): Renamed from mtr_defer_drop_ahi().
recv_sys_t::free_corrupted_page(): Only set_corrupt_fs()
if any log records exist for the page. We do not mind if read-ahead
produces corrupted (or all-zero) pages that were not actually needed
during recovery.
recv_recover_page(): Return whether the operation succeeded.
recv_sys_t::recover_low(): Simplify the logic. Check for recovery error.
Thanks to Matthias Leich for testing this extensively and to the
authors of https://rr-project.org for making it easy to diagnose
and fix any failures that were found during the testing.
3 years ago MDEV-34830: LSN in the future is not being treated as serious corruption
The invariant of write-ahead logging is that before any change to a
page is written to the data file, the corresponding log record must
must first have been durably written.
In crash recovery, there were some sloppy checks for this. Let us
implement accurate checks and flag an inconsistency as a hard error,
so that we can avoid further corruption of a corrupted database.
For data extraction from the corrupted database, innodb_force_recovery
can be used.
Before recovery is reading any data pages or invoking
buf_dblwr_t::recover() to recover torn pages from the
doublewrite buffer, InnoDB will have parsed the log until the
final LSN and updated log_sys.lsn to that. So, we can rely on
log_sys.lsn at all times. The doublewrite buffer recovery has been
refactored in such a way that the recv_sys.dblwr.pages may be consulted
while discovering files and their page sizes, but nothing will be
written back to data files before buf_dblwr_t::recover() is invoked.
recv_max_page_lsn, recv_lsn_checks_on: Remove.
recv_sys_t::validate_checkpoint(): Validate the write-ahead-logging
condition at the end of the recovery.
recv_dblwr_t::validate_page(): Keep track of the maximum LSN
(if we are checking a non-doublewrite copy of a page) but
do not complain LSN being in the future. The doublewrite buffer
is a special case, because it will be read early during recovery.
Besides, starting with commit 762bcb81b5bf9bbde61fed59afb26417f4ce1e86
the dblwr=true copies of pages may legitimately be "too new".
recv_dblwr_t::find_page(): Find a valid page with the smallest
FIL_PAGE_LSN that is in the valid range for recovery.
recv_dblwr_t::restore_first_page(): Replaced by find_page().
Only buf_dblwr_t::recover() will write to data files.
buf_dblwr_t::recover(): Simplify the message output. Do attempt
doublewrite recovery on user page read error. Ignore doublewrite
pages whose FIL_PAGE_LSN is outside the usable bounds. Previously,
we could wrongly recover a too new page from the doublewrite buffer.
It is unlikely that this could have lead to an actual error.
Write back all recovered pages from the doublewrite buffer here,
including for the first page of any tablespace.
buf_page_is_corrupted(): Distinguish the return values
CORRUPTED_FUTURE_LSN and CORRUPTED_OTHER.
buf_page_check_corrupt(): Return the error code DB_CORRUPTION
in case the LSN is in the future.
Datafile::read_first_page_flags(): Split from read_first_page().
Take a copy of the first page as a parameter.
recv_sys_t::free_corrupted_page(): Take the file as a parameter
and return whether a message was displayed. This avoids some duplicated
and incomplete error messages.
buf_page_t::read_complete(): Remove some redundant output and always
display the name of the corrupted file. Never return DB_FAIL;
use it only in internal error handling.
IORequest::read_complete(): Assume that buf_page_t::read_complete()
will have reported any error.
fil_space_t::set_corrupted(): Return whether this is the first time
the tablespace had been flagged as corrupted.
Datafile::validate_first_page(), fil_node_open_file_low(),
fil_node_open_file(), fil_space_t::read_page0(),
fil_node_t::read_page0(): Add a parameter for a copy of the
first page, and a parameter to indicate whether the FIL_PAGE_LSN
check should be suppressed. Before buf_dblwr_t::recover() is
invoked, we cannot validate the FIL_PAGE_LSN, but we can trust the
FSP_SPACE_FLAGS and the tablespace ID that may be present in a
potentially too new copy of a page.
Reviewed by: Debarun Banerjee
12 months ago MDEV-13542: Crashing on corrupted page is unhelpful
The approach to handling corruption that was chosen by Oracle in
commit 177d8b0c125b841c0650d27d735e3b87509dc286
is not really useful. Not only did it actually fail to prevent InnoDB
from crashing, but it is making things worse by blocking attempts to
rescue data from or rebuild a partially readable table.
We will try to prevent crashes in a different way: by propagating
errors up the call stack. We will never mark the clustered index
persistently corrupted, so that data recovery may be attempted by
reading from the table, or by rebuilding the table.
This should also fix MDEV-13680 (crash on btr_page_alloc() failure);
it was extensively tested with innodb_file_per_table=0 and a
non-autoextend system tablespace.
We should now avoid crashes in many cases, such as when a page
cannot be read or allocated, or an inconsistency is detected when
attempting to update multiple pages. We will not crash on double-free,
such as on the recovery of DDL in system tablespace in case something
was corrupted.
Crashes on corrupted data are still possible. The fault injection mechanism
that is introduced in the subsequent commit may help catch more of them.
buf_page_import_corrupt_failure: Remove the fault injection, and instead
corrupt some pages using Perl code in the tests.
btr_cur_pessimistic_insert(): Always reserve extents (except for the
change buffer), in order to prevent a subsequent allocation failure.
btr_pcur_open_at_rnd_pos(): Merged to the only caller ibuf_merge_pages().
btr_assert_not_corrupted(), btr_corruption_report(): Remove.
Similar checks are already part of btr_block_get().
FSEG_MAGIC_N_BYTES: Replaces FSEG_MAGIC_N_VALUE.
dict_hdr_get(), trx_rsegf_get_new(), trx_undo_page_get(),
trx_undo_page_get_s_latched(): Replaced with error-checking calls.
trx_rseg_t::get(mtr_t*): Replaces trx_rsegf_get().
trx_rseg_header_create(): Let the caller update the TRX_SYS page if needed.
trx_sys_create_sys_pages(): Merged with trx_sysf_create().
dict_check_tablespaces_and_store_max_id(): Do not access
DICT_HDR_MAX_SPACE_ID, because it was already recovered in dict_boot().
Merge dict_check_sys_tables() with this function.
dir_pathname(): Replaces os_file_make_new_pathname().
row_undo_ins_remove_sec(): Do not modify the undo page by adding
a terminating NUL byte to the record.
btr_decryption_failed(): Report decryption failures
dict_set_corrupted_by_space(), dict_set_encrypted_by_space(),
dict_set_corrupted_index_cache_only(): Remove.
dict_set_corrupted(): Remove the constant parameter dict_locked=false.
Never flag the clustered index corrupted in SYS_INDEXES, because
that would deny further access to the table. It might be possible to
repair the table by executing ALTER TABLE or OPTIMIZE TABLE, in case
no B-tree leaf page is corrupted.
dict_table_skip_corrupt_index(), dict_table_next_uncorrupted_index(),
row_purge_skip_uncommitted_virtual_index(): Remove, and refactor
the callers to read dict_index_t::type only once.
dict_table_is_corrupted(): Remove.
dict_index_t::is_btree(): Determine if the index is a valid B-tree.
BUF_GET_NO_LATCH, BUF_EVICT_IF_IN_POOL: Remove.
UNIV_BTR_DEBUG: Remove. Any inconsistency will no longer trigger
assertion failures, but error codes being returned.
buf_corrupt_page_release(): Replaced with a direct call to
buf_pool.corrupted_evict().
fil_invalid_page_access_msg(): Never crash on an invalid read;
let the caller of buf_page_get_gen() decide.
btr_pcur_t::restore_position(): Propagate failure status to the caller
by returning CORRUPTED.
opt_search_plan_for_table(): Simplify the code.
row_purge_del_mark(), row_purge_upd_exist_or_extern_func(),
row_undo_ins_remove_sec_rec(), row_undo_mod_upd_del_sec(),
row_undo_mod_del_mark_sec(): Avoid mem_heap_create()/mem_heap_free()
when no secondary indexes exist.
row_undo_mod_upd_exist_sec(): Simplify the code.
row_upd_clust_step(), dict_load_table_one(): Return DB_TABLE_CORRUPT
if the clustered index (and therefore the table) is corrupted, similar
to what we do in row_insert_for_mysql().
fut_get_ptr(): Replace with buf_page_get_gen() calls.
buf_page_get_gen(): Return nullptr and *err=DB_CORRUPTION
if the page is marked as freed. For other modes than
BUF_GET_POSSIBLY_FREED or BUF_PEEK_IF_IN_POOL this will
trigger a debug assertion failure. For BUF_GET_POSSIBLY_FREED,
we will return nullptr for freed pages, so that the callers
can be simplified. The purge of transaction history will be
a new user of BUF_GET_POSSIBLY_FREED, to avoid crashes on
corrupted data.
buf_page_get_low(): Never crash on a corrupted page, but simply
return nullptr.
fseg_page_is_allocated(): Replaces fseg_page_is_free().
fts_drop_common_tables(): Return an error if the transaction
was rolled back.
fil_space_t::set_corrupted(): Report a tablespace as corrupted if
it was not reported already.
fil_space_t::io(): Invoke fil_space_t::set_corrupted() to report
out-of-bounds page access or other errors.
Clean up mtr_t::page_lock()
buf_page_get_low(): Validate the page identifier (to check for
recently read corrupted pages) after acquiring the page latch.
buf_page_t::read_complete(): Flag uninitialized (all-zero) pages
with DB_FAIL. Return DB_PAGE_CORRUPTED on page number mismatch.
mtr_t::defer_drop_ahi(): Renamed from mtr_defer_drop_ahi().
recv_sys_t::free_corrupted_page(): Only set_corrupt_fs()
if any log records exist for the page. We do not mind if read-ahead
produces corrupted (or all-zero) pages that were not actually needed
during recovery.
recv_recover_page(): Return whether the operation succeeded.
recv_sys_t::recover_low(): Simplify the logic. Check for recovery error.
Thanks to Matthias Leich for testing this extensively and to the
authors of https://rr-project.org for making it easy to diagnose
and fix any failures that were found during the testing.
3 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-23399: Performance regression with write workloads
The buffer pool refactoring in MDEV-15053 and MDEV-22871 shifted
the performance bottleneck to the page flushing.
The configuration parameters will be changed as follows:
innodb_lru_flush_size=32 (new: how many pages to flush on LRU eviction)
innodb_lru_scan_depth=1536 (old: 1024)
innodb_max_dirty_pages_pct=90 (old: 75)
innodb_max_dirty_pages_pct_lwm=75 (old: 0)
Note: The parameter innodb_lru_scan_depth will only affect LRU
eviction of buffer pool pages when a new page is being allocated. The
page cleaner thread will no longer evict any pages. It used to
guarantee that some pages will remain free in the buffer pool. Now, we
perform that eviction 'on demand' in buf_LRU_get_free_block().
The parameter innodb_lru_scan_depth(srv_LRU_scan_depth) is used as follows:
* When the buffer pool is being shrunk in buf_pool_t::withdraw_blocks()
* As a buf_pool.free limit in buf_LRU_list_batch() for terminating
the flushing that is initiated e.g., by buf_LRU_get_free_block()
The parameter also used to serve as an initial limit for unzip_LRU
eviction (evicting uncompressed page frames while retaining
ROW_FORMAT=COMPRESSED pages), but now we will use a hard-coded limit
of 100 or unlimited for invoking buf_LRU_scan_and_free_block().
The status variables will be changed as follows:
innodb_buffer_pool_pages_flushed: This includes also the count of
innodb_buffer_pool_pages_LRU_flushed and should work reliably,
updated one by one in buf_flush_page() to give more real-time
statistics. The function buf_flush_stats(), which we are removing,
was not called in every code path. For both counters, we will use
regular variables that are incremented in a critical section of
buf_pool.mutex. Note that show_innodb_vars() directly links to the
variables, and reads of the counters will *not* be protected by
buf_pool.mutex, so you cannot get a consistent snapshot of both variables.
The following INFORMATION_SCHEMA.INNODB_METRICS counters will be
removed, because the page cleaner no longer deals with writing or
evicting least recently used pages, and because the single-page writes
have been removed:
* buffer_LRU_batch_flush_avg_time_slot
* buffer_LRU_batch_flush_avg_time_thread
* buffer_LRU_batch_flush_avg_time_est
* buffer_LRU_batch_flush_avg_pass
* buffer_LRU_single_flush_scanned
* buffer_LRU_single_flush_num_scan
* buffer_LRU_single_flush_scanned_per_call
When moving to a single buffer pool instance in MDEV-15058, we missed
some opportunity to simplify the buf_flush_page_cleaner thread. It was
unnecessarily using a mutex and some complex data structures, even
though we always have a single page cleaner thread.
Furthermore, the buf_flush_page_cleaner thread had separate 'recovery'
and 'shutdown' modes where it was waiting to be triggered by some
other thread, adding unnecessary latency and potential for hangs in
relatively rarely executed startup or shutdown code.
The page cleaner was also running two kinds of batches in an
interleaved fashion: "LRU flush" (writing out some least recently used
pages and evicting them on write completion) and the normal batches
that aim to increase the MIN(oldest_modification) in the buffer pool,
to help the log checkpoint advance.
The buf_pool.flush_list flushing was being blocked by
buf_block_t::lock for no good reason. Furthermore, if the FIL_PAGE_LSN
of a page is ahead of log_sys.get_flushed_lsn(), that is, what has
been persistently written to the redo log, we would trigger a log
flush and then resume the page flushing. This would unnecessarily
limit the performance of the page cleaner thread and trigger the
infamous messages "InnoDB: page_cleaner: 1000ms intended loop took 4450ms.
The settings might not be optimal" that were suppressed in
commit d1ab89037a518fcffbc50c24e4bd94e4ec33aed0 unless log_warnings>2.
Our revised algorithm will make log_sys.get_flushed_lsn() advance at
the start of buf_flush_lists(), and then execute a 'best effort' to
write out all pages. The flush batches will skip pages that were modified
since the log was written, or are are currently exclusively locked.
The MDEV-13670 message "page_cleaner: 1000ms intended loop took" message
will be removed, because by design, the buf_flush_page_cleaner() should
not be blocked during a batch for extended periods of time.
We will remove the single-page flushing altogether. Related to this,
the debug parameter innodb_doublewrite_batch_size will be removed,
because all of the doublewrite buffer will be used for flushing
batches. If a page needs to be evicted from the buffer pool and all
100 least recently used pages in the buffer pool have unflushed
changes, buf_LRU_get_free_block() will execute buf_flush_lists() to
write out and evict innodb_lru_flush_size pages. At most one thread
will execute buf_flush_lists() in buf_LRU_get_free_block(); other
threads will wait for that LRU flushing batch to finish.
To improve concurrency, we will replace the InnoDB ib_mutex_t and
os_event_t native mutexes and condition variables in this area of code.
Most notably, this means that the buffer pool mutex (buf_pool.mutex)
is no longer instrumented via any InnoDB interfaces. It will continue
to be instrumented via PERFORMANCE_SCHEMA.
For now, both buf_pool.flush_list_mutex and buf_pool.mutex will be
declared with MY_MUTEX_INIT_FAST (PTHREAD_MUTEX_ADAPTIVE_NP). The critical
sections of buf_pool.flush_list_mutex should be shorter than those for
buf_pool.mutex, because in the worst case, they cover a linear scan of
buf_pool.flush_list, while the worst case of a critical section of
buf_pool.mutex covers a linear scan of the potentially much longer
buf_pool.LRU list.
mysql_mutex_is_owner(), safe_mutex_is_owner(): New predicate, usable
with SAFE_MUTEX. Some InnoDB debug assertions need this predicate
instead of mysql_mutex_assert_owner() or mysql_mutex_assert_not_owner().
buf_pool_t::n_flush_LRU, buf_pool_t::n_flush_list:
Replaces buf_pool_t::init_flush[] and buf_pool_t::n_flush[].
The number of active flush operations.
buf_pool_t::mutex, buf_pool_t::flush_list_mutex: Use mysql_mutex_t
instead of ib_mutex_t, to have native mutexes with PERFORMANCE_SCHEMA
and SAFE_MUTEX instrumentation.
buf_pool_t::done_flush_LRU: Condition variable for !n_flush_LRU.
buf_pool_t::done_flush_list: Condition variable for !n_flush_list.
buf_pool_t::do_flush_list: Condition variable to wake up the
buf_flush_page_cleaner when a log checkpoint needs to be written
or the server is being shut down. Replaces buf_flush_event.
We will keep using timed waits (the page cleaner thread will wake
_at least_ once per second), because the calculations for
innodb_adaptive_flushing depend on fixed time intervals.
buf_dblwr: Allocate statically, and move all code to member functions.
Use a native mutex and condition variable. Remove code to deal with
single-page flushing.
buf_dblwr_check_block(): Make the check debug-only. We were spending
a significant amount of execution time in page_simple_validate_new().
flush_counters_t::unzip_LRU_evicted: Remove.
IORequest: Make more members const. FIXME: m_fil_node should be removed.
buf_flush_sync_lsn: Protect by std::atomic, not page_cleaner.mutex
(which we are removing).
page_cleaner_slot_t, page_cleaner_t: Remove many redundant members.
pc_request_flush_slot(): Replaces pc_request() and pc_flush_slot().
recv_writer_thread: Remove. Recovery works just fine without it, if we
simply invoke buf_flush_sync() at the end of each batch in
recv_sys_t::apply().
recv_recovery_from_checkpoint_finish(): Remove. We can simply call
recv_sys.debug_free() directly.
srv_started_redo: Replaces srv_start_state.
SRV_SHUTDOWN_FLUSH_PHASE: Remove. logs_empty_and_mark_files_at_shutdown()
can communicate with the normal page cleaner loop via the new function
flush_buffer_pool().
buf_flush_remove(): Assert that the calling thread is holding
buf_pool.flush_list_mutex. This removes unnecessary mutex operations
from buf_flush_remove_pages() and buf_flush_dirty_pages(),
which replace buf_LRU_flush_or_remove_pages().
buf_flush_lists(): Renamed from buf_flush_batch(), with simplified
interface. Return the number of flushed pages. Clarified comments and
renamed min_n to max_n. Identify LRU batch by lsn=0. Merge all the functions
buf_flush_start(), buf_flush_batch(), buf_flush_end() directly to this
function, which was their only caller, and remove 2 unnecessary
buf_pool.mutex release/re-acquisition that we used to perform around
the buf_flush_batch() call. At the start, if not all log has been
durably written, wait for a background task to do it, or start a new
task to do it. This allows the log write to run concurrently with our
page flushing batch. Any pages that were skipped due to too recent
FIL_PAGE_LSN or due to them being latched by a writer should be flushed
during the next batch, unless there are further modifications to those
pages. It is possible that a page that we must flush due to small
oldest_modification also carries a recent FIL_PAGE_LSN or is being
constantly modified. In the worst case, all writers would then end up
waiting in log_free_check() to allow the flushing and the checkpoint
to complete.
buf_do_flush_list_batch(): Clarify comments, and rename min_n to max_n.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_flush_space(): Auxiliary function to look up a tablespace for
page flushing.
buf_flush_page(): Defer the computation of space->full_crc32(). Never
call log_write_up_to(), but instead skip persistent pages whose latest
modification (FIL_PAGE_LSN) is newer than the redo log. Also skip
pages on which we cannot acquire a shared latch without waiting.
buf_flush_try_neighbors(): Do not bother checking buf_fix_count
because buf_flush_page() will no longer wait for the page latch.
Take the tablespace as a parameter, and only execute this function
when innodb_flush_neighbors>0. Avoid repeated calls of page_id_t::fold().
buf_flush_relocate_on_flush_list(): Declare as cold, and push down
a condition from the callers.
buf_flush_check_neighbor(): Take id.fold() as a parameter.
buf_flush_sync(): Ensure that the buf_pool.flush_list is empty,
because the flushing batch will skip pages whose modifications have
not yet been written to the log or were latched for modification.
buf_free_from_unzip_LRU_list_batch(): Remove redundant local variables.
buf_flush_LRU_list_batch(): Let the caller buf_do_LRU_batch() initialize
the counters, and report n->evicted.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_do_LRU_batch(): Return the number of pages flushed.
buf_LRU_free_page(): Only release and re-acquire buf_pool.mutex if
adaptive hash index entries are pointing to the block.
buf_LRU_get_free_block(): Do not wake up the page cleaner, because it
will no longer perform any useful work for us, and we do not want it
to compete for I/O while buf_flush_lists(innodb_lru_flush_size, 0)
writes out and evicts at most innodb_lru_flush_size pages. (The
function buf_do_LRU_batch() may complete after writing fewer pages if
more than innodb_lru_scan_depth pages end up in buf_pool.free list.)
Eliminate some mutex release-acquire cycles, and wait for the LRU
flush batch to complete before rescanning.
buf_LRU_check_size_of_non_data_objects(): Simplify the code.
buf_page_write_complete(): Remove the parameter evict, and always
evict pages that were part of an LRU flush.
buf_page_create(): Take a pre-allocated page as a parameter.
buf_pool_t::free_block(): Free a pre-allocated block.
recv_sys_t::recover_low(), recv_sys_t::apply(): Preallocate the block
while not holding recv_sys.mutex. During page allocation, we may
initiate a page flush, which in turn may initiate a log flush, which
would require acquiring log_sys.mutex, which should always be acquired
before recv_sys.mutex in order to avoid deadlocks. Therefore, we must
not be holding recv_sys.mutex while allocating a buffer pool block.
BtrBulk::logFreeCheck(): Skip a redundant condition.
row_undo_step(): Do not invoke srv_inc_activity_count() for every row
that is being rolled back. It should suffice to invoke the function in
trx_flush_log_if_needed() during trx_t::commit_in_memory() when the
rollback completes.
sync_check_enable(): Remove. We will enable innodb_sync_debug from the
very beginning.
Reviewed by: Vladislav Vaintroub
5 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-34830: LSN in the future is not being treated as serious corruption
The invariant of write-ahead logging is that before any change to a
page is written to the data file, the corresponding log record must
must first have been durably written.
In crash recovery, there were some sloppy checks for this. Let us
implement accurate checks and flag an inconsistency as a hard error,
so that we can avoid further corruption of a corrupted database.
For data extraction from the corrupted database, innodb_force_recovery
can be used.
Before recovery is reading any data pages or invoking
buf_dblwr_t::recover() to recover torn pages from the
doublewrite buffer, InnoDB will have parsed the log until the
final LSN and updated log_sys.lsn to that. So, we can rely on
log_sys.lsn at all times. The doublewrite buffer recovery has been
refactored in such a way that the recv_sys.dblwr.pages may be consulted
while discovering files and their page sizes, but nothing will be
written back to data files before buf_dblwr_t::recover() is invoked.
recv_max_page_lsn, recv_lsn_checks_on: Remove.
recv_sys_t::validate_checkpoint(): Validate the write-ahead-logging
condition at the end of the recovery.
recv_dblwr_t::validate_page(): Keep track of the maximum LSN
(if we are checking a non-doublewrite copy of a page) but
do not complain LSN being in the future. The doublewrite buffer
is a special case, because it will be read early during recovery.
Besides, starting with commit 762bcb81b5bf9bbde61fed59afb26417f4ce1e86
the dblwr=true copies of pages may legitimately be "too new".
recv_dblwr_t::find_page(): Find a valid page with the smallest
FIL_PAGE_LSN that is in the valid range for recovery.
recv_dblwr_t::restore_first_page(): Replaced by find_page().
Only buf_dblwr_t::recover() will write to data files.
buf_dblwr_t::recover(): Simplify the message output. Do attempt
doublewrite recovery on user page read error. Ignore doublewrite
pages whose FIL_PAGE_LSN is outside the usable bounds. Previously,
we could wrongly recover a too new page from the doublewrite buffer.
It is unlikely that this could have lead to an actual error.
Write back all recovered pages from the doublewrite buffer here,
including for the first page of any tablespace.
buf_page_is_corrupted(): Distinguish the return values
CORRUPTED_FUTURE_LSN and CORRUPTED_OTHER.
buf_page_check_corrupt(): Return the error code DB_CORRUPTION
in case the LSN is in the future.
Datafile::read_first_page_flags(): Split from read_first_page().
Take a copy of the first page as a parameter.
recv_sys_t::free_corrupted_page(): Take the file as a parameter
and return whether a message was displayed. This avoids some duplicated
and incomplete error messages.
buf_page_t::read_complete(): Remove some redundant output and always
display the name of the corrupted file. Never return DB_FAIL;
use it only in internal error handling.
IORequest::read_complete(): Assume that buf_page_t::read_complete()
will have reported any error.
fil_space_t::set_corrupted(): Return whether this is the first time
the tablespace had been flagged as corrupted.
Datafile::validate_first_page(), fil_node_open_file_low(),
fil_node_open_file(), fil_space_t::read_page0(),
fil_node_t::read_page0(): Add a parameter for a copy of the
first page, and a parameter to indicate whether the FIL_PAGE_LSN
check should be suppressed. Before buf_dblwr_t::recover() is
invoked, we cannot validate the FIL_PAGE_LSN, but we can trust the
FSP_SPACE_FLAGS and the tablespace ID that may be present in a
potentially too new copy of a page.
Reviewed by: Debarun Banerjee
12 months ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-13542: Crashing on corrupted page is unhelpful
The approach to handling corruption that was chosen by Oracle in
commit 177d8b0c125b841c0650d27d735e3b87509dc286
is not really useful. Not only did it actually fail to prevent InnoDB
from crashing, but it is making things worse by blocking attempts to
rescue data from or rebuild a partially readable table.
We will try to prevent crashes in a different way: by propagating
errors up the call stack. We will never mark the clustered index
persistently corrupted, so that data recovery may be attempted by
reading from the table, or by rebuilding the table.
This should also fix MDEV-13680 (crash on btr_page_alloc() failure);
it was extensively tested with innodb_file_per_table=0 and a
non-autoextend system tablespace.
We should now avoid crashes in many cases, such as when a page
cannot be read or allocated, or an inconsistency is detected when
attempting to update multiple pages. We will not crash on double-free,
such as on the recovery of DDL in system tablespace in case something
was corrupted.
Crashes on corrupted data are still possible. The fault injection mechanism
that is introduced in the subsequent commit may help catch more of them.
buf_page_import_corrupt_failure: Remove the fault injection, and instead
corrupt some pages using Perl code in the tests.
btr_cur_pessimistic_insert(): Always reserve extents (except for the
change buffer), in order to prevent a subsequent allocation failure.
btr_pcur_open_at_rnd_pos(): Merged to the only caller ibuf_merge_pages().
btr_assert_not_corrupted(), btr_corruption_report(): Remove.
Similar checks are already part of btr_block_get().
FSEG_MAGIC_N_BYTES: Replaces FSEG_MAGIC_N_VALUE.
dict_hdr_get(), trx_rsegf_get_new(), trx_undo_page_get(),
trx_undo_page_get_s_latched(): Replaced with error-checking calls.
trx_rseg_t::get(mtr_t*): Replaces trx_rsegf_get().
trx_rseg_header_create(): Let the caller update the TRX_SYS page if needed.
trx_sys_create_sys_pages(): Merged with trx_sysf_create().
dict_check_tablespaces_and_store_max_id(): Do not access
DICT_HDR_MAX_SPACE_ID, because it was already recovered in dict_boot().
Merge dict_check_sys_tables() with this function.
dir_pathname(): Replaces os_file_make_new_pathname().
row_undo_ins_remove_sec(): Do not modify the undo page by adding
a terminating NUL byte to the record.
btr_decryption_failed(): Report decryption failures
dict_set_corrupted_by_space(), dict_set_encrypted_by_space(),
dict_set_corrupted_index_cache_only(): Remove.
dict_set_corrupted(): Remove the constant parameter dict_locked=false.
Never flag the clustered index corrupted in SYS_INDEXES, because
that would deny further access to the table. It might be possible to
repair the table by executing ALTER TABLE or OPTIMIZE TABLE, in case
no B-tree leaf page is corrupted.
dict_table_skip_corrupt_index(), dict_table_next_uncorrupted_index(),
row_purge_skip_uncommitted_virtual_index(): Remove, and refactor
the callers to read dict_index_t::type only once.
dict_table_is_corrupted(): Remove.
dict_index_t::is_btree(): Determine if the index is a valid B-tree.
BUF_GET_NO_LATCH, BUF_EVICT_IF_IN_POOL: Remove.
UNIV_BTR_DEBUG: Remove. Any inconsistency will no longer trigger
assertion failures, but error codes being returned.
buf_corrupt_page_release(): Replaced with a direct call to
buf_pool.corrupted_evict().
fil_invalid_page_access_msg(): Never crash on an invalid read;
let the caller of buf_page_get_gen() decide.
btr_pcur_t::restore_position(): Propagate failure status to the caller
by returning CORRUPTED.
opt_search_plan_for_table(): Simplify the code.
row_purge_del_mark(), row_purge_upd_exist_or_extern_func(),
row_undo_ins_remove_sec_rec(), row_undo_mod_upd_del_sec(),
row_undo_mod_del_mark_sec(): Avoid mem_heap_create()/mem_heap_free()
when no secondary indexes exist.
row_undo_mod_upd_exist_sec(): Simplify the code.
row_upd_clust_step(), dict_load_table_one(): Return DB_TABLE_CORRUPT
if the clustered index (and therefore the table) is corrupted, similar
to what we do in row_insert_for_mysql().
fut_get_ptr(): Replace with buf_page_get_gen() calls.
buf_page_get_gen(): Return nullptr and *err=DB_CORRUPTION
if the page is marked as freed. For other modes than
BUF_GET_POSSIBLY_FREED or BUF_PEEK_IF_IN_POOL this will
trigger a debug assertion failure. For BUF_GET_POSSIBLY_FREED,
we will return nullptr for freed pages, so that the callers
can be simplified. The purge of transaction history will be
a new user of BUF_GET_POSSIBLY_FREED, to avoid crashes on
corrupted data.
buf_page_get_low(): Never crash on a corrupted page, but simply
return nullptr.
fseg_page_is_allocated(): Replaces fseg_page_is_free().
fts_drop_common_tables(): Return an error if the transaction
was rolled back.
fil_space_t::set_corrupted(): Report a tablespace as corrupted if
it was not reported already.
fil_space_t::io(): Invoke fil_space_t::set_corrupted() to report
out-of-bounds page access or other errors.
Clean up mtr_t::page_lock()
buf_page_get_low(): Validate the page identifier (to check for
recently read corrupted pages) after acquiring the page latch.
buf_page_t::read_complete(): Flag uninitialized (all-zero) pages
with DB_FAIL. Return DB_PAGE_CORRUPTED on page number mismatch.
mtr_t::defer_drop_ahi(): Renamed from mtr_defer_drop_ahi().
recv_sys_t::free_corrupted_page(): Only set_corrupt_fs()
if any log records exist for the page. We do not mind if read-ahead
produces corrupted (or all-zero) pages that were not actually needed
during recovery.
recv_recover_page(): Return whether the operation succeeded.
recv_sys_t::recover_low(): Simplify the logic. Check for recovery error.
Thanks to Matthias Leich for testing this extensively and to the
authors of https://rr-project.org for making it easy to diagnose
and fix any failures that were found during the testing.
3 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-34830: LSN in the future is not being treated as serious corruption
The invariant of write-ahead logging is that before any change to a
page is written to the data file, the corresponding log record must
must first have been durably written.
In crash recovery, there were some sloppy checks for this. Let us
implement accurate checks and flag an inconsistency as a hard error,
so that we can avoid further corruption of a corrupted database.
For data extraction from the corrupted database, innodb_force_recovery
can be used.
Before recovery is reading any data pages or invoking
buf_dblwr_t::recover() to recover torn pages from the
doublewrite buffer, InnoDB will have parsed the log until the
final LSN and updated log_sys.lsn to that. So, we can rely on
log_sys.lsn at all times. The doublewrite buffer recovery has been
refactored in such a way that the recv_sys.dblwr.pages may be consulted
while discovering files and their page sizes, but nothing will be
written back to data files before buf_dblwr_t::recover() is invoked.
recv_max_page_lsn, recv_lsn_checks_on: Remove.
recv_sys_t::validate_checkpoint(): Validate the write-ahead-logging
condition at the end of the recovery.
recv_dblwr_t::validate_page(): Keep track of the maximum LSN
(if we are checking a non-doublewrite copy of a page) but
do not complain LSN being in the future. The doublewrite buffer
is a special case, because it will be read early during recovery.
Besides, starting with commit 762bcb81b5bf9bbde61fed59afb26417f4ce1e86
the dblwr=true copies of pages may legitimately be "too new".
recv_dblwr_t::find_page(): Find a valid page with the smallest
FIL_PAGE_LSN that is in the valid range for recovery.
recv_dblwr_t::restore_first_page(): Replaced by find_page().
Only buf_dblwr_t::recover() will write to data files.
buf_dblwr_t::recover(): Simplify the message output. Do attempt
doublewrite recovery on user page read error. Ignore doublewrite
pages whose FIL_PAGE_LSN is outside the usable bounds. Previously,
we could wrongly recover a too new page from the doublewrite buffer.
It is unlikely that this could have lead to an actual error.
Write back all recovered pages from the doublewrite buffer here,
including for the first page of any tablespace.
buf_page_is_corrupted(): Distinguish the return values
CORRUPTED_FUTURE_LSN and CORRUPTED_OTHER.
buf_page_check_corrupt(): Return the error code DB_CORRUPTION
in case the LSN is in the future.
Datafile::read_first_page_flags(): Split from read_first_page().
Take a copy of the first page as a parameter.
recv_sys_t::free_corrupted_page(): Take the file as a parameter
and return whether a message was displayed. This avoids some duplicated
and incomplete error messages.
buf_page_t::read_complete(): Remove some redundant output and always
display the name of the corrupted file. Never return DB_FAIL;
use it only in internal error handling.
IORequest::read_complete(): Assume that buf_page_t::read_complete()
will have reported any error.
fil_space_t::set_corrupted(): Return whether this is the first time
the tablespace had been flagged as corrupted.
Datafile::validate_first_page(), fil_node_open_file_low(),
fil_node_open_file(), fil_space_t::read_page0(),
fil_node_t::read_page0(): Add a parameter for a copy of the
first page, and a parameter to indicate whether the FIL_PAGE_LSN
check should be suppressed. Before buf_dblwr_t::recover() is
invoked, we cannot validate the FIL_PAGE_LSN, but we can trust the
FSP_SPACE_FLAGS and the tablespace ID that may be present in a
potentially too new copy of a page.
Reviewed by: Debarun Banerjee
12 months ago MDEV-13542: Crashing on corrupted page is unhelpful
The approach to handling corruption that was chosen by Oracle in
commit 177d8b0c125b841c0650d27d735e3b87509dc286
is not really useful. Not only did it actually fail to prevent InnoDB
from crashing, but it is making things worse by blocking attempts to
rescue data from or rebuild a partially readable table.
We will try to prevent crashes in a different way: by propagating
errors up the call stack. We will never mark the clustered index
persistently corrupted, so that data recovery may be attempted by
reading from the table, or by rebuilding the table.
This should also fix MDEV-13680 (crash on btr_page_alloc() failure);
it was extensively tested with innodb_file_per_table=0 and a
non-autoextend system tablespace.
We should now avoid crashes in many cases, such as when a page
cannot be read or allocated, or an inconsistency is detected when
attempting to update multiple pages. We will not crash on double-free,
such as on the recovery of DDL in system tablespace in case something
was corrupted.
Crashes on corrupted data are still possible. The fault injection mechanism
that is introduced in the subsequent commit may help catch more of them.
buf_page_import_corrupt_failure: Remove the fault injection, and instead
corrupt some pages using Perl code in the tests.
btr_cur_pessimistic_insert(): Always reserve extents (except for the
change buffer), in order to prevent a subsequent allocation failure.
btr_pcur_open_at_rnd_pos(): Merged to the only caller ibuf_merge_pages().
btr_assert_not_corrupted(), btr_corruption_report(): Remove.
Similar checks are already part of btr_block_get().
FSEG_MAGIC_N_BYTES: Replaces FSEG_MAGIC_N_VALUE.
dict_hdr_get(), trx_rsegf_get_new(), trx_undo_page_get(),
trx_undo_page_get_s_latched(): Replaced with error-checking calls.
trx_rseg_t::get(mtr_t*): Replaces trx_rsegf_get().
trx_rseg_header_create(): Let the caller update the TRX_SYS page if needed.
trx_sys_create_sys_pages(): Merged with trx_sysf_create().
dict_check_tablespaces_and_store_max_id(): Do not access
DICT_HDR_MAX_SPACE_ID, because it was already recovered in dict_boot().
Merge dict_check_sys_tables() with this function.
dir_pathname(): Replaces os_file_make_new_pathname().
row_undo_ins_remove_sec(): Do not modify the undo page by adding
a terminating NUL byte to the record.
btr_decryption_failed(): Report decryption failures
dict_set_corrupted_by_space(), dict_set_encrypted_by_space(),
dict_set_corrupted_index_cache_only(): Remove.
dict_set_corrupted(): Remove the constant parameter dict_locked=false.
Never flag the clustered index corrupted in SYS_INDEXES, because
that would deny further access to the table. It might be possible to
repair the table by executing ALTER TABLE or OPTIMIZE TABLE, in case
no B-tree leaf page is corrupted.
dict_table_skip_corrupt_index(), dict_table_next_uncorrupted_index(),
row_purge_skip_uncommitted_virtual_index(): Remove, and refactor
the callers to read dict_index_t::type only once.
dict_table_is_corrupted(): Remove.
dict_index_t::is_btree(): Determine if the index is a valid B-tree.
BUF_GET_NO_LATCH, BUF_EVICT_IF_IN_POOL: Remove.
UNIV_BTR_DEBUG: Remove. Any inconsistency will no longer trigger
assertion failures, but error codes being returned.
buf_corrupt_page_release(): Replaced with a direct call to
buf_pool.corrupted_evict().
fil_invalid_page_access_msg(): Never crash on an invalid read;
let the caller of buf_page_get_gen() decide.
btr_pcur_t::restore_position(): Propagate failure status to the caller
by returning CORRUPTED.
opt_search_plan_for_table(): Simplify the code.
row_purge_del_mark(), row_purge_upd_exist_or_extern_func(),
row_undo_ins_remove_sec_rec(), row_undo_mod_upd_del_sec(),
row_undo_mod_del_mark_sec(): Avoid mem_heap_create()/mem_heap_free()
when no secondary indexes exist.
row_undo_mod_upd_exist_sec(): Simplify the code.
row_upd_clust_step(), dict_load_table_one(): Return DB_TABLE_CORRUPT
if the clustered index (and therefore the table) is corrupted, similar
to what we do in row_insert_for_mysql().
fut_get_ptr(): Replace with buf_page_get_gen() calls.
buf_page_get_gen(): Return nullptr and *err=DB_CORRUPTION
if the page is marked as freed. For other modes than
BUF_GET_POSSIBLY_FREED or BUF_PEEK_IF_IN_POOL this will
trigger a debug assertion failure. For BUF_GET_POSSIBLY_FREED,
we will return nullptr for freed pages, so that the callers
can be simplified. The purge of transaction history will be
a new user of BUF_GET_POSSIBLY_FREED, to avoid crashes on
corrupted data.
buf_page_get_low(): Never crash on a corrupted page, but simply
return nullptr.
fseg_page_is_allocated(): Replaces fseg_page_is_free().
fts_drop_common_tables(): Return an error if the transaction
was rolled back.
fil_space_t::set_corrupted(): Report a tablespace as corrupted if
it was not reported already.
fil_space_t::io(): Invoke fil_space_t::set_corrupted() to report
out-of-bounds page access or other errors.
Clean up mtr_t::page_lock()
buf_page_get_low(): Validate the page identifier (to check for
recently read corrupted pages) after acquiring the page latch.
buf_page_t::read_complete(): Flag uninitialized (all-zero) pages
with DB_FAIL. Return DB_PAGE_CORRUPTED on page number mismatch.
mtr_t::defer_drop_ahi(): Renamed from mtr_defer_drop_ahi().
recv_sys_t::free_corrupted_page(): Only set_corrupt_fs()
if any log records exist for the page. We do not mind if read-ahead
produces corrupted (or all-zero) pages that were not actually needed
during recovery.
recv_recover_page(): Return whether the operation succeeded.
recv_sys_t::recover_low(): Simplify the logic. Check for recovery error.
Thanks to Matthias Leich for testing this extensively and to the
authors of https://rr-project.org for making it easy to diagnose
and fix any failures that were found during the testing.
3 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-13542: Crashing on corrupted page is unhelpful
The approach to handling corruption that was chosen by Oracle in
commit 177d8b0c125b841c0650d27d735e3b87509dc286
is not really useful. Not only did it actually fail to prevent InnoDB
from crashing, but it is making things worse by blocking attempts to
rescue data from or rebuild a partially readable table.
We will try to prevent crashes in a different way: by propagating
errors up the call stack. We will never mark the clustered index
persistently corrupted, so that data recovery may be attempted by
reading from the table, or by rebuilding the table.
This should also fix MDEV-13680 (crash on btr_page_alloc() failure);
it was extensively tested with innodb_file_per_table=0 and a
non-autoextend system tablespace.
We should now avoid crashes in many cases, such as when a page
cannot be read or allocated, or an inconsistency is detected when
attempting to update multiple pages. We will not crash on double-free,
such as on the recovery of DDL in system tablespace in case something
was corrupted.
Crashes on corrupted data are still possible. The fault injection mechanism
that is introduced in the subsequent commit may help catch more of them.
buf_page_import_corrupt_failure: Remove the fault injection, and instead
corrupt some pages using Perl code in the tests.
btr_cur_pessimistic_insert(): Always reserve extents (except for the
change buffer), in order to prevent a subsequent allocation failure.
btr_pcur_open_at_rnd_pos(): Merged to the only caller ibuf_merge_pages().
btr_assert_not_corrupted(), btr_corruption_report(): Remove.
Similar checks are already part of btr_block_get().
FSEG_MAGIC_N_BYTES: Replaces FSEG_MAGIC_N_VALUE.
dict_hdr_get(), trx_rsegf_get_new(), trx_undo_page_get(),
trx_undo_page_get_s_latched(): Replaced with error-checking calls.
trx_rseg_t::get(mtr_t*): Replaces trx_rsegf_get().
trx_rseg_header_create(): Let the caller update the TRX_SYS page if needed.
trx_sys_create_sys_pages(): Merged with trx_sysf_create().
dict_check_tablespaces_and_store_max_id(): Do not access
DICT_HDR_MAX_SPACE_ID, because it was already recovered in dict_boot().
Merge dict_check_sys_tables() with this function.
dir_pathname(): Replaces os_file_make_new_pathname().
row_undo_ins_remove_sec(): Do not modify the undo page by adding
a terminating NUL byte to the record.
btr_decryption_failed(): Report decryption failures
dict_set_corrupted_by_space(), dict_set_encrypted_by_space(),
dict_set_corrupted_index_cache_only(): Remove.
dict_set_corrupted(): Remove the constant parameter dict_locked=false.
Never flag the clustered index corrupted in SYS_INDEXES, because
that would deny further access to the table. It might be possible to
repair the table by executing ALTER TABLE or OPTIMIZE TABLE, in case
no B-tree leaf page is corrupted.
dict_table_skip_corrupt_index(), dict_table_next_uncorrupted_index(),
row_purge_skip_uncommitted_virtual_index(): Remove, and refactor
the callers to read dict_index_t::type only once.
dict_table_is_corrupted(): Remove.
dict_index_t::is_btree(): Determine if the index is a valid B-tree.
BUF_GET_NO_LATCH, BUF_EVICT_IF_IN_POOL: Remove.
UNIV_BTR_DEBUG: Remove. Any inconsistency will no longer trigger
assertion failures, but error codes being returned.
buf_corrupt_page_release(): Replaced with a direct call to
buf_pool.corrupted_evict().
fil_invalid_page_access_msg(): Never crash on an invalid read;
let the caller of buf_page_get_gen() decide.
btr_pcur_t::restore_position(): Propagate failure status to the caller
by returning CORRUPTED.
opt_search_plan_for_table(): Simplify the code.
row_purge_del_mark(), row_purge_upd_exist_or_extern_func(),
row_undo_ins_remove_sec_rec(), row_undo_mod_upd_del_sec(),
row_undo_mod_del_mark_sec(): Avoid mem_heap_create()/mem_heap_free()
when no secondary indexes exist.
row_undo_mod_upd_exist_sec(): Simplify the code.
row_upd_clust_step(), dict_load_table_one(): Return DB_TABLE_CORRUPT
if the clustered index (and therefore the table) is corrupted, similar
to what we do in row_insert_for_mysql().
fut_get_ptr(): Replace with buf_page_get_gen() calls.
buf_page_get_gen(): Return nullptr and *err=DB_CORRUPTION
if the page is marked as freed. For other modes than
BUF_GET_POSSIBLY_FREED or BUF_PEEK_IF_IN_POOL this will
trigger a debug assertion failure. For BUF_GET_POSSIBLY_FREED,
we will return nullptr for freed pages, so that the callers
can be simplified. The purge of transaction history will be
a new user of BUF_GET_POSSIBLY_FREED, to avoid crashes on
corrupted data.
buf_page_get_low(): Never crash on a corrupted page, but simply
return nullptr.
fseg_page_is_allocated(): Replaces fseg_page_is_free().
fts_drop_common_tables(): Return an error if the transaction
was rolled back.
fil_space_t::set_corrupted(): Report a tablespace as corrupted if
it was not reported already.
fil_space_t::io(): Invoke fil_space_t::set_corrupted() to report
out-of-bounds page access or other errors.
Clean up mtr_t::page_lock()
buf_page_get_low(): Validate the page identifier (to check for
recently read corrupted pages) after acquiring the page latch.
buf_page_t::read_complete(): Flag uninitialized (all-zero) pages
with DB_FAIL. Return DB_PAGE_CORRUPTED on page number mismatch.
mtr_t::defer_drop_ahi(): Renamed from mtr_defer_drop_ahi().
recv_sys_t::free_corrupted_page(): Only set_corrupt_fs()
if any log records exist for the page. We do not mind if read-ahead
produces corrupted (or all-zero) pages that were not actually needed
during recovery.
recv_recover_page(): Return whether the operation succeeded.
recv_sys_t::recover_low(): Simplify the logic. Check for recovery error.
Thanks to Matthias Leich for testing this extensively and to the
authors of https://rr-project.org for making it easy to diagnose
and fix any failures that were found during the testing.
3 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-24142: Replace InnoDB rw_lock_t with sux_lock
InnoDB buffer pool block and index tree latches depend on a
special kind of read-update-write lock that allows reentrant
(recursive) acquisition of the 'update' and 'write' locks
as well as an upgrade from 'update' lock to 'write' lock.
The 'update' lock allows any number of reader locks from
other threads, but no concurrent 'update' or 'write' lock.
If there were no requirement to support an upgrade from 'update'
to 'write', we could compose the lock out of two srw_lock
(implemented as any type of native rw-lock, such as SRWLOCK on
Microsoft Windows). Removing this requirement is very difficult,
so in commit f7e7f487d4b06695f91f6fbeb0396b9d87fc7bbf we
implemented an 'update' mode to our srw_lock.
Re-entrant or recursive locking is mostly needed when writing or
freeing BLOB pages, but also in crash recovery or when merging
buffered changes to an index page. The re-entrancy allows us to
attach a previously acquired page to a sub-mini-transaction that
will be committed before whatever else is holding the page latch.
The SUX lock supports Shared ('read'), Update, and eXclusive ('write')
locking modes. The S latches are not re-entrant, but a single S latch
may be acquired even if the thread already holds an U latch.
The idea of the U latch is to allow a write of something that concurrent
readers do not care about (such as the contents of BTR_SEG_LEAF,
BTR_SEG_TOP and other page allocation metadata structures, or
the MDEV-6076 PAGE_ROOT_AUTO_INC). (The PAGE_ROOT_AUTO_INC field
is only updated when a dict_table_t for the table exists, and only
read when a dict_table_t for the table is being added to dict_sys.)
block_lock::u_lock_try(bool for_io=true) is used in buf_flush_page()
to allow concurrent readers but no concurrent modifications while the
page is being written to the data file. That latch will be released
by buf_page_write_complete() in a different thread. Hence, we use
the special lock owner value FOR_IO.
The index_lock::u_lock() improves concurrency on operations that
involve non-leaf index pages.
The interface has been cleaned up a little. We will use
x_lock_recursive() instead of x_lock() when we know that a
lock is already held by the current thread. Similarly,
a lock upgrade from U to X is only allowed via u_x_upgrade()
or x_lock_upgraded() but not via x_lock().
We will disable the LatchDebug and sync_array interfaces to
InnoDB rw-locks.
The SEMAPHORES section of SHOW ENGINE INNODB STATUS output
will no longer include any information about InnoDB rw-locks,
only TTASEventMutex (cmake -DMUTEXTYPE=event) waits.
This will make a part of the 'innotop' script dead code.
The block_lock buf_block_t::lock will not be covered by any
PERFORMANCE_SCHEMA instrumentation.
SHOW ENGINE INNODB MUTEX and INFORMATION_SCHEMA.INNODB_MUTEXES
will no longer output source code file names or line numbers.
The dict_index_t::lock will be identified by index and table names,
which should be much more useful. PERFORMANCE_SCHEMA is lumping
information about all dict_index_t::lock together as
event_name='wait/synch/sxlock/innodb/index_tree_rw_lock'.
buf_page_free(): Remove the file,line parameters. The sux_lock will
not store such diagnostic information.
buf_block_dbg_add_level(): Define as empty macro, to be removed
in a subsequent commit.
Unless the build was configured with cmake -DPLUGIN_PERFSCHEMA=NO
the index_lock dict_index_t::lock will be instrumented via
PERFORMANCE_SCHEMA. Similar to
commit 1669c8890ca2e9092213626e5b047e58ca8b1e77
we will distinguish lock waits by registering shared_lock,exclusive_lock
events instead of try_shared_lock,try_exclusive_lock.
Actual 'try' operations will not be instrumented at all.
rw_lock_list: Remove. After MDEV-24167, this only covered
buf_block_t::lock and dict_index_t::lock. We will output their
information by traversing buf_pool or dict_sys.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-24142: Replace InnoDB rw_lock_t with sux_lock
InnoDB buffer pool block and index tree latches depend on a
special kind of read-update-write lock that allows reentrant
(recursive) acquisition of the 'update' and 'write' locks
as well as an upgrade from 'update' lock to 'write' lock.
The 'update' lock allows any number of reader locks from
other threads, but no concurrent 'update' or 'write' lock.
If there were no requirement to support an upgrade from 'update'
to 'write', we could compose the lock out of two srw_lock
(implemented as any type of native rw-lock, such as SRWLOCK on
Microsoft Windows). Removing this requirement is very difficult,
so in commit f7e7f487d4b06695f91f6fbeb0396b9d87fc7bbf we
implemented an 'update' mode to our srw_lock.
Re-entrant or recursive locking is mostly needed when writing or
freeing BLOB pages, but also in crash recovery or when merging
buffered changes to an index page. The re-entrancy allows us to
attach a previously acquired page to a sub-mini-transaction that
will be committed before whatever else is holding the page latch.
The SUX lock supports Shared ('read'), Update, and eXclusive ('write')
locking modes. The S latches are not re-entrant, but a single S latch
may be acquired even if the thread already holds an U latch.
The idea of the U latch is to allow a write of something that concurrent
readers do not care about (such as the contents of BTR_SEG_LEAF,
BTR_SEG_TOP and other page allocation metadata structures, or
the MDEV-6076 PAGE_ROOT_AUTO_INC). (The PAGE_ROOT_AUTO_INC field
is only updated when a dict_table_t for the table exists, and only
read when a dict_table_t for the table is being added to dict_sys.)
block_lock::u_lock_try(bool for_io=true) is used in buf_flush_page()
to allow concurrent readers but no concurrent modifications while the
page is being written to the data file. That latch will be released
by buf_page_write_complete() in a different thread. Hence, we use
the special lock owner value FOR_IO.
The index_lock::u_lock() improves concurrency on operations that
involve non-leaf index pages.
The interface has been cleaned up a little. We will use
x_lock_recursive() instead of x_lock() when we know that a
lock is already held by the current thread. Similarly,
a lock upgrade from U to X is only allowed via u_x_upgrade()
or x_lock_upgraded() but not via x_lock().
We will disable the LatchDebug and sync_array interfaces to
InnoDB rw-locks.
The SEMAPHORES section of SHOW ENGINE INNODB STATUS output
will no longer include any information about InnoDB rw-locks,
only TTASEventMutex (cmake -DMUTEXTYPE=event) waits.
This will make a part of the 'innotop' script dead code.
The block_lock buf_block_t::lock will not be covered by any
PERFORMANCE_SCHEMA instrumentation.
SHOW ENGINE INNODB MUTEX and INFORMATION_SCHEMA.INNODB_MUTEXES
will no longer output source code file names or line numbers.
The dict_index_t::lock will be identified by index and table names,
which should be much more useful. PERFORMANCE_SCHEMA is lumping
information about all dict_index_t::lock together as
event_name='wait/synch/sxlock/innodb/index_tree_rw_lock'.
buf_page_free(): Remove the file,line parameters. The sux_lock will
not store such diagnostic information.
buf_block_dbg_add_level(): Define as empty macro, to be removed
in a subsequent commit.
Unless the build was configured with cmake -DPLUGIN_PERFSCHEMA=NO
the index_lock dict_index_t::lock will be instrumented via
PERFORMANCE_SCHEMA. Similar to
commit 1669c8890ca2e9092213626e5b047e58ca8b1e77
we will distinguish lock waits by registering shared_lock,exclusive_lock
events instead of try_shared_lock,try_exclusive_lock.
Actual 'try' operations will not be instrumented at all.
rw_lock_list: Remove. After MDEV-24167, this only covered
buf_block_t::lock and dict_index_t::lock. We will output their
information by traversing buf_pool or dict_sys.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-23399: Performance regression with write workloads
The buffer pool refactoring in MDEV-15053 and MDEV-22871 shifted
the performance bottleneck to the page flushing.
The configuration parameters will be changed as follows:
innodb_lru_flush_size=32 (new: how many pages to flush on LRU eviction)
innodb_lru_scan_depth=1536 (old: 1024)
innodb_max_dirty_pages_pct=90 (old: 75)
innodb_max_dirty_pages_pct_lwm=75 (old: 0)
Note: The parameter innodb_lru_scan_depth will only affect LRU
eviction of buffer pool pages when a new page is being allocated. The
page cleaner thread will no longer evict any pages. It used to
guarantee that some pages will remain free in the buffer pool. Now, we
perform that eviction 'on demand' in buf_LRU_get_free_block().
The parameter innodb_lru_scan_depth(srv_LRU_scan_depth) is used as follows:
* When the buffer pool is being shrunk in buf_pool_t::withdraw_blocks()
* As a buf_pool.free limit in buf_LRU_list_batch() for terminating
the flushing that is initiated e.g., by buf_LRU_get_free_block()
The parameter also used to serve as an initial limit for unzip_LRU
eviction (evicting uncompressed page frames while retaining
ROW_FORMAT=COMPRESSED pages), but now we will use a hard-coded limit
of 100 or unlimited for invoking buf_LRU_scan_and_free_block().
The status variables will be changed as follows:
innodb_buffer_pool_pages_flushed: This includes also the count of
innodb_buffer_pool_pages_LRU_flushed and should work reliably,
updated one by one in buf_flush_page() to give more real-time
statistics. The function buf_flush_stats(), which we are removing,
was not called in every code path. For both counters, we will use
regular variables that are incremented in a critical section of
buf_pool.mutex. Note that show_innodb_vars() directly links to the
variables, and reads of the counters will *not* be protected by
buf_pool.mutex, so you cannot get a consistent snapshot of both variables.
The following INFORMATION_SCHEMA.INNODB_METRICS counters will be
removed, because the page cleaner no longer deals with writing or
evicting least recently used pages, and because the single-page writes
have been removed:
* buffer_LRU_batch_flush_avg_time_slot
* buffer_LRU_batch_flush_avg_time_thread
* buffer_LRU_batch_flush_avg_time_est
* buffer_LRU_batch_flush_avg_pass
* buffer_LRU_single_flush_scanned
* buffer_LRU_single_flush_num_scan
* buffer_LRU_single_flush_scanned_per_call
When moving to a single buffer pool instance in MDEV-15058, we missed
some opportunity to simplify the buf_flush_page_cleaner thread. It was
unnecessarily using a mutex and some complex data structures, even
though we always have a single page cleaner thread.
Furthermore, the buf_flush_page_cleaner thread had separate 'recovery'
and 'shutdown' modes where it was waiting to be triggered by some
other thread, adding unnecessary latency and potential for hangs in
relatively rarely executed startup or shutdown code.
The page cleaner was also running two kinds of batches in an
interleaved fashion: "LRU flush" (writing out some least recently used
pages and evicting them on write completion) and the normal batches
that aim to increase the MIN(oldest_modification) in the buffer pool,
to help the log checkpoint advance.
The buf_pool.flush_list flushing was being blocked by
buf_block_t::lock for no good reason. Furthermore, if the FIL_PAGE_LSN
of a page is ahead of log_sys.get_flushed_lsn(), that is, what has
been persistently written to the redo log, we would trigger a log
flush and then resume the page flushing. This would unnecessarily
limit the performance of the page cleaner thread and trigger the
infamous messages "InnoDB: page_cleaner: 1000ms intended loop took 4450ms.
The settings might not be optimal" that were suppressed in
commit d1ab89037a518fcffbc50c24e4bd94e4ec33aed0 unless log_warnings>2.
Our revised algorithm will make log_sys.get_flushed_lsn() advance at
the start of buf_flush_lists(), and then execute a 'best effort' to
write out all pages. The flush batches will skip pages that were modified
since the log was written, or are are currently exclusively locked.
The MDEV-13670 message "page_cleaner: 1000ms intended loop took" message
will be removed, because by design, the buf_flush_page_cleaner() should
not be blocked during a batch for extended periods of time.
We will remove the single-page flushing altogether. Related to this,
the debug parameter innodb_doublewrite_batch_size will be removed,
because all of the doublewrite buffer will be used for flushing
batches. If a page needs to be evicted from the buffer pool and all
100 least recently used pages in the buffer pool have unflushed
changes, buf_LRU_get_free_block() will execute buf_flush_lists() to
write out and evict innodb_lru_flush_size pages. At most one thread
will execute buf_flush_lists() in buf_LRU_get_free_block(); other
threads will wait for that LRU flushing batch to finish.
To improve concurrency, we will replace the InnoDB ib_mutex_t and
os_event_t native mutexes and condition variables in this area of code.
Most notably, this means that the buffer pool mutex (buf_pool.mutex)
is no longer instrumented via any InnoDB interfaces. It will continue
to be instrumented via PERFORMANCE_SCHEMA.
For now, both buf_pool.flush_list_mutex and buf_pool.mutex will be
declared with MY_MUTEX_INIT_FAST (PTHREAD_MUTEX_ADAPTIVE_NP). The critical
sections of buf_pool.flush_list_mutex should be shorter than those for
buf_pool.mutex, because in the worst case, they cover a linear scan of
buf_pool.flush_list, while the worst case of a critical section of
buf_pool.mutex covers a linear scan of the potentially much longer
buf_pool.LRU list.
mysql_mutex_is_owner(), safe_mutex_is_owner(): New predicate, usable
with SAFE_MUTEX. Some InnoDB debug assertions need this predicate
instead of mysql_mutex_assert_owner() or mysql_mutex_assert_not_owner().
buf_pool_t::n_flush_LRU, buf_pool_t::n_flush_list:
Replaces buf_pool_t::init_flush[] and buf_pool_t::n_flush[].
The number of active flush operations.
buf_pool_t::mutex, buf_pool_t::flush_list_mutex: Use mysql_mutex_t
instead of ib_mutex_t, to have native mutexes with PERFORMANCE_SCHEMA
and SAFE_MUTEX instrumentation.
buf_pool_t::done_flush_LRU: Condition variable for !n_flush_LRU.
buf_pool_t::done_flush_list: Condition variable for !n_flush_list.
buf_pool_t::do_flush_list: Condition variable to wake up the
buf_flush_page_cleaner when a log checkpoint needs to be written
or the server is being shut down. Replaces buf_flush_event.
We will keep using timed waits (the page cleaner thread will wake
_at least_ once per second), because the calculations for
innodb_adaptive_flushing depend on fixed time intervals.
buf_dblwr: Allocate statically, and move all code to member functions.
Use a native mutex and condition variable. Remove code to deal with
single-page flushing.
buf_dblwr_check_block(): Make the check debug-only. We were spending
a significant amount of execution time in page_simple_validate_new().
flush_counters_t::unzip_LRU_evicted: Remove.
IORequest: Make more members const. FIXME: m_fil_node should be removed.
buf_flush_sync_lsn: Protect by std::atomic, not page_cleaner.mutex
(which we are removing).
page_cleaner_slot_t, page_cleaner_t: Remove many redundant members.
pc_request_flush_slot(): Replaces pc_request() and pc_flush_slot().
recv_writer_thread: Remove. Recovery works just fine without it, if we
simply invoke buf_flush_sync() at the end of each batch in
recv_sys_t::apply().
recv_recovery_from_checkpoint_finish(): Remove. We can simply call
recv_sys.debug_free() directly.
srv_started_redo: Replaces srv_start_state.
SRV_SHUTDOWN_FLUSH_PHASE: Remove. logs_empty_and_mark_files_at_shutdown()
can communicate with the normal page cleaner loop via the new function
flush_buffer_pool().
buf_flush_remove(): Assert that the calling thread is holding
buf_pool.flush_list_mutex. This removes unnecessary mutex operations
from buf_flush_remove_pages() and buf_flush_dirty_pages(),
which replace buf_LRU_flush_or_remove_pages().
buf_flush_lists(): Renamed from buf_flush_batch(), with simplified
interface. Return the number of flushed pages. Clarified comments and
renamed min_n to max_n. Identify LRU batch by lsn=0. Merge all the functions
buf_flush_start(), buf_flush_batch(), buf_flush_end() directly to this
function, which was their only caller, and remove 2 unnecessary
buf_pool.mutex release/re-acquisition that we used to perform around
the buf_flush_batch() call. At the start, if not all log has been
durably written, wait for a background task to do it, or start a new
task to do it. This allows the log write to run concurrently with our
page flushing batch. Any pages that were skipped due to too recent
FIL_PAGE_LSN or due to them being latched by a writer should be flushed
during the next batch, unless there are further modifications to those
pages. It is possible that a page that we must flush due to small
oldest_modification also carries a recent FIL_PAGE_LSN or is being
constantly modified. In the worst case, all writers would then end up
waiting in log_free_check() to allow the flushing and the checkpoint
to complete.
buf_do_flush_list_batch(): Clarify comments, and rename min_n to max_n.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_flush_space(): Auxiliary function to look up a tablespace for
page flushing.
buf_flush_page(): Defer the computation of space->full_crc32(). Never
call log_write_up_to(), but instead skip persistent pages whose latest
modification (FIL_PAGE_LSN) is newer than the redo log. Also skip
pages on which we cannot acquire a shared latch without waiting.
buf_flush_try_neighbors(): Do not bother checking buf_fix_count
because buf_flush_page() will no longer wait for the page latch.
Take the tablespace as a parameter, and only execute this function
when innodb_flush_neighbors>0. Avoid repeated calls of page_id_t::fold().
buf_flush_relocate_on_flush_list(): Declare as cold, and push down
a condition from the callers.
buf_flush_check_neighbor(): Take id.fold() as a parameter.
buf_flush_sync(): Ensure that the buf_pool.flush_list is empty,
because the flushing batch will skip pages whose modifications have
not yet been written to the log or were latched for modification.
buf_free_from_unzip_LRU_list_batch(): Remove redundant local variables.
buf_flush_LRU_list_batch(): Let the caller buf_do_LRU_batch() initialize
the counters, and report n->evicted.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_do_LRU_batch(): Return the number of pages flushed.
buf_LRU_free_page(): Only release and re-acquire buf_pool.mutex if
adaptive hash index entries are pointing to the block.
buf_LRU_get_free_block(): Do not wake up the page cleaner, because it
will no longer perform any useful work for us, and we do not want it
to compete for I/O while buf_flush_lists(innodb_lru_flush_size, 0)
writes out and evicts at most innodb_lru_flush_size pages. (The
function buf_do_LRU_batch() may complete after writing fewer pages if
more than innodb_lru_scan_depth pages end up in buf_pool.free list.)
Eliminate some mutex release-acquire cycles, and wait for the LRU
flush batch to complete before rescanning.
buf_LRU_check_size_of_non_data_objects(): Simplify the code.
buf_page_write_complete(): Remove the parameter evict, and always
evict pages that were part of an LRU flush.
buf_page_create(): Take a pre-allocated page as a parameter.
buf_pool_t::free_block(): Free a pre-allocated block.
recv_sys_t::recover_low(), recv_sys_t::apply(): Preallocate the block
while not holding recv_sys.mutex. During page allocation, we may
initiate a page flush, which in turn may initiate a log flush, which
would require acquiring log_sys.mutex, which should always be acquired
before recv_sys.mutex in order to avoid deadlocks. Therefore, we must
not be holding recv_sys.mutex while allocating a buffer pool block.
BtrBulk::logFreeCheck(): Skip a redundant condition.
row_undo_step(): Do not invoke srv_inc_activity_count() for every row
that is being rolled back. It should suffice to invoke the function in
trx_flush_log_if_needed() during trx_t::commit_in_memory() when the
rollback completes.
sync_check_enable(): Remove. We will enable innodb_sync_debug from the
very beginning.
Reviewed by: Vladislav Vaintroub
5 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-23399: Performance regression with write workloads
The buffer pool refactoring in MDEV-15053 and MDEV-22871 shifted
the performance bottleneck to the page flushing.
The configuration parameters will be changed as follows:
innodb_lru_flush_size=32 (new: how many pages to flush on LRU eviction)
innodb_lru_scan_depth=1536 (old: 1024)
innodb_max_dirty_pages_pct=90 (old: 75)
innodb_max_dirty_pages_pct_lwm=75 (old: 0)
Note: The parameter innodb_lru_scan_depth will only affect LRU
eviction of buffer pool pages when a new page is being allocated. The
page cleaner thread will no longer evict any pages. It used to
guarantee that some pages will remain free in the buffer pool. Now, we
perform that eviction 'on demand' in buf_LRU_get_free_block().
The parameter innodb_lru_scan_depth(srv_LRU_scan_depth) is used as follows:
* When the buffer pool is being shrunk in buf_pool_t::withdraw_blocks()
* As a buf_pool.free limit in buf_LRU_list_batch() for terminating
the flushing that is initiated e.g., by buf_LRU_get_free_block()
The parameter also used to serve as an initial limit for unzip_LRU
eviction (evicting uncompressed page frames while retaining
ROW_FORMAT=COMPRESSED pages), but now we will use a hard-coded limit
of 100 or unlimited for invoking buf_LRU_scan_and_free_block().
The status variables will be changed as follows:
innodb_buffer_pool_pages_flushed: This includes also the count of
innodb_buffer_pool_pages_LRU_flushed and should work reliably,
updated one by one in buf_flush_page() to give more real-time
statistics. The function buf_flush_stats(), which we are removing,
was not called in every code path. For both counters, we will use
regular variables that are incremented in a critical section of
buf_pool.mutex. Note that show_innodb_vars() directly links to the
variables, and reads of the counters will *not* be protected by
buf_pool.mutex, so you cannot get a consistent snapshot of both variables.
The following INFORMATION_SCHEMA.INNODB_METRICS counters will be
removed, because the page cleaner no longer deals with writing or
evicting least recently used pages, and because the single-page writes
have been removed:
* buffer_LRU_batch_flush_avg_time_slot
* buffer_LRU_batch_flush_avg_time_thread
* buffer_LRU_batch_flush_avg_time_est
* buffer_LRU_batch_flush_avg_pass
* buffer_LRU_single_flush_scanned
* buffer_LRU_single_flush_num_scan
* buffer_LRU_single_flush_scanned_per_call
When moving to a single buffer pool instance in MDEV-15058, we missed
some opportunity to simplify the buf_flush_page_cleaner thread. It was
unnecessarily using a mutex and some complex data structures, even
though we always have a single page cleaner thread.
Furthermore, the buf_flush_page_cleaner thread had separate 'recovery'
and 'shutdown' modes where it was waiting to be triggered by some
other thread, adding unnecessary latency and potential for hangs in
relatively rarely executed startup or shutdown code.
The page cleaner was also running two kinds of batches in an
interleaved fashion: "LRU flush" (writing out some least recently used
pages and evicting them on write completion) and the normal batches
that aim to increase the MIN(oldest_modification) in the buffer pool,
to help the log checkpoint advance.
The buf_pool.flush_list flushing was being blocked by
buf_block_t::lock for no good reason. Furthermore, if the FIL_PAGE_LSN
of a page is ahead of log_sys.get_flushed_lsn(), that is, what has
been persistently written to the redo log, we would trigger a log
flush and then resume the page flushing. This would unnecessarily
limit the performance of the page cleaner thread and trigger the
infamous messages "InnoDB: page_cleaner: 1000ms intended loop took 4450ms.
The settings might not be optimal" that were suppressed in
commit d1ab89037a518fcffbc50c24e4bd94e4ec33aed0 unless log_warnings>2.
Our revised algorithm will make log_sys.get_flushed_lsn() advance at
the start of buf_flush_lists(), and then execute a 'best effort' to
write out all pages. The flush batches will skip pages that were modified
since the log was written, or are are currently exclusively locked.
The MDEV-13670 message "page_cleaner: 1000ms intended loop took" message
will be removed, because by design, the buf_flush_page_cleaner() should
not be blocked during a batch for extended periods of time.
We will remove the single-page flushing altogether. Related to this,
the debug parameter innodb_doublewrite_batch_size will be removed,
because all of the doublewrite buffer will be used for flushing
batches. If a page needs to be evicted from the buffer pool and all
100 least recently used pages in the buffer pool have unflushed
changes, buf_LRU_get_free_block() will execute buf_flush_lists() to
write out and evict innodb_lru_flush_size pages. At most one thread
will execute buf_flush_lists() in buf_LRU_get_free_block(); other
threads will wait for that LRU flushing batch to finish.
To improve concurrency, we will replace the InnoDB ib_mutex_t and
os_event_t native mutexes and condition variables in this area of code.
Most notably, this means that the buffer pool mutex (buf_pool.mutex)
is no longer instrumented via any InnoDB interfaces. It will continue
to be instrumented via PERFORMANCE_SCHEMA.
For now, both buf_pool.flush_list_mutex and buf_pool.mutex will be
declared with MY_MUTEX_INIT_FAST (PTHREAD_MUTEX_ADAPTIVE_NP). The critical
sections of buf_pool.flush_list_mutex should be shorter than those for
buf_pool.mutex, because in the worst case, they cover a linear scan of
buf_pool.flush_list, while the worst case of a critical section of
buf_pool.mutex covers a linear scan of the potentially much longer
buf_pool.LRU list.
mysql_mutex_is_owner(), safe_mutex_is_owner(): New predicate, usable
with SAFE_MUTEX. Some InnoDB debug assertions need this predicate
instead of mysql_mutex_assert_owner() or mysql_mutex_assert_not_owner().
buf_pool_t::n_flush_LRU, buf_pool_t::n_flush_list:
Replaces buf_pool_t::init_flush[] and buf_pool_t::n_flush[].
The number of active flush operations.
buf_pool_t::mutex, buf_pool_t::flush_list_mutex: Use mysql_mutex_t
instead of ib_mutex_t, to have native mutexes with PERFORMANCE_SCHEMA
and SAFE_MUTEX instrumentation.
buf_pool_t::done_flush_LRU: Condition variable for !n_flush_LRU.
buf_pool_t::done_flush_list: Condition variable for !n_flush_list.
buf_pool_t::do_flush_list: Condition variable to wake up the
buf_flush_page_cleaner when a log checkpoint needs to be written
or the server is being shut down. Replaces buf_flush_event.
We will keep using timed waits (the page cleaner thread will wake
_at least_ once per second), because the calculations for
innodb_adaptive_flushing depend on fixed time intervals.
buf_dblwr: Allocate statically, and move all code to member functions.
Use a native mutex and condition variable. Remove code to deal with
single-page flushing.
buf_dblwr_check_block(): Make the check debug-only. We were spending
a significant amount of execution time in page_simple_validate_new().
flush_counters_t::unzip_LRU_evicted: Remove.
IORequest: Make more members const. FIXME: m_fil_node should be removed.
buf_flush_sync_lsn: Protect by std::atomic, not page_cleaner.mutex
(which we are removing).
page_cleaner_slot_t, page_cleaner_t: Remove many redundant members.
pc_request_flush_slot(): Replaces pc_request() and pc_flush_slot().
recv_writer_thread: Remove. Recovery works just fine without it, if we
simply invoke buf_flush_sync() at the end of each batch in
recv_sys_t::apply().
recv_recovery_from_checkpoint_finish(): Remove. We can simply call
recv_sys.debug_free() directly.
srv_started_redo: Replaces srv_start_state.
SRV_SHUTDOWN_FLUSH_PHASE: Remove. logs_empty_and_mark_files_at_shutdown()
can communicate with the normal page cleaner loop via the new function
flush_buffer_pool().
buf_flush_remove(): Assert that the calling thread is holding
buf_pool.flush_list_mutex. This removes unnecessary mutex operations
from buf_flush_remove_pages() and buf_flush_dirty_pages(),
which replace buf_LRU_flush_or_remove_pages().
buf_flush_lists(): Renamed from buf_flush_batch(), with simplified
interface. Return the number of flushed pages. Clarified comments and
renamed min_n to max_n. Identify LRU batch by lsn=0. Merge all the functions
buf_flush_start(), buf_flush_batch(), buf_flush_end() directly to this
function, which was their only caller, and remove 2 unnecessary
buf_pool.mutex release/re-acquisition that we used to perform around
the buf_flush_batch() call. At the start, if not all log has been
durably written, wait for a background task to do it, or start a new
task to do it. This allows the log write to run concurrently with our
page flushing batch. Any pages that were skipped due to too recent
FIL_PAGE_LSN or due to them being latched by a writer should be flushed
during the next batch, unless there are further modifications to those
pages. It is possible that a page that we must flush due to small
oldest_modification also carries a recent FIL_PAGE_LSN or is being
constantly modified. In the worst case, all writers would then end up
waiting in log_free_check() to allow the flushing and the checkpoint
to complete.
buf_do_flush_list_batch(): Clarify comments, and rename min_n to max_n.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_flush_space(): Auxiliary function to look up a tablespace for
page flushing.
buf_flush_page(): Defer the computation of space->full_crc32(). Never
call log_write_up_to(), but instead skip persistent pages whose latest
modification (FIL_PAGE_LSN) is newer than the redo log. Also skip
pages on which we cannot acquire a shared latch without waiting.
buf_flush_try_neighbors(): Do not bother checking buf_fix_count
because buf_flush_page() will no longer wait for the page latch.
Take the tablespace as a parameter, and only execute this function
when innodb_flush_neighbors>0. Avoid repeated calls of page_id_t::fold().
buf_flush_relocate_on_flush_list(): Declare as cold, and push down
a condition from the callers.
buf_flush_check_neighbor(): Take id.fold() as a parameter.
buf_flush_sync(): Ensure that the buf_pool.flush_list is empty,
because the flushing batch will skip pages whose modifications have
not yet been written to the log or were latched for modification.
buf_free_from_unzip_LRU_list_batch(): Remove redundant local variables.
buf_flush_LRU_list_batch(): Let the caller buf_do_LRU_batch() initialize
the counters, and report n->evicted.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_do_LRU_batch(): Return the number of pages flushed.
buf_LRU_free_page(): Only release and re-acquire buf_pool.mutex if
adaptive hash index entries are pointing to the block.
buf_LRU_get_free_block(): Do not wake up the page cleaner, because it
will no longer perform any useful work for us, and we do not want it
to compete for I/O while buf_flush_lists(innodb_lru_flush_size, 0)
writes out and evicts at most innodb_lru_flush_size pages. (The
function buf_do_LRU_batch() may complete after writing fewer pages if
more than innodb_lru_scan_depth pages end up in buf_pool.free list.)
Eliminate some mutex release-acquire cycles, and wait for the LRU
flush batch to complete before rescanning.
buf_LRU_check_size_of_non_data_objects(): Simplify the code.
buf_page_write_complete(): Remove the parameter evict, and always
evict pages that were part of an LRU flush.
buf_page_create(): Take a pre-allocated page as a parameter.
buf_pool_t::free_block(): Free a pre-allocated block.
recv_sys_t::recover_low(), recv_sys_t::apply(): Preallocate the block
while not holding recv_sys.mutex. During page allocation, we may
initiate a page flush, which in turn may initiate a log flush, which
would require acquiring log_sys.mutex, which should always be acquired
before recv_sys.mutex in order to avoid deadlocks. Therefore, we must
not be holding recv_sys.mutex while allocating a buffer pool block.
BtrBulk::logFreeCheck(): Skip a redundant condition.
row_undo_step(): Do not invoke srv_inc_activity_count() for every row
that is being rolled back. It should suffice to invoke the function in
trx_flush_log_if_needed() during trx_t::commit_in_memory() when the
rollback completes.
sync_check_enable(): Remove. We will enable innodb_sync_debug from the
very beginning.
Reviewed by: Vladislav Vaintroub
5 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-23399: Performance regression with write workloads
The buffer pool refactoring in MDEV-15053 and MDEV-22871 shifted
the performance bottleneck to the page flushing.
The configuration parameters will be changed as follows:
innodb_lru_flush_size=32 (new: how many pages to flush on LRU eviction)
innodb_lru_scan_depth=1536 (old: 1024)
innodb_max_dirty_pages_pct=90 (old: 75)
innodb_max_dirty_pages_pct_lwm=75 (old: 0)
Note: The parameter innodb_lru_scan_depth will only affect LRU
eviction of buffer pool pages when a new page is being allocated. The
page cleaner thread will no longer evict any pages. It used to
guarantee that some pages will remain free in the buffer pool. Now, we
perform that eviction 'on demand' in buf_LRU_get_free_block().
The parameter innodb_lru_scan_depth(srv_LRU_scan_depth) is used as follows:
* When the buffer pool is being shrunk in buf_pool_t::withdraw_blocks()
* As a buf_pool.free limit in buf_LRU_list_batch() for terminating
the flushing that is initiated e.g., by buf_LRU_get_free_block()
The parameter also used to serve as an initial limit for unzip_LRU
eviction (evicting uncompressed page frames while retaining
ROW_FORMAT=COMPRESSED pages), but now we will use a hard-coded limit
of 100 or unlimited for invoking buf_LRU_scan_and_free_block().
The status variables will be changed as follows:
innodb_buffer_pool_pages_flushed: This includes also the count of
innodb_buffer_pool_pages_LRU_flushed and should work reliably,
updated one by one in buf_flush_page() to give more real-time
statistics. The function buf_flush_stats(), which we are removing,
was not called in every code path. For both counters, we will use
regular variables that are incremented in a critical section of
buf_pool.mutex. Note that show_innodb_vars() directly links to the
variables, and reads of the counters will *not* be protected by
buf_pool.mutex, so you cannot get a consistent snapshot of both variables.
The following INFORMATION_SCHEMA.INNODB_METRICS counters will be
removed, because the page cleaner no longer deals with writing or
evicting least recently used pages, and because the single-page writes
have been removed:
* buffer_LRU_batch_flush_avg_time_slot
* buffer_LRU_batch_flush_avg_time_thread
* buffer_LRU_batch_flush_avg_time_est
* buffer_LRU_batch_flush_avg_pass
* buffer_LRU_single_flush_scanned
* buffer_LRU_single_flush_num_scan
* buffer_LRU_single_flush_scanned_per_call
When moving to a single buffer pool instance in MDEV-15058, we missed
some opportunity to simplify the buf_flush_page_cleaner thread. It was
unnecessarily using a mutex and some complex data structures, even
though we always have a single page cleaner thread.
Furthermore, the buf_flush_page_cleaner thread had separate 'recovery'
and 'shutdown' modes where it was waiting to be triggered by some
other thread, adding unnecessary latency and potential for hangs in
relatively rarely executed startup or shutdown code.
The page cleaner was also running two kinds of batches in an
interleaved fashion: "LRU flush" (writing out some least recently used
pages and evicting them on write completion) and the normal batches
that aim to increase the MIN(oldest_modification) in the buffer pool,
to help the log checkpoint advance.
The buf_pool.flush_list flushing was being blocked by
buf_block_t::lock for no good reason. Furthermore, if the FIL_PAGE_LSN
of a page is ahead of log_sys.get_flushed_lsn(), that is, what has
been persistently written to the redo log, we would trigger a log
flush and then resume the page flushing. This would unnecessarily
limit the performance of the page cleaner thread and trigger the
infamous messages "InnoDB: page_cleaner: 1000ms intended loop took 4450ms.
The settings might not be optimal" that were suppressed in
commit d1ab89037a518fcffbc50c24e4bd94e4ec33aed0 unless log_warnings>2.
Our revised algorithm will make log_sys.get_flushed_lsn() advance at
the start of buf_flush_lists(), and then execute a 'best effort' to
write out all pages. The flush batches will skip pages that were modified
since the log was written, or are are currently exclusively locked.
The MDEV-13670 message "page_cleaner: 1000ms intended loop took" message
will be removed, because by design, the buf_flush_page_cleaner() should
not be blocked during a batch for extended periods of time.
We will remove the single-page flushing altogether. Related to this,
the debug parameter innodb_doublewrite_batch_size will be removed,
because all of the doublewrite buffer will be used for flushing
batches. If a page needs to be evicted from the buffer pool and all
100 least recently used pages in the buffer pool have unflushed
changes, buf_LRU_get_free_block() will execute buf_flush_lists() to
write out and evict innodb_lru_flush_size pages. At most one thread
will execute buf_flush_lists() in buf_LRU_get_free_block(); other
threads will wait for that LRU flushing batch to finish.
To improve concurrency, we will replace the InnoDB ib_mutex_t and
os_event_t native mutexes and condition variables in this area of code.
Most notably, this means that the buffer pool mutex (buf_pool.mutex)
is no longer instrumented via any InnoDB interfaces. It will continue
to be instrumented via PERFORMANCE_SCHEMA.
For now, both buf_pool.flush_list_mutex and buf_pool.mutex will be
declared with MY_MUTEX_INIT_FAST (PTHREAD_MUTEX_ADAPTIVE_NP). The critical
sections of buf_pool.flush_list_mutex should be shorter than those for
buf_pool.mutex, because in the worst case, they cover a linear scan of
buf_pool.flush_list, while the worst case of a critical section of
buf_pool.mutex covers a linear scan of the potentially much longer
buf_pool.LRU list.
mysql_mutex_is_owner(), safe_mutex_is_owner(): New predicate, usable
with SAFE_MUTEX. Some InnoDB debug assertions need this predicate
instead of mysql_mutex_assert_owner() or mysql_mutex_assert_not_owner().
buf_pool_t::n_flush_LRU, buf_pool_t::n_flush_list:
Replaces buf_pool_t::init_flush[] and buf_pool_t::n_flush[].
The number of active flush operations.
buf_pool_t::mutex, buf_pool_t::flush_list_mutex: Use mysql_mutex_t
instead of ib_mutex_t, to have native mutexes with PERFORMANCE_SCHEMA
and SAFE_MUTEX instrumentation.
buf_pool_t::done_flush_LRU: Condition variable for !n_flush_LRU.
buf_pool_t::done_flush_list: Condition variable for !n_flush_list.
buf_pool_t::do_flush_list: Condition variable to wake up the
buf_flush_page_cleaner when a log checkpoint needs to be written
or the server is being shut down. Replaces buf_flush_event.
We will keep using timed waits (the page cleaner thread will wake
_at least_ once per second), because the calculations for
innodb_adaptive_flushing depend on fixed time intervals.
buf_dblwr: Allocate statically, and move all code to member functions.
Use a native mutex and condition variable. Remove code to deal with
single-page flushing.
buf_dblwr_check_block(): Make the check debug-only. We were spending
a significant amount of execution time in page_simple_validate_new().
flush_counters_t::unzip_LRU_evicted: Remove.
IORequest: Make more members const. FIXME: m_fil_node should be removed.
buf_flush_sync_lsn: Protect by std::atomic, not page_cleaner.mutex
(which we are removing).
page_cleaner_slot_t, page_cleaner_t: Remove many redundant members.
pc_request_flush_slot(): Replaces pc_request() and pc_flush_slot().
recv_writer_thread: Remove. Recovery works just fine without it, if we
simply invoke buf_flush_sync() at the end of each batch in
recv_sys_t::apply().
recv_recovery_from_checkpoint_finish(): Remove. We can simply call
recv_sys.debug_free() directly.
srv_started_redo: Replaces srv_start_state.
SRV_SHUTDOWN_FLUSH_PHASE: Remove. logs_empty_and_mark_files_at_shutdown()
can communicate with the normal page cleaner loop via the new function
flush_buffer_pool().
buf_flush_remove(): Assert that the calling thread is holding
buf_pool.flush_list_mutex. This removes unnecessary mutex operations
from buf_flush_remove_pages() and buf_flush_dirty_pages(),
which replace buf_LRU_flush_or_remove_pages().
buf_flush_lists(): Renamed from buf_flush_batch(), with simplified
interface. Return the number of flushed pages. Clarified comments and
renamed min_n to max_n. Identify LRU batch by lsn=0. Merge all the functions
buf_flush_start(), buf_flush_batch(), buf_flush_end() directly to this
function, which was their only caller, and remove 2 unnecessary
buf_pool.mutex release/re-acquisition that we used to perform around
the buf_flush_batch() call. At the start, if not all log has been
durably written, wait for a background task to do it, or start a new
task to do it. This allows the log write to run concurrently with our
page flushing batch. Any pages that were skipped due to too recent
FIL_PAGE_LSN or due to them being latched by a writer should be flushed
during the next batch, unless there are further modifications to those
pages. It is possible that a page that we must flush due to small
oldest_modification also carries a recent FIL_PAGE_LSN or is being
constantly modified. In the worst case, all writers would then end up
waiting in log_free_check() to allow the flushing and the checkpoint
to complete.
buf_do_flush_list_batch(): Clarify comments, and rename min_n to max_n.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_flush_space(): Auxiliary function to look up a tablespace for
page flushing.
buf_flush_page(): Defer the computation of space->full_crc32(). Never
call log_write_up_to(), but instead skip persistent pages whose latest
modification (FIL_PAGE_LSN) is newer than the redo log. Also skip
pages on which we cannot acquire a shared latch without waiting.
buf_flush_try_neighbors(): Do not bother checking buf_fix_count
because buf_flush_page() will no longer wait for the page latch.
Take the tablespace as a parameter, and only execute this function
when innodb_flush_neighbors>0. Avoid repeated calls of page_id_t::fold().
buf_flush_relocate_on_flush_list(): Declare as cold, and push down
a condition from the callers.
buf_flush_check_neighbor(): Take id.fold() as a parameter.
buf_flush_sync(): Ensure that the buf_pool.flush_list is empty,
because the flushing batch will skip pages whose modifications have
not yet been written to the log or were latched for modification.
buf_free_from_unzip_LRU_list_batch(): Remove redundant local variables.
buf_flush_LRU_list_batch(): Let the caller buf_do_LRU_batch() initialize
the counters, and report n->evicted.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_do_LRU_batch(): Return the number of pages flushed.
buf_LRU_free_page(): Only release and re-acquire buf_pool.mutex if
adaptive hash index entries are pointing to the block.
buf_LRU_get_free_block(): Do not wake up the page cleaner, because it
will no longer perform any useful work for us, and we do not want it
to compete for I/O while buf_flush_lists(innodb_lru_flush_size, 0)
writes out and evicts at most innodb_lru_flush_size pages. (The
function buf_do_LRU_batch() may complete after writing fewer pages if
more than innodb_lru_scan_depth pages end up in buf_pool.free list.)
Eliminate some mutex release-acquire cycles, and wait for the LRU
flush batch to complete before rescanning.
buf_LRU_check_size_of_non_data_objects(): Simplify the code.
buf_page_write_complete(): Remove the parameter evict, and always
evict pages that were part of an LRU flush.
buf_page_create(): Take a pre-allocated page as a parameter.
buf_pool_t::free_block(): Free a pre-allocated block.
recv_sys_t::recover_low(), recv_sys_t::apply(): Preallocate the block
while not holding recv_sys.mutex. During page allocation, we may
initiate a page flush, which in turn may initiate a log flush, which
would require acquiring log_sys.mutex, which should always be acquired
before recv_sys.mutex in order to avoid deadlocks. Therefore, we must
not be holding recv_sys.mutex while allocating a buffer pool block.
BtrBulk::logFreeCheck(): Skip a redundant condition.
row_undo_step(): Do not invoke srv_inc_activity_count() for every row
that is being rolled back. It should suffice to invoke the function in
trx_flush_log_if_needed() during trx_t::commit_in_memory() when the
rollback completes.
sync_check_enable(): Remove. We will enable innodb_sync_debug from the
very beginning.
Reviewed by: Vladislav Vaintroub
5 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-23399: Performance regression with write workloads
The buffer pool refactoring in MDEV-15053 and MDEV-22871 shifted
the performance bottleneck to the page flushing.
The configuration parameters will be changed as follows:
innodb_lru_flush_size=32 (new: how many pages to flush on LRU eviction)
innodb_lru_scan_depth=1536 (old: 1024)
innodb_max_dirty_pages_pct=90 (old: 75)
innodb_max_dirty_pages_pct_lwm=75 (old: 0)
Note: The parameter innodb_lru_scan_depth will only affect LRU
eviction of buffer pool pages when a new page is being allocated. The
page cleaner thread will no longer evict any pages. It used to
guarantee that some pages will remain free in the buffer pool. Now, we
perform that eviction 'on demand' in buf_LRU_get_free_block().
The parameter innodb_lru_scan_depth(srv_LRU_scan_depth) is used as follows:
* When the buffer pool is being shrunk in buf_pool_t::withdraw_blocks()
* As a buf_pool.free limit in buf_LRU_list_batch() for terminating
the flushing that is initiated e.g., by buf_LRU_get_free_block()
The parameter also used to serve as an initial limit for unzip_LRU
eviction (evicting uncompressed page frames while retaining
ROW_FORMAT=COMPRESSED pages), but now we will use a hard-coded limit
of 100 or unlimited for invoking buf_LRU_scan_and_free_block().
The status variables will be changed as follows:
innodb_buffer_pool_pages_flushed: This includes also the count of
innodb_buffer_pool_pages_LRU_flushed and should work reliably,
updated one by one in buf_flush_page() to give more real-time
statistics. The function buf_flush_stats(), which we are removing,
was not called in every code path. For both counters, we will use
regular variables that are incremented in a critical section of
buf_pool.mutex. Note that show_innodb_vars() directly links to the
variables, and reads of the counters will *not* be protected by
buf_pool.mutex, so you cannot get a consistent snapshot of both variables.
The following INFORMATION_SCHEMA.INNODB_METRICS counters will be
removed, because the page cleaner no longer deals with writing or
evicting least recently used pages, and because the single-page writes
have been removed:
* buffer_LRU_batch_flush_avg_time_slot
* buffer_LRU_batch_flush_avg_time_thread
* buffer_LRU_batch_flush_avg_time_est
* buffer_LRU_batch_flush_avg_pass
* buffer_LRU_single_flush_scanned
* buffer_LRU_single_flush_num_scan
* buffer_LRU_single_flush_scanned_per_call
When moving to a single buffer pool instance in MDEV-15058, we missed
some opportunity to simplify the buf_flush_page_cleaner thread. It was
unnecessarily using a mutex and some complex data structures, even
though we always have a single page cleaner thread.
Furthermore, the buf_flush_page_cleaner thread had separate 'recovery'
and 'shutdown' modes where it was waiting to be triggered by some
other thread, adding unnecessary latency and potential for hangs in
relatively rarely executed startup or shutdown code.
The page cleaner was also running two kinds of batches in an
interleaved fashion: "LRU flush" (writing out some least recently used
pages and evicting them on write completion) and the normal batches
that aim to increase the MIN(oldest_modification) in the buffer pool,
to help the log checkpoint advance.
The buf_pool.flush_list flushing was being blocked by
buf_block_t::lock for no good reason. Furthermore, if the FIL_PAGE_LSN
of a page is ahead of log_sys.get_flushed_lsn(), that is, what has
been persistently written to the redo log, we would trigger a log
flush and then resume the page flushing. This would unnecessarily
limit the performance of the page cleaner thread and trigger the
infamous messages "InnoDB: page_cleaner: 1000ms intended loop took 4450ms.
The settings might not be optimal" that were suppressed in
commit d1ab89037a518fcffbc50c24e4bd94e4ec33aed0 unless log_warnings>2.
Our revised algorithm will make log_sys.get_flushed_lsn() advance at
the start of buf_flush_lists(), and then execute a 'best effort' to
write out all pages. The flush batches will skip pages that were modified
since the log was written, or are are currently exclusively locked.
The MDEV-13670 message "page_cleaner: 1000ms intended loop took" message
will be removed, because by design, the buf_flush_page_cleaner() should
not be blocked during a batch for extended periods of time.
We will remove the single-page flushing altogether. Related to this,
the debug parameter innodb_doublewrite_batch_size will be removed,
because all of the doublewrite buffer will be used for flushing
batches. If a page needs to be evicted from the buffer pool and all
100 least recently used pages in the buffer pool have unflushed
changes, buf_LRU_get_free_block() will execute buf_flush_lists() to
write out and evict innodb_lru_flush_size pages. At most one thread
will execute buf_flush_lists() in buf_LRU_get_free_block(); other
threads will wait for that LRU flushing batch to finish.
To improve concurrency, we will replace the InnoDB ib_mutex_t and
os_event_t native mutexes and condition variables in this area of code.
Most notably, this means that the buffer pool mutex (buf_pool.mutex)
is no longer instrumented via any InnoDB interfaces. It will continue
to be instrumented via PERFORMANCE_SCHEMA.
For now, both buf_pool.flush_list_mutex and buf_pool.mutex will be
declared with MY_MUTEX_INIT_FAST (PTHREAD_MUTEX_ADAPTIVE_NP). The critical
sections of buf_pool.flush_list_mutex should be shorter than those for
buf_pool.mutex, because in the worst case, they cover a linear scan of
buf_pool.flush_list, while the worst case of a critical section of
buf_pool.mutex covers a linear scan of the potentially much longer
buf_pool.LRU list.
mysql_mutex_is_owner(), safe_mutex_is_owner(): New predicate, usable
with SAFE_MUTEX. Some InnoDB debug assertions need this predicate
instead of mysql_mutex_assert_owner() or mysql_mutex_assert_not_owner().
buf_pool_t::n_flush_LRU, buf_pool_t::n_flush_list:
Replaces buf_pool_t::init_flush[] and buf_pool_t::n_flush[].
The number of active flush operations.
buf_pool_t::mutex, buf_pool_t::flush_list_mutex: Use mysql_mutex_t
instead of ib_mutex_t, to have native mutexes with PERFORMANCE_SCHEMA
and SAFE_MUTEX instrumentation.
buf_pool_t::done_flush_LRU: Condition variable for !n_flush_LRU.
buf_pool_t::done_flush_list: Condition variable for !n_flush_list.
buf_pool_t::do_flush_list: Condition variable to wake up the
buf_flush_page_cleaner when a log checkpoint needs to be written
or the server is being shut down. Replaces buf_flush_event.
We will keep using timed waits (the page cleaner thread will wake
_at least_ once per second), because the calculations for
innodb_adaptive_flushing depend on fixed time intervals.
buf_dblwr: Allocate statically, and move all code to member functions.
Use a native mutex and condition variable. Remove code to deal with
single-page flushing.
buf_dblwr_check_block(): Make the check debug-only. We were spending
a significant amount of execution time in page_simple_validate_new().
flush_counters_t::unzip_LRU_evicted: Remove.
IORequest: Make more members const. FIXME: m_fil_node should be removed.
buf_flush_sync_lsn: Protect by std::atomic, not page_cleaner.mutex
(which we are removing).
page_cleaner_slot_t, page_cleaner_t: Remove many redundant members.
pc_request_flush_slot(): Replaces pc_request() and pc_flush_slot().
recv_writer_thread: Remove. Recovery works just fine without it, if we
simply invoke buf_flush_sync() at the end of each batch in
recv_sys_t::apply().
recv_recovery_from_checkpoint_finish(): Remove. We can simply call
recv_sys.debug_free() directly.
srv_started_redo: Replaces srv_start_state.
SRV_SHUTDOWN_FLUSH_PHASE: Remove. logs_empty_and_mark_files_at_shutdown()
can communicate with the normal page cleaner loop via the new function
flush_buffer_pool().
buf_flush_remove(): Assert that the calling thread is holding
buf_pool.flush_list_mutex. This removes unnecessary mutex operations
from buf_flush_remove_pages() and buf_flush_dirty_pages(),
which replace buf_LRU_flush_or_remove_pages().
buf_flush_lists(): Renamed from buf_flush_batch(), with simplified
interface. Return the number of flushed pages. Clarified comments and
renamed min_n to max_n. Identify LRU batch by lsn=0. Merge all the functions
buf_flush_start(), buf_flush_batch(), buf_flush_end() directly to this
function, which was their only caller, and remove 2 unnecessary
buf_pool.mutex release/re-acquisition that we used to perform around
the buf_flush_batch() call. At the start, if not all log has been
durably written, wait for a background task to do it, or start a new
task to do it. This allows the log write to run concurrently with our
page flushing batch. Any pages that were skipped due to too recent
FIL_PAGE_LSN or due to them being latched by a writer should be flushed
during the next batch, unless there are further modifications to those
pages. It is possible that a page that we must flush due to small
oldest_modification also carries a recent FIL_PAGE_LSN or is being
constantly modified. In the worst case, all writers would then end up
waiting in log_free_check() to allow the flushing and the checkpoint
to complete.
buf_do_flush_list_batch(): Clarify comments, and rename min_n to max_n.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_flush_space(): Auxiliary function to look up a tablespace for
page flushing.
buf_flush_page(): Defer the computation of space->full_crc32(). Never
call log_write_up_to(), but instead skip persistent pages whose latest
modification (FIL_PAGE_LSN) is newer than the redo log. Also skip
pages on which we cannot acquire a shared latch without waiting.
buf_flush_try_neighbors(): Do not bother checking buf_fix_count
because buf_flush_page() will no longer wait for the page latch.
Take the tablespace as a parameter, and only execute this function
when innodb_flush_neighbors>0. Avoid repeated calls of page_id_t::fold().
buf_flush_relocate_on_flush_list(): Declare as cold, and push down
a condition from the callers.
buf_flush_check_neighbor(): Take id.fold() as a parameter.
buf_flush_sync(): Ensure that the buf_pool.flush_list is empty,
because the flushing batch will skip pages whose modifications have
not yet been written to the log or were latched for modification.
buf_free_from_unzip_LRU_list_batch(): Remove redundant local variables.
buf_flush_LRU_list_batch(): Let the caller buf_do_LRU_batch() initialize
the counters, and report n->evicted.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_do_LRU_batch(): Return the number of pages flushed.
buf_LRU_free_page(): Only release and re-acquire buf_pool.mutex if
adaptive hash index entries are pointing to the block.
buf_LRU_get_free_block(): Do not wake up the page cleaner, because it
will no longer perform any useful work for us, and we do not want it
to compete for I/O while buf_flush_lists(innodb_lru_flush_size, 0)
writes out and evicts at most innodb_lru_flush_size pages. (The
function buf_do_LRU_batch() may complete after writing fewer pages if
more than innodb_lru_scan_depth pages end up in buf_pool.free list.)
Eliminate some mutex release-acquire cycles, and wait for the LRU
flush batch to complete before rescanning.
buf_LRU_check_size_of_non_data_objects(): Simplify the code.
buf_page_write_complete(): Remove the parameter evict, and always
evict pages that were part of an LRU flush.
buf_page_create(): Take a pre-allocated page as a parameter.
buf_pool_t::free_block(): Free a pre-allocated block.
recv_sys_t::recover_low(), recv_sys_t::apply(): Preallocate the block
while not holding recv_sys.mutex. During page allocation, we may
initiate a page flush, which in turn may initiate a log flush, which
would require acquiring log_sys.mutex, which should always be acquired
before recv_sys.mutex in order to avoid deadlocks. Therefore, we must
not be holding recv_sys.mutex while allocating a buffer pool block.
BtrBulk::logFreeCheck(): Skip a redundant condition.
row_undo_step(): Do not invoke srv_inc_activity_count() for every row
that is being rolled back. It should suffice to invoke the function in
trx_flush_log_if_needed() during trx_t::commit_in_memory() when the
rollback completes.
sync_check_enable(): Remove. We will enable innodb_sync_debug from the
very beginning.
Reviewed by: Vladislav Vaintroub
5 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-23399: Performance regression with write workloads
The buffer pool refactoring in MDEV-15053 and MDEV-22871 shifted
the performance bottleneck to the page flushing.
The configuration parameters will be changed as follows:
innodb_lru_flush_size=32 (new: how many pages to flush on LRU eviction)
innodb_lru_scan_depth=1536 (old: 1024)
innodb_max_dirty_pages_pct=90 (old: 75)
innodb_max_dirty_pages_pct_lwm=75 (old: 0)
Note: The parameter innodb_lru_scan_depth will only affect LRU
eviction of buffer pool pages when a new page is being allocated. The
page cleaner thread will no longer evict any pages. It used to
guarantee that some pages will remain free in the buffer pool. Now, we
perform that eviction 'on demand' in buf_LRU_get_free_block().
The parameter innodb_lru_scan_depth(srv_LRU_scan_depth) is used as follows:
* When the buffer pool is being shrunk in buf_pool_t::withdraw_blocks()
* As a buf_pool.free limit in buf_LRU_list_batch() for terminating
the flushing that is initiated e.g., by buf_LRU_get_free_block()
The parameter also used to serve as an initial limit for unzip_LRU
eviction (evicting uncompressed page frames while retaining
ROW_FORMAT=COMPRESSED pages), but now we will use a hard-coded limit
of 100 or unlimited for invoking buf_LRU_scan_and_free_block().
The status variables will be changed as follows:
innodb_buffer_pool_pages_flushed: This includes also the count of
innodb_buffer_pool_pages_LRU_flushed and should work reliably,
updated one by one in buf_flush_page() to give more real-time
statistics. The function buf_flush_stats(), which we are removing,
was not called in every code path. For both counters, we will use
regular variables that are incremented in a critical section of
buf_pool.mutex. Note that show_innodb_vars() directly links to the
variables, and reads of the counters will *not* be protected by
buf_pool.mutex, so you cannot get a consistent snapshot of both variables.
The following INFORMATION_SCHEMA.INNODB_METRICS counters will be
removed, because the page cleaner no longer deals with writing or
evicting least recently used pages, and because the single-page writes
have been removed:
* buffer_LRU_batch_flush_avg_time_slot
* buffer_LRU_batch_flush_avg_time_thread
* buffer_LRU_batch_flush_avg_time_est
* buffer_LRU_batch_flush_avg_pass
* buffer_LRU_single_flush_scanned
* buffer_LRU_single_flush_num_scan
* buffer_LRU_single_flush_scanned_per_call
When moving to a single buffer pool instance in MDEV-15058, we missed
some opportunity to simplify the buf_flush_page_cleaner thread. It was
unnecessarily using a mutex and some complex data structures, even
though we always have a single page cleaner thread.
Furthermore, the buf_flush_page_cleaner thread had separate 'recovery'
and 'shutdown' modes where it was waiting to be triggered by some
other thread, adding unnecessary latency and potential for hangs in
relatively rarely executed startup or shutdown code.
The page cleaner was also running two kinds of batches in an
interleaved fashion: "LRU flush" (writing out some least recently used
pages and evicting them on write completion) and the normal batches
that aim to increase the MIN(oldest_modification) in the buffer pool,
to help the log checkpoint advance.
The buf_pool.flush_list flushing was being blocked by
buf_block_t::lock for no good reason. Furthermore, if the FIL_PAGE_LSN
of a page is ahead of log_sys.get_flushed_lsn(), that is, what has
been persistently written to the redo log, we would trigger a log
flush and then resume the page flushing. This would unnecessarily
limit the performance of the page cleaner thread and trigger the
infamous messages "InnoDB: page_cleaner: 1000ms intended loop took 4450ms.
The settings might not be optimal" that were suppressed in
commit d1ab89037a518fcffbc50c24e4bd94e4ec33aed0 unless log_warnings>2.
Our revised algorithm will make log_sys.get_flushed_lsn() advance at
the start of buf_flush_lists(), and then execute a 'best effort' to
write out all pages. The flush batches will skip pages that were modified
since the log was written, or are are currently exclusively locked.
The MDEV-13670 message "page_cleaner: 1000ms intended loop took" message
will be removed, because by design, the buf_flush_page_cleaner() should
not be blocked during a batch for extended periods of time.
We will remove the single-page flushing altogether. Related to this,
the debug parameter innodb_doublewrite_batch_size will be removed,
because all of the doublewrite buffer will be used for flushing
batches. If a page needs to be evicted from the buffer pool and all
100 least recently used pages in the buffer pool have unflushed
changes, buf_LRU_get_free_block() will execute buf_flush_lists() to
write out and evict innodb_lru_flush_size pages. At most one thread
will execute buf_flush_lists() in buf_LRU_get_free_block(); other
threads will wait for that LRU flushing batch to finish.
To improve concurrency, we will replace the InnoDB ib_mutex_t and
os_event_t native mutexes and condition variables in this area of code.
Most notably, this means that the buffer pool mutex (buf_pool.mutex)
is no longer instrumented via any InnoDB interfaces. It will continue
to be instrumented via PERFORMANCE_SCHEMA.
For now, both buf_pool.flush_list_mutex and buf_pool.mutex will be
declared with MY_MUTEX_INIT_FAST (PTHREAD_MUTEX_ADAPTIVE_NP). The critical
sections of buf_pool.flush_list_mutex should be shorter than those for
buf_pool.mutex, because in the worst case, they cover a linear scan of
buf_pool.flush_list, while the worst case of a critical section of
buf_pool.mutex covers a linear scan of the potentially much longer
buf_pool.LRU list.
mysql_mutex_is_owner(), safe_mutex_is_owner(): New predicate, usable
with SAFE_MUTEX. Some InnoDB debug assertions need this predicate
instead of mysql_mutex_assert_owner() or mysql_mutex_assert_not_owner().
buf_pool_t::n_flush_LRU, buf_pool_t::n_flush_list:
Replaces buf_pool_t::init_flush[] and buf_pool_t::n_flush[].
The number of active flush operations.
buf_pool_t::mutex, buf_pool_t::flush_list_mutex: Use mysql_mutex_t
instead of ib_mutex_t, to have native mutexes with PERFORMANCE_SCHEMA
and SAFE_MUTEX instrumentation.
buf_pool_t::done_flush_LRU: Condition variable for !n_flush_LRU.
buf_pool_t::done_flush_list: Condition variable for !n_flush_list.
buf_pool_t::do_flush_list: Condition variable to wake up the
buf_flush_page_cleaner when a log checkpoint needs to be written
or the server is being shut down. Replaces buf_flush_event.
We will keep using timed waits (the page cleaner thread will wake
_at least_ once per second), because the calculations for
innodb_adaptive_flushing depend on fixed time intervals.
buf_dblwr: Allocate statically, and move all code to member functions.
Use a native mutex and condition variable. Remove code to deal with
single-page flushing.
buf_dblwr_check_block(): Make the check debug-only. We were spending
a significant amount of execution time in page_simple_validate_new().
flush_counters_t::unzip_LRU_evicted: Remove.
IORequest: Make more members const. FIXME: m_fil_node should be removed.
buf_flush_sync_lsn: Protect by std::atomic, not page_cleaner.mutex
(which we are removing).
page_cleaner_slot_t, page_cleaner_t: Remove many redundant members.
pc_request_flush_slot(): Replaces pc_request() and pc_flush_slot().
recv_writer_thread: Remove. Recovery works just fine without it, if we
simply invoke buf_flush_sync() at the end of each batch in
recv_sys_t::apply().
recv_recovery_from_checkpoint_finish(): Remove. We can simply call
recv_sys.debug_free() directly.
srv_started_redo: Replaces srv_start_state.
SRV_SHUTDOWN_FLUSH_PHASE: Remove. logs_empty_and_mark_files_at_shutdown()
can communicate with the normal page cleaner loop via the new function
flush_buffer_pool().
buf_flush_remove(): Assert that the calling thread is holding
buf_pool.flush_list_mutex. This removes unnecessary mutex operations
from buf_flush_remove_pages() and buf_flush_dirty_pages(),
which replace buf_LRU_flush_or_remove_pages().
buf_flush_lists(): Renamed from buf_flush_batch(), with simplified
interface. Return the number of flushed pages. Clarified comments and
renamed min_n to max_n. Identify LRU batch by lsn=0. Merge all the functions
buf_flush_start(), buf_flush_batch(), buf_flush_end() directly to this
function, which was their only caller, and remove 2 unnecessary
buf_pool.mutex release/re-acquisition that we used to perform around
the buf_flush_batch() call. At the start, if not all log has been
durably written, wait for a background task to do it, or start a new
task to do it. This allows the log write to run concurrently with our
page flushing batch. Any pages that were skipped due to too recent
FIL_PAGE_LSN or due to them being latched by a writer should be flushed
during the next batch, unless there are further modifications to those
pages. It is possible that a page that we must flush due to small
oldest_modification also carries a recent FIL_PAGE_LSN or is being
constantly modified. In the worst case, all writers would then end up
waiting in log_free_check() to allow the flushing and the checkpoint
to complete.
buf_do_flush_list_batch(): Clarify comments, and rename min_n to max_n.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_flush_space(): Auxiliary function to look up a tablespace for
page flushing.
buf_flush_page(): Defer the computation of space->full_crc32(). Never
call log_write_up_to(), but instead skip persistent pages whose latest
modification (FIL_PAGE_LSN) is newer than the redo log. Also skip
pages on which we cannot acquire a shared latch without waiting.
buf_flush_try_neighbors(): Do not bother checking buf_fix_count
because buf_flush_page() will no longer wait for the page latch.
Take the tablespace as a parameter, and only execute this function
when innodb_flush_neighbors>0. Avoid repeated calls of page_id_t::fold().
buf_flush_relocate_on_flush_list(): Declare as cold, and push down
a condition from the callers.
buf_flush_check_neighbor(): Take id.fold() as a parameter.
buf_flush_sync(): Ensure that the buf_pool.flush_list is empty,
because the flushing batch will skip pages whose modifications have
not yet been written to the log or were latched for modification.
buf_free_from_unzip_LRU_list_batch(): Remove redundant local variables.
buf_flush_LRU_list_batch(): Let the caller buf_do_LRU_batch() initialize
the counters, and report n->evicted.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_do_LRU_batch(): Return the number of pages flushed.
buf_LRU_free_page(): Only release and re-acquire buf_pool.mutex if
adaptive hash index entries are pointing to the block.
buf_LRU_get_free_block(): Do not wake up the page cleaner, because it
will no longer perform any useful work for us, and we do not want it
to compete for I/O while buf_flush_lists(innodb_lru_flush_size, 0)
writes out and evicts at most innodb_lru_flush_size pages. (The
function buf_do_LRU_batch() may complete after writing fewer pages if
more than innodb_lru_scan_depth pages end up in buf_pool.free list.)
Eliminate some mutex release-acquire cycles, and wait for the LRU
flush batch to complete before rescanning.
buf_LRU_check_size_of_non_data_objects(): Simplify the code.
buf_page_write_complete(): Remove the parameter evict, and always
evict pages that were part of an LRU flush.
buf_page_create(): Take a pre-allocated page as a parameter.
buf_pool_t::free_block(): Free a pre-allocated block.
recv_sys_t::recover_low(), recv_sys_t::apply(): Preallocate the block
while not holding recv_sys.mutex. During page allocation, we may
initiate a page flush, which in turn may initiate a log flush, which
would require acquiring log_sys.mutex, which should always be acquired
before recv_sys.mutex in order to avoid deadlocks. Therefore, we must
not be holding recv_sys.mutex while allocating a buffer pool block.
BtrBulk::logFreeCheck(): Skip a redundant condition.
row_undo_step(): Do not invoke srv_inc_activity_count() for every row
that is being rolled back. It should suffice to invoke the function in
trx_flush_log_if_needed() during trx_t::commit_in_memory() when the
rollback completes.
sync_check_enable(): Remove. We will enable innodb_sync_debug from the
very beginning.
Reviewed by: Vladislav Vaintroub
5 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-23399: Performance regression with write workloads
The buffer pool refactoring in MDEV-15053 and MDEV-22871 shifted
the performance bottleneck to the page flushing.
The configuration parameters will be changed as follows:
innodb_lru_flush_size=32 (new: how many pages to flush on LRU eviction)
innodb_lru_scan_depth=1536 (old: 1024)
innodb_max_dirty_pages_pct=90 (old: 75)
innodb_max_dirty_pages_pct_lwm=75 (old: 0)
Note: The parameter innodb_lru_scan_depth will only affect LRU
eviction of buffer pool pages when a new page is being allocated. The
page cleaner thread will no longer evict any pages. It used to
guarantee that some pages will remain free in the buffer pool. Now, we
perform that eviction 'on demand' in buf_LRU_get_free_block().
The parameter innodb_lru_scan_depth(srv_LRU_scan_depth) is used as follows:
* When the buffer pool is being shrunk in buf_pool_t::withdraw_blocks()
* As a buf_pool.free limit in buf_LRU_list_batch() for terminating
the flushing that is initiated e.g., by buf_LRU_get_free_block()
The parameter also used to serve as an initial limit for unzip_LRU
eviction (evicting uncompressed page frames while retaining
ROW_FORMAT=COMPRESSED pages), but now we will use a hard-coded limit
of 100 or unlimited for invoking buf_LRU_scan_and_free_block().
The status variables will be changed as follows:
innodb_buffer_pool_pages_flushed: This includes also the count of
innodb_buffer_pool_pages_LRU_flushed and should work reliably,
updated one by one in buf_flush_page() to give more real-time
statistics. The function buf_flush_stats(), which we are removing,
was not called in every code path. For both counters, we will use
regular variables that are incremented in a critical section of
buf_pool.mutex. Note that show_innodb_vars() directly links to the
variables, and reads of the counters will *not* be protected by
buf_pool.mutex, so you cannot get a consistent snapshot of both variables.
The following INFORMATION_SCHEMA.INNODB_METRICS counters will be
removed, because the page cleaner no longer deals with writing or
evicting least recently used pages, and because the single-page writes
have been removed:
* buffer_LRU_batch_flush_avg_time_slot
* buffer_LRU_batch_flush_avg_time_thread
* buffer_LRU_batch_flush_avg_time_est
* buffer_LRU_batch_flush_avg_pass
* buffer_LRU_single_flush_scanned
* buffer_LRU_single_flush_num_scan
* buffer_LRU_single_flush_scanned_per_call
When moving to a single buffer pool instance in MDEV-15058, we missed
some opportunity to simplify the buf_flush_page_cleaner thread. It was
unnecessarily using a mutex and some complex data structures, even
though we always have a single page cleaner thread.
Furthermore, the buf_flush_page_cleaner thread had separate 'recovery'
and 'shutdown' modes where it was waiting to be triggered by some
other thread, adding unnecessary latency and potential for hangs in
relatively rarely executed startup or shutdown code.
The page cleaner was also running two kinds of batches in an
interleaved fashion: "LRU flush" (writing out some least recently used
pages and evicting them on write completion) and the normal batches
that aim to increase the MIN(oldest_modification) in the buffer pool,
to help the log checkpoint advance.
The buf_pool.flush_list flushing was being blocked by
buf_block_t::lock for no good reason. Furthermore, if the FIL_PAGE_LSN
of a page is ahead of log_sys.get_flushed_lsn(), that is, what has
been persistently written to the redo log, we would trigger a log
flush and then resume the page flushing. This would unnecessarily
limit the performance of the page cleaner thread and trigger the
infamous messages "InnoDB: page_cleaner: 1000ms intended loop took 4450ms.
The settings might not be optimal" that were suppressed in
commit d1ab89037a518fcffbc50c24e4bd94e4ec33aed0 unless log_warnings>2.
Our revised algorithm will make log_sys.get_flushed_lsn() advance at
the start of buf_flush_lists(), and then execute a 'best effort' to
write out all pages. The flush batches will skip pages that were modified
since the log was written, or are are currently exclusively locked.
The MDEV-13670 message "page_cleaner: 1000ms intended loop took" message
will be removed, because by design, the buf_flush_page_cleaner() should
not be blocked during a batch for extended periods of time.
We will remove the single-page flushing altogether. Related to this,
the debug parameter innodb_doublewrite_batch_size will be removed,
because all of the doublewrite buffer will be used for flushing
batches. If a page needs to be evicted from the buffer pool and all
100 least recently used pages in the buffer pool have unflushed
changes, buf_LRU_get_free_block() will execute buf_flush_lists() to
write out and evict innodb_lru_flush_size pages. At most one thread
will execute buf_flush_lists() in buf_LRU_get_free_block(); other
threads will wait for that LRU flushing batch to finish.
To improve concurrency, we will replace the InnoDB ib_mutex_t and
os_event_t native mutexes and condition variables in this area of code.
Most notably, this means that the buffer pool mutex (buf_pool.mutex)
is no longer instrumented via any InnoDB interfaces. It will continue
to be instrumented via PERFORMANCE_SCHEMA.
For now, both buf_pool.flush_list_mutex and buf_pool.mutex will be
declared with MY_MUTEX_INIT_FAST (PTHREAD_MUTEX_ADAPTIVE_NP). The critical
sections of buf_pool.flush_list_mutex should be shorter than those for
buf_pool.mutex, because in the worst case, they cover a linear scan of
buf_pool.flush_list, while the worst case of a critical section of
buf_pool.mutex covers a linear scan of the potentially much longer
buf_pool.LRU list.
mysql_mutex_is_owner(), safe_mutex_is_owner(): New predicate, usable
with SAFE_MUTEX. Some InnoDB debug assertions need this predicate
instead of mysql_mutex_assert_owner() or mysql_mutex_assert_not_owner().
buf_pool_t::n_flush_LRU, buf_pool_t::n_flush_list:
Replaces buf_pool_t::init_flush[] and buf_pool_t::n_flush[].
The number of active flush operations.
buf_pool_t::mutex, buf_pool_t::flush_list_mutex: Use mysql_mutex_t
instead of ib_mutex_t, to have native mutexes with PERFORMANCE_SCHEMA
and SAFE_MUTEX instrumentation.
buf_pool_t::done_flush_LRU: Condition variable for !n_flush_LRU.
buf_pool_t::done_flush_list: Condition variable for !n_flush_list.
buf_pool_t::do_flush_list: Condition variable to wake up the
buf_flush_page_cleaner when a log checkpoint needs to be written
or the server is being shut down. Replaces buf_flush_event.
We will keep using timed waits (the page cleaner thread will wake
_at least_ once per second), because the calculations for
innodb_adaptive_flushing depend on fixed time intervals.
buf_dblwr: Allocate statically, and move all code to member functions.
Use a native mutex and condition variable. Remove code to deal with
single-page flushing.
buf_dblwr_check_block(): Make the check debug-only. We were spending
a significant amount of execution time in page_simple_validate_new().
flush_counters_t::unzip_LRU_evicted: Remove.
IORequest: Make more members const. FIXME: m_fil_node should be removed.
buf_flush_sync_lsn: Protect by std::atomic, not page_cleaner.mutex
(which we are removing).
page_cleaner_slot_t, page_cleaner_t: Remove many redundant members.
pc_request_flush_slot(): Replaces pc_request() and pc_flush_slot().
recv_writer_thread: Remove. Recovery works just fine without it, if we
simply invoke buf_flush_sync() at the end of each batch in
recv_sys_t::apply().
recv_recovery_from_checkpoint_finish(): Remove. We can simply call
recv_sys.debug_free() directly.
srv_started_redo: Replaces srv_start_state.
SRV_SHUTDOWN_FLUSH_PHASE: Remove. logs_empty_and_mark_files_at_shutdown()
can communicate with the normal page cleaner loop via the new function
flush_buffer_pool().
buf_flush_remove(): Assert that the calling thread is holding
buf_pool.flush_list_mutex. This removes unnecessary mutex operations
from buf_flush_remove_pages() and buf_flush_dirty_pages(),
which replace buf_LRU_flush_or_remove_pages().
buf_flush_lists(): Renamed from buf_flush_batch(), with simplified
interface. Return the number of flushed pages. Clarified comments and
renamed min_n to max_n. Identify LRU batch by lsn=0. Merge all the functions
buf_flush_start(), buf_flush_batch(), buf_flush_end() directly to this
function, which was their only caller, and remove 2 unnecessary
buf_pool.mutex release/re-acquisition that we used to perform around
the buf_flush_batch() call. At the start, if not all log has been
durably written, wait for a background task to do it, or start a new
task to do it. This allows the log write to run concurrently with our
page flushing batch. Any pages that were skipped due to too recent
FIL_PAGE_LSN or due to them being latched by a writer should be flushed
during the next batch, unless there are further modifications to those
pages. It is possible that a page that we must flush due to small
oldest_modification also carries a recent FIL_PAGE_LSN or is being
constantly modified. In the worst case, all writers would then end up
waiting in log_free_check() to allow the flushing and the checkpoint
to complete.
buf_do_flush_list_batch(): Clarify comments, and rename min_n to max_n.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_flush_space(): Auxiliary function to look up a tablespace for
page flushing.
buf_flush_page(): Defer the computation of space->full_crc32(). Never
call log_write_up_to(), but instead skip persistent pages whose latest
modification (FIL_PAGE_LSN) is newer than the redo log. Also skip
pages on which we cannot acquire a shared latch without waiting.
buf_flush_try_neighbors(): Do not bother checking buf_fix_count
because buf_flush_page() will no longer wait for the page latch.
Take the tablespace as a parameter, and only execute this function
when innodb_flush_neighbors>0. Avoid repeated calls of page_id_t::fold().
buf_flush_relocate_on_flush_list(): Declare as cold, and push down
a condition from the callers.
buf_flush_check_neighbor(): Take id.fold() as a parameter.
buf_flush_sync(): Ensure that the buf_pool.flush_list is empty,
because the flushing batch will skip pages whose modifications have
not yet been written to the log or were latched for modification.
buf_free_from_unzip_LRU_list_batch(): Remove redundant local variables.
buf_flush_LRU_list_batch(): Let the caller buf_do_LRU_batch() initialize
the counters, and report n->evicted.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_do_LRU_batch(): Return the number of pages flushed.
buf_LRU_free_page(): Only release and re-acquire buf_pool.mutex if
adaptive hash index entries are pointing to the block.
buf_LRU_get_free_block(): Do not wake up the page cleaner, because it
will no longer perform any useful work for us, and we do not want it
to compete for I/O while buf_flush_lists(innodb_lru_flush_size, 0)
writes out and evicts at most innodb_lru_flush_size pages. (The
function buf_do_LRU_batch() may complete after writing fewer pages if
more than innodb_lru_scan_depth pages end up in buf_pool.free list.)
Eliminate some mutex release-acquire cycles, and wait for the LRU
flush batch to complete before rescanning.
buf_LRU_check_size_of_non_data_objects(): Simplify the code.
buf_page_write_complete(): Remove the parameter evict, and always
evict pages that were part of an LRU flush.
buf_page_create(): Take a pre-allocated page as a parameter.
buf_pool_t::free_block(): Free a pre-allocated block.
recv_sys_t::recover_low(), recv_sys_t::apply(): Preallocate the block
while not holding recv_sys.mutex. During page allocation, we may
initiate a page flush, which in turn may initiate a log flush, which
would require acquiring log_sys.mutex, which should always be acquired
before recv_sys.mutex in order to avoid deadlocks. Therefore, we must
not be holding recv_sys.mutex while allocating a buffer pool block.
BtrBulk::logFreeCheck(): Skip a redundant condition.
row_undo_step(): Do not invoke srv_inc_activity_count() for every row
that is being rolled back. It should suffice to invoke the function in
trx_flush_log_if_needed() during trx_t::commit_in_memory() when the
rollback completes.
sync_check_enable(): Remove. We will enable innodb_sync_debug from the
very beginning.
Reviewed by: Vladislav Vaintroub
5 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-23399: Performance regression with write workloads
The buffer pool refactoring in MDEV-15053 and MDEV-22871 shifted
the performance bottleneck to the page flushing.
The configuration parameters will be changed as follows:
innodb_lru_flush_size=32 (new: how many pages to flush on LRU eviction)
innodb_lru_scan_depth=1536 (old: 1024)
innodb_max_dirty_pages_pct=90 (old: 75)
innodb_max_dirty_pages_pct_lwm=75 (old: 0)
Note: The parameter innodb_lru_scan_depth will only affect LRU
eviction of buffer pool pages when a new page is being allocated. The
page cleaner thread will no longer evict any pages. It used to
guarantee that some pages will remain free in the buffer pool. Now, we
perform that eviction 'on demand' in buf_LRU_get_free_block().
The parameter innodb_lru_scan_depth(srv_LRU_scan_depth) is used as follows:
* When the buffer pool is being shrunk in buf_pool_t::withdraw_blocks()
* As a buf_pool.free limit in buf_LRU_list_batch() for terminating
the flushing that is initiated e.g., by buf_LRU_get_free_block()
The parameter also used to serve as an initial limit for unzip_LRU
eviction (evicting uncompressed page frames while retaining
ROW_FORMAT=COMPRESSED pages), but now we will use a hard-coded limit
of 100 or unlimited for invoking buf_LRU_scan_and_free_block().
The status variables will be changed as follows:
innodb_buffer_pool_pages_flushed: This includes also the count of
innodb_buffer_pool_pages_LRU_flushed and should work reliably,
updated one by one in buf_flush_page() to give more real-time
statistics. The function buf_flush_stats(), which we are removing,
was not called in every code path. For both counters, we will use
regular variables that are incremented in a critical section of
buf_pool.mutex. Note that show_innodb_vars() directly links to the
variables, and reads of the counters will *not* be protected by
buf_pool.mutex, so you cannot get a consistent snapshot of both variables.
The following INFORMATION_SCHEMA.INNODB_METRICS counters will be
removed, because the page cleaner no longer deals with writing or
evicting least recently used pages, and because the single-page writes
have been removed:
* buffer_LRU_batch_flush_avg_time_slot
* buffer_LRU_batch_flush_avg_time_thread
* buffer_LRU_batch_flush_avg_time_est
* buffer_LRU_batch_flush_avg_pass
* buffer_LRU_single_flush_scanned
* buffer_LRU_single_flush_num_scan
* buffer_LRU_single_flush_scanned_per_call
When moving to a single buffer pool instance in MDEV-15058, we missed
some opportunity to simplify the buf_flush_page_cleaner thread. It was
unnecessarily using a mutex and some complex data structures, even
though we always have a single page cleaner thread.
Furthermore, the buf_flush_page_cleaner thread had separate 'recovery'
and 'shutdown' modes where it was waiting to be triggered by some
other thread, adding unnecessary latency and potential for hangs in
relatively rarely executed startup or shutdown code.
The page cleaner was also running two kinds of batches in an
interleaved fashion: "LRU flush" (writing out some least recently used
pages and evicting them on write completion) and the normal batches
that aim to increase the MIN(oldest_modification) in the buffer pool,
to help the log checkpoint advance.
The buf_pool.flush_list flushing was being blocked by
buf_block_t::lock for no good reason. Furthermore, if the FIL_PAGE_LSN
of a page is ahead of log_sys.get_flushed_lsn(), that is, what has
been persistently written to the redo log, we would trigger a log
flush and then resume the page flushing. This would unnecessarily
limit the performance of the page cleaner thread and trigger the
infamous messages "InnoDB: page_cleaner: 1000ms intended loop took 4450ms.
The settings might not be optimal" that were suppressed in
commit d1ab89037a518fcffbc50c24e4bd94e4ec33aed0 unless log_warnings>2.
Our revised algorithm will make log_sys.get_flushed_lsn() advance at
the start of buf_flush_lists(), and then execute a 'best effort' to
write out all pages. The flush batches will skip pages that were modified
since the log was written, or are are currently exclusively locked.
The MDEV-13670 message "page_cleaner: 1000ms intended loop took" message
will be removed, because by design, the buf_flush_page_cleaner() should
not be blocked during a batch for extended periods of time.
We will remove the single-page flushing altogether. Related to this,
the debug parameter innodb_doublewrite_batch_size will be removed,
because all of the doublewrite buffer will be used for flushing
batches. If a page needs to be evicted from the buffer pool and all
100 least recently used pages in the buffer pool have unflushed
changes, buf_LRU_get_free_block() will execute buf_flush_lists() to
write out and evict innodb_lru_flush_size pages. At most one thread
will execute buf_flush_lists() in buf_LRU_get_free_block(); other
threads will wait for that LRU flushing batch to finish.
To improve concurrency, we will replace the InnoDB ib_mutex_t and
os_event_t native mutexes and condition variables in this area of code.
Most notably, this means that the buffer pool mutex (buf_pool.mutex)
is no longer instrumented via any InnoDB interfaces. It will continue
to be instrumented via PERFORMANCE_SCHEMA.
For now, both buf_pool.flush_list_mutex and buf_pool.mutex will be
declared with MY_MUTEX_INIT_FAST (PTHREAD_MUTEX_ADAPTIVE_NP). The critical
sections of buf_pool.flush_list_mutex should be shorter than those for
buf_pool.mutex, because in the worst case, they cover a linear scan of
buf_pool.flush_list, while the worst case of a critical section of
buf_pool.mutex covers a linear scan of the potentially much longer
buf_pool.LRU list.
mysql_mutex_is_owner(), safe_mutex_is_owner(): New predicate, usable
with SAFE_MUTEX. Some InnoDB debug assertions need this predicate
instead of mysql_mutex_assert_owner() or mysql_mutex_assert_not_owner().
buf_pool_t::n_flush_LRU, buf_pool_t::n_flush_list:
Replaces buf_pool_t::init_flush[] and buf_pool_t::n_flush[].
The number of active flush operations.
buf_pool_t::mutex, buf_pool_t::flush_list_mutex: Use mysql_mutex_t
instead of ib_mutex_t, to have native mutexes with PERFORMANCE_SCHEMA
and SAFE_MUTEX instrumentation.
buf_pool_t::done_flush_LRU: Condition variable for !n_flush_LRU.
buf_pool_t::done_flush_list: Condition variable for !n_flush_list.
buf_pool_t::do_flush_list: Condition variable to wake up the
buf_flush_page_cleaner when a log checkpoint needs to be written
or the server is being shut down. Replaces buf_flush_event.
We will keep using timed waits (the page cleaner thread will wake
_at least_ once per second), because the calculations for
innodb_adaptive_flushing depend on fixed time intervals.
buf_dblwr: Allocate statically, and move all code to member functions.
Use a native mutex and condition variable. Remove code to deal with
single-page flushing.
buf_dblwr_check_block(): Make the check debug-only. We were spending
a significant amount of execution time in page_simple_validate_new().
flush_counters_t::unzip_LRU_evicted: Remove.
IORequest: Make more members const. FIXME: m_fil_node should be removed.
buf_flush_sync_lsn: Protect by std::atomic, not page_cleaner.mutex
(which we are removing).
page_cleaner_slot_t, page_cleaner_t: Remove many redundant members.
pc_request_flush_slot(): Replaces pc_request() and pc_flush_slot().
recv_writer_thread: Remove. Recovery works just fine without it, if we
simply invoke buf_flush_sync() at the end of each batch in
recv_sys_t::apply().
recv_recovery_from_checkpoint_finish(): Remove. We can simply call
recv_sys.debug_free() directly.
srv_started_redo: Replaces srv_start_state.
SRV_SHUTDOWN_FLUSH_PHASE: Remove. logs_empty_and_mark_files_at_shutdown()
can communicate with the normal page cleaner loop via the new function
flush_buffer_pool().
buf_flush_remove(): Assert that the calling thread is holding
buf_pool.flush_list_mutex. This removes unnecessary mutex operations
from buf_flush_remove_pages() and buf_flush_dirty_pages(),
which replace buf_LRU_flush_or_remove_pages().
buf_flush_lists(): Renamed from buf_flush_batch(), with simplified
interface. Return the number of flushed pages. Clarified comments and
renamed min_n to max_n. Identify LRU batch by lsn=0. Merge all the functions
buf_flush_start(), buf_flush_batch(), buf_flush_end() directly to this
function, which was their only caller, and remove 2 unnecessary
buf_pool.mutex release/re-acquisition that we used to perform around
the buf_flush_batch() call. At the start, if not all log has been
durably written, wait for a background task to do it, or start a new
task to do it. This allows the log write to run concurrently with our
page flushing batch. Any pages that were skipped due to too recent
FIL_PAGE_LSN or due to them being latched by a writer should be flushed
during the next batch, unless there are further modifications to those
pages. It is possible that a page that we must flush due to small
oldest_modification also carries a recent FIL_PAGE_LSN or is being
constantly modified. In the worst case, all writers would then end up
waiting in log_free_check() to allow the flushing and the checkpoint
to complete.
buf_do_flush_list_batch(): Clarify comments, and rename min_n to max_n.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_flush_space(): Auxiliary function to look up a tablespace for
page flushing.
buf_flush_page(): Defer the computation of space->full_crc32(). Never
call log_write_up_to(), but instead skip persistent pages whose latest
modification (FIL_PAGE_LSN) is newer than the redo log. Also skip
pages on which we cannot acquire a shared latch without waiting.
buf_flush_try_neighbors(): Do not bother checking buf_fix_count
because buf_flush_page() will no longer wait for the page latch.
Take the tablespace as a parameter, and only execute this function
when innodb_flush_neighbors>0. Avoid repeated calls of page_id_t::fold().
buf_flush_relocate_on_flush_list(): Declare as cold, and push down
a condition from the callers.
buf_flush_check_neighbor(): Take id.fold() as a parameter.
buf_flush_sync(): Ensure that the buf_pool.flush_list is empty,
because the flushing batch will skip pages whose modifications have
not yet been written to the log or were latched for modification.
buf_free_from_unzip_LRU_list_batch(): Remove redundant local variables.
buf_flush_LRU_list_batch(): Let the caller buf_do_LRU_batch() initialize
the counters, and report n->evicted.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_do_LRU_batch(): Return the number of pages flushed.
buf_LRU_free_page(): Only release and re-acquire buf_pool.mutex if
adaptive hash index entries are pointing to the block.
buf_LRU_get_free_block(): Do not wake up the page cleaner, because it
will no longer perform any useful work for us, and we do not want it
to compete for I/O while buf_flush_lists(innodb_lru_flush_size, 0)
writes out and evicts at most innodb_lru_flush_size pages. (The
function buf_do_LRU_batch() may complete after writing fewer pages if
more than innodb_lru_scan_depth pages end up in buf_pool.free list.)
Eliminate some mutex release-acquire cycles, and wait for the LRU
flush batch to complete before rescanning.
buf_LRU_check_size_of_non_data_objects(): Simplify the code.
buf_page_write_complete(): Remove the parameter evict, and always
evict pages that were part of an LRU flush.
buf_page_create(): Take a pre-allocated page as a parameter.
buf_pool_t::free_block(): Free a pre-allocated block.
recv_sys_t::recover_low(), recv_sys_t::apply(): Preallocate the block
while not holding recv_sys.mutex. During page allocation, we may
initiate a page flush, which in turn may initiate a log flush, which
would require acquiring log_sys.mutex, which should always be acquired
before recv_sys.mutex in order to avoid deadlocks. Therefore, we must
not be holding recv_sys.mutex while allocating a buffer pool block.
BtrBulk::logFreeCheck(): Skip a redundant condition.
row_undo_step(): Do not invoke srv_inc_activity_count() for every row
that is being rolled back. It should suffice to invoke the function in
trx_flush_log_if_needed() during trx_t::commit_in_memory() when the
rollback completes.
sync_check_enable(): Remove. We will enable innodb_sync_debug from the
very beginning.
Reviewed by: Vladislav Vaintroub
5 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-23399: Performance regression with write workloads
The buffer pool refactoring in MDEV-15053 and MDEV-22871 shifted
the performance bottleneck to the page flushing.
The configuration parameters will be changed as follows:
innodb_lru_flush_size=32 (new: how many pages to flush on LRU eviction)
innodb_lru_scan_depth=1536 (old: 1024)
innodb_max_dirty_pages_pct=90 (old: 75)
innodb_max_dirty_pages_pct_lwm=75 (old: 0)
Note: The parameter innodb_lru_scan_depth will only affect LRU
eviction of buffer pool pages when a new page is being allocated. The
page cleaner thread will no longer evict any pages. It used to
guarantee that some pages will remain free in the buffer pool. Now, we
perform that eviction 'on demand' in buf_LRU_get_free_block().
The parameter innodb_lru_scan_depth(srv_LRU_scan_depth) is used as follows:
* When the buffer pool is being shrunk in buf_pool_t::withdraw_blocks()
* As a buf_pool.free limit in buf_LRU_list_batch() for terminating
the flushing that is initiated e.g., by buf_LRU_get_free_block()
The parameter also used to serve as an initial limit for unzip_LRU
eviction (evicting uncompressed page frames while retaining
ROW_FORMAT=COMPRESSED pages), but now we will use a hard-coded limit
of 100 or unlimited for invoking buf_LRU_scan_and_free_block().
The status variables will be changed as follows:
innodb_buffer_pool_pages_flushed: This includes also the count of
innodb_buffer_pool_pages_LRU_flushed and should work reliably,
updated one by one in buf_flush_page() to give more real-time
statistics. The function buf_flush_stats(), which we are removing,
was not called in every code path. For both counters, we will use
regular variables that are incremented in a critical section of
buf_pool.mutex. Note that show_innodb_vars() directly links to the
variables, and reads of the counters will *not* be protected by
buf_pool.mutex, so you cannot get a consistent snapshot of both variables.
The following INFORMATION_SCHEMA.INNODB_METRICS counters will be
removed, because the page cleaner no longer deals with writing or
evicting least recently used pages, and because the single-page writes
have been removed:
* buffer_LRU_batch_flush_avg_time_slot
* buffer_LRU_batch_flush_avg_time_thread
* buffer_LRU_batch_flush_avg_time_est
* buffer_LRU_batch_flush_avg_pass
* buffer_LRU_single_flush_scanned
* buffer_LRU_single_flush_num_scan
* buffer_LRU_single_flush_scanned_per_call
When moving to a single buffer pool instance in MDEV-15058, we missed
some opportunity to simplify the buf_flush_page_cleaner thread. It was
unnecessarily using a mutex and some complex data structures, even
though we always have a single page cleaner thread.
Furthermore, the buf_flush_page_cleaner thread had separate 'recovery'
and 'shutdown' modes where it was waiting to be triggered by some
other thread, adding unnecessary latency and potential for hangs in
relatively rarely executed startup or shutdown code.
The page cleaner was also running two kinds of batches in an
interleaved fashion: "LRU flush" (writing out some least recently used
pages and evicting them on write completion) and the normal batches
that aim to increase the MIN(oldest_modification) in the buffer pool,
to help the log checkpoint advance.
The buf_pool.flush_list flushing was being blocked by
buf_block_t::lock for no good reason. Furthermore, if the FIL_PAGE_LSN
of a page is ahead of log_sys.get_flushed_lsn(), that is, what has
been persistently written to the redo log, we would trigger a log
flush and then resume the page flushing. This would unnecessarily
limit the performance of the page cleaner thread and trigger the
infamous messages "InnoDB: page_cleaner: 1000ms intended loop took 4450ms.
The settings might not be optimal" that were suppressed in
commit d1ab89037a518fcffbc50c24e4bd94e4ec33aed0 unless log_warnings>2.
Our revised algorithm will make log_sys.get_flushed_lsn() advance at
the start of buf_flush_lists(), and then execute a 'best effort' to
write out all pages. The flush batches will skip pages that were modified
since the log was written, or are are currently exclusively locked.
The MDEV-13670 message "page_cleaner: 1000ms intended loop took" message
will be removed, because by design, the buf_flush_page_cleaner() should
not be blocked during a batch for extended periods of time.
We will remove the single-page flushing altogether. Related to this,
the debug parameter innodb_doublewrite_batch_size will be removed,
because all of the doublewrite buffer will be used for flushing
batches. If a page needs to be evicted from the buffer pool and all
100 least recently used pages in the buffer pool have unflushed
changes, buf_LRU_get_free_block() will execute buf_flush_lists() to
write out and evict innodb_lru_flush_size pages. At most one thread
will execute buf_flush_lists() in buf_LRU_get_free_block(); other
threads will wait for that LRU flushing batch to finish.
To improve concurrency, we will replace the InnoDB ib_mutex_t and
os_event_t native mutexes and condition variables in this area of code.
Most notably, this means that the buffer pool mutex (buf_pool.mutex)
is no longer instrumented via any InnoDB interfaces. It will continue
to be instrumented via PERFORMANCE_SCHEMA.
For now, both buf_pool.flush_list_mutex and buf_pool.mutex will be
declared with MY_MUTEX_INIT_FAST (PTHREAD_MUTEX_ADAPTIVE_NP). The critical
sections of buf_pool.flush_list_mutex should be shorter than those for
buf_pool.mutex, because in the worst case, they cover a linear scan of
buf_pool.flush_list, while the worst case of a critical section of
buf_pool.mutex covers a linear scan of the potentially much longer
buf_pool.LRU list.
mysql_mutex_is_owner(), safe_mutex_is_owner(): New predicate, usable
with SAFE_MUTEX. Some InnoDB debug assertions need this predicate
instead of mysql_mutex_assert_owner() or mysql_mutex_assert_not_owner().
buf_pool_t::n_flush_LRU, buf_pool_t::n_flush_list:
Replaces buf_pool_t::init_flush[] and buf_pool_t::n_flush[].
The number of active flush operations.
buf_pool_t::mutex, buf_pool_t::flush_list_mutex: Use mysql_mutex_t
instead of ib_mutex_t, to have native mutexes with PERFORMANCE_SCHEMA
and SAFE_MUTEX instrumentation.
buf_pool_t::done_flush_LRU: Condition variable for !n_flush_LRU.
buf_pool_t::done_flush_list: Condition variable for !n_flush_list.
buf_pool_t::do_flush_list: Condition variable to wake up the
buf_flush_page_cleaner when a log checkpoint needs to be written
or the server is being shut down. Replaces buf_flush_event.
We will keep using timed waits (the page cleaner thread will wake
_at least_ once per second), because the calculations for
innodb_adaptive_flushing depend on fixed time intervals.
buf_dblwr: Allocate statically, and move all code to member functions.
Use a native mutex and condition variable. Remove code to deal with
single-page flushing.
buf_dblwr_check_block(): Make the check debug-only. We were spending
a significant amount of execution time in page_simple_validate_new().
flush_counters_t::unzip_LRU_evicted: Remove.
IORequest: Make more members const. FIXME: m_fil_node should be removed.
buf_flush_sync_lsn: Protect by std::atomic, not page_cleaner.mutex
(which we are removing).
page_cleaner_slot_t, page_cleaner_t: Remove many redundant members.
pc_request_flush_slot(): Replaces pc_request() and pc_flush_slot().
recv_writer_thread: Remove. Recovery works just fine without it, if we
simply invoke buf_flush_sync() at the end of each batch in
recv_sys_t::apply().
recv_recovery_from_checkpoint_finish(): Remove. We can simply call
recv_sys.debug_free() directly.
srv_started_redo: Replaces srv_start_state.
SRV_SHUTDOWN_FLUSH_PHASE: Remove. logs_empty_and_mark_files_at_shutdown()
can communicate with the normal page cleaner loop via the new function
flush_buffer_pool().
buf_flush_remove(): Assert that the calling thread is holding
buf_pool.flush_list_mutex. This removes unnecessary mutex operations
from buf_flush_remove_pages() and buf_flush_dirty_pages(),
which replace buf_LRU_flush_or_remove_pages().
buf_flush_lists(): Renamed from buf_flush_batch(), with simplified
interface. Return the number of flushed pages. Clarified comments and
renamed min_n to max_n. Identify LRU batch by lsn=0. Merge all the functions
buf_flush_start(), buf_flush_batch(), buf_flush_end() directly to this
function, which was their only caller, and remove 2 unnecessary
buf_pool.mutex release/re-acquisition that we used to perform around
the buf_flush_batch() call. At the start, if not all log has been
durably written, wait for a background task to do it, or start a new
task to do it. This allows the log write to run concurrently with our
page flushing batch. Any pages that were skipped due to too recent
FIL_PAGE_LSN or due to them being latched by a writer should be flushed
during the next batch, unless there are further modifications to those
pages. It is possible that a page that we must flush due to small
oldest_modification also carries a recent FIL_PAGE_LSN or is being
constantly modified. In the worst case, all writers would then end up
waiting in log_free_check() to allow the flushing and the checkpoint
to complete.
buf_do_flush_list_batch(): Clarify comments, and rename min_n to max_n.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_flush_space(): Auxiliary function to look up a tablespace for
page flushing.
buf_flush_page(): Defer the computation of space->full_crc32(). Never
call log_write_up_to(), but instead skip persistent pages whose latest
modification (FIL_PAGE_LSN) is newer than the redo log. Also skip
pages on which we cannot acquire a shared latch without waiting.
buf_flush_try_neighbors(): Do not bother checking buf_fix_count
because buf_flush_page() will no longer wait for the page latch.
Take the tablespace as a parameter, and only execute this function
when innodb_flush_neighbors>0. Avoid repeated calls of page_id_t::fold().
buf_flush_relocate_on_flush_list(): Declare as cold, and push down
a condition from the callers.
buf_flush_check_neighbor(): Take id.fold() as a parameter.
buf_flush_sync(): Ensure that the buf_pool.flush_list is empty,
because the flushing batch will skip pages whose modifications have
not yet been written to the log or were latched for modification.
buf_free_from_unzip_LRU_list_batch(): Remove redundant local variables.
buf_flush_LRU_list_batch(): Let the caller buf_do_LRU_batch() initialize
the counters, and report n->evicted.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_do_LRU_batch(): Return the number of pages flushed.
buf_LRU_free_page(): Only release and re-acquire buf_pool.mutex if
adaptive hash index entries are pointing to the block.
buf_LRU_get_free_block(): Do not wake up the page cleaner, because it
will no longer perform any useful work for us, and we do not want it
to compete for I/O while buf_flush_lists(innodb_lru_flush_size, 0)
writes out and evicts at most innodb_lru_flush_size pages. (The
function buf_do_LRU_batch() may complete after writing fewer pages if
more than innodb_lru_scan_depth pages end up in buf_pool.free list.)
Eliminate some mutex release-acquire cycles, and wait for the LRU
flush batch to complete before rescanning.
buf_LRU_check_size_of_non_data_objects(): Simplify the code.
buf_page_write_complete(): Remove the parameter evict, and always
evict pages that were part of an LRU flush.
buf_page_create(): Take a pre-allocated page as a parameter.
buf_pool_t::free_block(): Free a pre-allocated block.
recv_sys_t::recover_low(), recv_sys_t::apply(): Preallocate the block
while not holding recv_sys.mutex. During page allocation, we may
initiate a page flush, which in turn may initiate a log flush, which
would require acquiring log_sys.mutex, which should always be acquired
before recv_sys.mutex in order to avoid deadlocks. Therefore, we must
not be holding recv_sys.mutex while allocating a buffer pool block.
BtrBulk::logFreeCheck(): Skip a redundant condition.
row_undo_step(): Do not invoke srv_inc_activity_count() for every row
that is being rolled back. It should suffice to invoke the function in
trx_flush_log_if_needed() during trx_t::commit_in_memory() when the
rollback completes.
sync_check_enable(): Remove. We will enable innodb_sync_debug from the
very beginning.
Reviewed by: Vladislav Vaintroub
5 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-26827 Make page flushing even faster
For more convenient monitoring of something that could greatly affect
the volume of page writes, we add the status variable
Innodb_buffer_pool_pages_split that was previously only available
via information_schema.innodb_metrics as "innodb_page_splits".
This was suggested by Axel Schwenke.
buf_flush_page_count: Replaced with buf_pool.stat.n_pages_written.
We protect buf_pool.stat (except n_page_gets) with buf_pool.mutex
and remove unnecessary export_vars indirection.
buf_pool.flush_list_bytes: Moved from buf_pool.stat.flush_list_bytes.
Protected by buf_pool.flush_list_mutex.
buf_pool_t::page_cleaner_status: Replaces buf_pool_t::n_flush_LRU_,
buf_pool_t::n_flush_list_, and buf_pool_t::page_cleaner_is_idle.
Protected by buf_pool.flush_list_mutex. We will exclusively broadcast
buf_pool.done_flush_list by the buf_flush_page_cleaner thread,
and only wait for it when communicating with buf_flush_page_cleaner.
There is no need to keep a count of pending writes by the
buf_pool.flush_list processing. A single flag suffices for that.
Waits for page write completion can be performed by
simply waiting on block->page.lock, or by invoking
buf_dblwr.wait_for_page_writes().
buf_LRU_block_free_non_file_page(): Broadcast buf_pool.done_free and
set buf_pool.try_LRU_scan when freeing a page. This would be
executed also as part of buf_page_write_complete().
buf_page_write_complete(): Do not broadcast buf_pool.done_flush_list,
and do not acquire buf_pool.mutex unless buf_pool.LRU eviction is needed.
Let buf_dblwr count all writes to persistent pages and broadcast a
condition variable when no outstanding writes remain.
buf_flush_page_cleaner(): Prioritize LRU flushing and eviction right after
"furious flushing" (lsn_limit). Simplify the conditions and reduce the
hold time of buf_pool.flush_list_mutex. Refuse to shut down
or sleep if buf_pool.ran_out(), that is, LRU eviction is needed.
buf_pool_t::page_cleaner_wakeup(): Add the optional parameter for_LRU.
buf_LRU_get_free_block(): Protect buf_lru_free_blocks_error_printed
with buf_pool.mutex. Invoke buf_pool.page_cleaner_wakeup(true) to
to ensure that buf_flush_page_cleaner() will process the LRU flush
request.
buf_do_LRU_batch(), buf_flush_list(), buf_flush_list_space():
Update buf_pool.stat.n_pages_written when submitting writes
(while holding buf_pool.mutex), not when completing them.
buf_page_t::flush(), buf_flush_discard_page(): Require that
the page U-latch be acquired upfront, and remove
buf_page_t::ready_for_flush().
buf_pool_t::delete_from_flush_list(): Remove the parameter "bool clear".
buf_flush_page(): Count pending page writes via buf_dblwr.
buf_flush_try_neighbors(): Take the block of page_id as a parameter.
If the tablespace is dropped before our page has been written out,
release the page U-latch.
buf_pool_invalidate(): Let the caller ensure that there are no
outstanding writes.
buf_flush_wait_batch_end(false),
buf_flush_wait_batch_end_acquiring_mutex(false):
Replaced with buf_dblwr.wait_for_page_writes().
buf_flush_wait_LRU_batch_end(): Replaces buf_flush_wait_batch_end(true).
buf_flush_list(): Remove some broadcast of buf_pool.done_flush_list.
buf_flush_buffer_pool(): Invoke also buf_dblwr.wait_for_page_writes().
buf_pool_t::io_pending(), buf_pool_t::n_flush_list(): Remove.
Outstanding writes are reflected by buf_dblwr.pending_writes().
buf_dblwr_t::init(): New function, to initialize the mutex and
the condition variables, but not the backing store.
buf_dblwr_t::is_created(): Replaces buf_dblwr_t::is_initialised().
buf_dblwr_t::pending_writes(), buf_dblwr_t::writes_pending:
Keeps track of writes of persistent data pages.
buf_flush_LRU(): Allow calls while LRU flushing may be in progress
in another thread.
Tested by Matthias Leich (correctness) and Axel Schwenke (performance)
3 years ago MDEV-23399: Performance regression with write workloads
The buffer pool refactoring in MDEV-15053 and MDEV-22871 shifted
the performance bottleneck to the page flushing.
The configuration parameters will be changed as follows:
innodb_lru_flush_size=32 (new: how many pages to flush on LRU eviction)
innodb_lru_scan_depth=1536 (old: 1024)
innodb_max_dirty_pages_pct=90 (old: 75)
innodb_max_dirty_pages_pct_lwm=75 (old: 0)
Note: The parameter innodb_lru_scan_depth will only affect LRU
eviction of buffer pool pages when a new page is being allocated. The
page cleaner thread will no longer evict any pages. It used to
guarantee that some pages will remain free in the buffer pool. Now, we
perform that eviction 'on demand' in buf_LRU_get_free_block().
The parameter innodb_lru_scan_depth(srv_LRU_scan_depth) is used as follows:
* When the buffer pool is being shrunk in buf_pool_t::withdraw_blocks()
* As a buf_pool.free limit in buf_LRU_list_batch() for terminating
the flushing that is initiated e.g., by buf_LRU_get_free_block()
The parameter also used to serve as an initial limit for unzip_LRU
eviction (evicting uncompressed page frames while retaining
ROW_FORMAT=COMPRESSED pages), but now we will use a hard-coded limit
of 100 or unlimited for invoking buf_LRU_scan_and_free_block().
The status variables will be changed as follows:
innodb_buffer_pool_pages_flushed: This includes also the count of
innodb_buffer_pool_pages_LRU_flushed and should work reliably,
updated one by one in buf_flush_page() to give more real-time
statistics. The function buf_flush_stats(), which we are removing,
was not called in every code path. For both counters, we will use
regular variables that are incremented in a critical section of
buf_pool.mutex. Note that show_innodb_vars() directly links to the
variables, and reads of the counters will *not* be protected by
buf_pool.mutex, so you cannot get a consistent snapshot of both variables.
The following INFORMATION_SCHEMA.INNODB_METRICS counters will be
removed, because the page cleaner no longer deals with writing or
evicting least recently used pages, and because the single-page writes
have been removed:
* buffer_LRU_batch_flush_avg_time_slot
* buffer_LRU_batch_flush_avg_time_thread
* buffer_LRU_batch_flush_avg_time_est
* buffer_LRU_batch_flush_avg_pass
* buffer_LRU_single_flush_scanned
* buffer_LRU_single_flush_num_scan
* buffer_LRU_single_flush_scanned_per_call
When moving to a single buffer pool instance in MDEV-15058, we missed
some opportunity to simplify the buf_flush_page_cleaner thread. It was
unnecessarily using a mutex and some complex data structures, even
though we always have a single page cleaner thread.
Furthermore, the buf_flush_page_cleaner thread had separate 'recovery'
and 'shutdown' modes where it was waiting to be triggered by some
other thread, adding unnecessary latency and potential for hangs in
relatively rarely executed startup or shutdown code.
The page cleaner was also running two kinds of batches in an
interleaved fashion: "LRU flush" (writing out some least recently used
pages and evicting them on write completion) and the normal batches
that aim to increase the MIN(oldest_modification) in the buffer pool,
to help the log checkpoint advance.
The buf_pool.flush_list flushing was being blocked by
buf_block_t::lock for no good reason. Furthermore, if the FIL_PAGE_LSN
of a page is ahead of log_sys.get_flushed_lsn(), that is, what has
been persistently written to the redo log, we would trigger a log
flush and then resume the page flushing. This would unnecessarily
limit the performance of the page cleaner thread and trigger the
infamous messages "InnoDB: page_cleaner: 1000ms intended loop took 4450ms.
The settings might not be optimal" that were suppressed in
commit d1ab89037a518fcffbc50c24e4bd94e4ec33aed0 unless log_warnings>2.
Our revised algorithm will make log_sys.get_flushed_lsn() advance at
the start of buf_flush_lists(), and then execute a 'best effort' to
write out all pages. The flush batches will skip pages that were modified
since the log was written, or are are currently exclusively locked.
The MDEV-13670 message "page_cleaner: 1000ms intended loop took" message
will be removed, because by design, the buf_flush_page_cleaner() should
not be blocked during a batch for extended periods of time.
We will remove the single-page flushing altogether. Related to this,
the debug parameter innodb_doublewrite_batch_size will be removed,
because all of the doublewrite buffer will be used for flushing
batches. If a page needs to be evicted from the buffer pool and all
100 least recently used pages in the buffer pool have unflushed
changes, buf_LRU_get_free_block() will execute buf_flush_lists() to
write out and evict innodb_lru_flush_size pages. At most one thread
will execute buf_flush_lists() in buf_LRU_get_free_block(); other
threads will wait for that LRU flushing batch to finish.
To improve concurrency, we will replace the InnoDB ib_mutex_t and
os_event_t native mutexes and condition variables in this area of code.
Most notably, this means that the buffer pool mutex (buf_pool.mutex)
is no longer instrumented via any InnoDB interfaces. It will continue
to be instrumented via PERFORMANCE_SCHEMA.
For now, both buf_pool.flush_list_mutex and buf_pool.mutex will be
declared with MY_MUTEX_INIT_FAST (PTHREAD_MUTEX_ADAPTIVE_NP). The critical
sections of buf_pool.flush_list_mutex should be shorter than those for
buf_pool.mutex, because in the worst case, they cover a linear scan of
buf_pool.flush_list, while the worst case of a critical section of
buf_pool.mutex covers a linear scan of the potentially much longer
buf_pool.LRU list.
mysql_mutex_is_owner(), safe_mutex_is_owner(): New predicate, usable
with SAFE_MUTEX. Some InnoDB debug assertions need this predicate
instead of mysql_mutex_assert_owner() or mysql_mutex_assert_not_owner().
buf_pool_t::n_flush_LRU, buf_pool_t::n_flush_list:
Replaces buf_pool_t::init_flush[] and buf_pool_t::n_flush[].
The number of active flush operations.
buf_pool_t::mutex, buf_pool_t::flush_list_mutex: Use mysql_mutex_t
instead of ib_mutex_t, to have native mutexes with PERFORMANCE_SCHEMA
and SAFE_MUTEX instrumentation.
buf_pool_t::done_flush_LRU: Condition variable for !n_flush_LRU.
buf_pool_t::done_flush_list: Condition variable for !n_flush_list.
buf_pool_t::do_flush_list: Condition variable to wake up the
buf_flush_page_cleaner when a log checkpoint needs to be written
or the server is being shut down. Replaces buf_flush_event.
We will keep using timed waits (the page cleaner thread will wake
_at least_ once per second), because the calculations for
innodb_adaptive_flushing depend on fixed time intervals.
buf_dblwr: Allocate statically, and move all code to member functions.
Use a native mutex and condition variable. Remove code to deal with
single-page flushing.
buf_dblwr_check_block(): Make the check debug-only. We were spending
a significant amount of execution time in page_simple_validate_new().
flush_counters_t::unzip_LRU_evicted: Remove.
IORequest: Make more members const. FIXME: m_fil_node should be removed.
buf_flush_sync_lsn: Protect by std::atomic, not page_cleaner.mutex
(which we are removing).
page_cleaner_slot_t, page_cleaner_t: Remove many redundant members.
pc_request_flush_slot(): Replaces pc_request() and pc_flush_slot().
recv_writer_thread: Remove. Recovery works just fine without it, if we
simply invoke buf_flush_sync() at the end of each batch in
recv_sys_t::apply().
recv_recovery_from_checkpoint_finish(): Remove. We can simply call
recv_sys.debug_free() directly.
srv_started_redo: Replaces srv_start_state.
SRV_SHUTDOWN_FLUSH_PHASE: Remove. logs_empty_and_mark_files_at_shutdown()
can communicate with the normal page cleaner loop via the new function
flush_buffer_pool().
buf_flush_remove(): Assert that the calling thread is holding
buf_pool.flush_list_mutex. This removes unnecessary mutex operations
from buf_flush_remove_pages() and buf_flush_dirty_pages(),
which replace buf_LRU_flush_or_remove_pages().
buf_flush_lists(): Renamed from buf_flush_batch(), with simplified
interface. Return the number of flushed pages. Clarified comments and
renamed min_n to max_n. Identify LRU batch by lsn=0. Merge all the functions
buf_flush_start(), buf_flush_batch(), buf_flush_end() directly to this
function, which was their only caller, and remove 2 unnecessary
buf_pool.mutex release/re-acquisition that we used to perform around
the buf_flush_batch() call. At the start, if not all log has been
durably written, wait for a background task to do it, or start a new
task to do it. This allows the log write to run concurrently with our
page flushing batch. Any pages that were skipped due to too recent
FIL_PAGE_LSN or due to them being latched by a writer should be flushed
during the next batch, unless there are further modifications to those
pages. It is possible that a page that we must flush due to small
oldest_modification also carries a recent FIL_PAGE_LSN or is being
constantly modified. In the worst case, all writers would then end up
waiting in log_free_check() to allow the flushing and the checkpoint
to complete.
buf_do_flush_list_batch(): Clarify comments, and rename min_n to max_n.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_flush_space(): Auxiliary function to look up a tablespace for
page flushing.
buf_flush_page(): Defer the computation of space->full_crc32(). Never
call log_write_up_to(), but instead skip persistent pages whose latest
modification (FIL_PAGE_LSN) is newer than the redo log. Also skip
pages on which we cannot acquire a shared latch without waiting.
buf_flush_try_neighbors(): Do not bother checking buf_fix_count
because buf_flush_page() will no longer wait for the page latch.
Take the tablespace as a parameter, and only execute this function
when innodb_flush_neighbors>0. Avoid repeated calls of page_id_t::fold().
buf_flush_relocate_on_flush_list(): Declare as cold, and push down
a condition from the callers.
buf_flush_check_neighbor(): Take id.fold() as a parameter.
buf_flush_sync(): Ensure that the buf_pool.flush_list is empty,
because the flushing batch will skip pages whose modifications have
not yet been written to the log or were latched for modification.
buf_free_from_unzip_LRU_list_batch(): Remove redundant local variables.
buf_flush_LRU_list_batch(): Let the caller buf_do_LRU_batch() initialize
the counters, and report n->evicted.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_do_LRU_batch(): Return the number of pages flushed.
buf_LRU_free_page(): Only release and re-acquire buf_pool.mutex if
adaptive hash index entries are pointing to the block.
buf_LRU_get_free_block(): Do not wake up the page cleaner, because it
will no longer perform any useful work for us, and we do not want it
to compete for I/O while buf_flush_lists(innodb_lru_flush_size, 0)
writes out and evicts at most innodb_lru_flush_size pages. (The
function buf_do_LRU_batch() may complete after writing fewer pages if
more than innodb_lru_scan_depth pages end up in buf_pool.free list.)
Eliminate some mutex release-acquire cycles, and wait for the LRU
flush batch to complete before rescanning.
buf_LRU_check_size_of_non_data_objects(): Simplify the code.
buf_page_write_complete(): Remove the parameter evict, and always
evict pages that were part of an LRU flush.
buf_page_create(): Take a pre-allocated page as a parameter.
buf_pool_t::free_block(): Free a pre-allocated block.
recv_sys_t::recover_low(), recv_sys_t::apply(): Preallocate the block
while not holding recv_sys.mutex. During page allocation, we may
initiate a page flush, which in turn may initiate a log flush, which
would require acquiring log_sys.mutex, which should always be acquired
before recv_sys.mutex in order to avoid deadlocks. Therefore, we must
not be holding recv_sys.mutex while allocating a buffer pool block.
BtrBulk::logFreeCheck(): Skip a redundant condition.
row_undo_step(): Do not invoke srv_inc_activity_count() for every row
that is being rolled back. It should suffice to invoke the function in
trx_flush_log_if_needed() during trx_t::commit_in_memory() when the
rollback completes.
sync_check_enable(): Remove. We will enable innodb_sync_debug from the
very beginning.
Reviewed by: Vladislav Vaintroub
5 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-23399: Performance regression with write workloads
The buffer pool refactoring in MDEV-15053 and MDEV-22871 shifted
the performance bottleneck to the page flushing.
The configuration parameters will be changed as follows:
innodb_lru_flush_size=32 (new: how many pages to flush on LRU eviction)
innodb_lru_scan_depth=1536 (old: 1024)
innodb_max_dirty_pages_pct=90 (old: 75)
innodb_max_dirty_pages_pct_lwm=75 (old: 0)
Note: The parameter innodb_lru_scan_depth will only affect LRU
eviction of buffer pool pages when a new page is being allocated. The
page cleaner thread will no longer evict any pages. It used to
guarantee that some pages will remain free in the buffer pool. Now, we
perform that eviction 'on demand' in buf_LRU_get_free_block().
The parameter innodb_lru_scan_depth(srv_LRU_scan_depth) is used as follows:
* When the buffer pool is being shrunk in buf_pool_t::withdraw_blocks()
* As a buf_pool.free limit in buf_LRU_list_batch() for terminating
the flushing that is initiated e.g., by buf_LRU_get_free_block()
The parameter also used to serve as an initial limit for unzip_LRU
eviction (evicting uncompressed page frames while retaining
ROW_FORMAT=COMPRESSED pages), but now we will use a hard-coded limit
of 100 or unlimited for invoking buf_LRU_scan_and_free_block().
The status variables will be changed as follows:
innodb_buffer_pool_pages_flushed: This includes also the count of
innodb_buffer_pool_pages_LRU_flushed and should work reliably,
updated one by one in buf_flush_page() to give more real-time
statistics. The function buf_flush_stats(), which we are removing,
was not called in every code path. For both counters, we will use
regular variables that are incremented in a critical section of
buf_pool.mutex. Note that show_innodb_vars() directly links to the
variables, and reads of the counters will *not* be protected by
buf_pool.mutex, so you cannot get a consistent snapshot of both variables.
The following INFORMATION_SCHEMA.INNODB_METRICS counters will be
removed, because the page cleaner no longer deals with writing or
evicting least recently used pages, and because the single-page writes
have been removed:
* buffer_LRU_batch_flush_avg_time_slot
* buffer_LRU_batch_flush_avg_time_thread
* buffer_LRU_batch_flush_avg_time_est
* buffer_LRU_batch_flush_avg_pass
* buffer_LRU_single_flush_scanned
* buffer_LRU_single_flush_num_scan
* buffer_LRU_single_flush_scanned_per_call
When moving to a single buffer pool instance in MDEV-15058, we missed
some opportunity to simplify the buf_flush_page_cleaner thread. It was
unnecessarily using a mutex and some complex data structures, even
though we always have a single page cleaner thread.
Furthermore, the buf_flush_page_cleaner thread had separate 'recovery'
and 'shutdown' modes where it was waiting to be triggered by some
other thread, adding unnecessary latency and potential for hangs in
relatively rarely executed startup or shutdown code.
The page cleaner was also running two kinds of batches in an
interleaved fashion: "LRU flush" (writing out some least recently used
pages and evicting them on write completion) and the normal batches
that aim to increase the MIN(oldest_modification) in the buffer pool,
to help the log checkpoint advance.
The buf_pool.flush_list flushing was being blocked by
buf_block_t::lock for no good reason. Furthermore, if the FIL_PAGE_LSN
of a page is ahead of log_sys.get_flushed_lsn(), that is, what has
been persistently written to the redo log, we would trigger a log
flush and then resume the page flushing. This would unnecessarily
limit the performance of the page cleaner thread and trigger the
infamous messages "InnoDB: page_cleaner: 1000ms intended loop took 4450ms.
The settings might not be optimal" that were suppressed in
commit d1ab89037a518fcffbc50c24e4bd94e4ec33aed0 unless log_warnings>2.
Our revised algorithm will make log_sys.get_flushed_lsn() advance at
the start of buf_flush_lists(), and then execute a 'best effort' to
write out all pages. The flush batches will skip pages that were modified
since the log was written, or are are currently exclusively locked.
The MDEV-13670 message "page_cleaner: 1000ms intended loop took" message
will be removed, because by design, the buf_flush_page_cleaner() should
not be blocked during a batch for extended periods of time.
We will remove the single-page flushing altogether. Related to this,
the debug parameter innodb_doublewrite_batch_size will be removed,
because all of the doublewrite buffer will be used for flushing
batches. If a page needs to be evicted from the buffer pool and all
100 least recently used pages in the buffer pool have unflushed
changes, buf_LRU_get_free_block() will execute buf_flush_lists() to
write out and evict innodb_lru_flush_size pages. At most one thread
will execute buf_flush_lists() in buf_LRU_get_free_block(); other
threads will wait for that LRU flushing batch to finish.
To improve concurrency, we will replace the InnoDB ib_mutex_t and
os_event_t native mutexes and condition variables in this area of code.
Most notably, this means that the buffer pool mutex (buf_pool.mutex)
is no longer instrumented via any InnoDB interfaces. It will continue
to be instrumented via PERFORMANCE_SCHEMA.
For now, both buf_pool.flush_list_mutex and buf_pool.mutex will be
declared with MY_MUTEX_INIT_FAST (PTHREAD_MUTEX_ADAPTIVE_NP). The critical
sections of buf_pool.flush_list_mutex should be shorter than those for
buf_pool.mutex, because in the worst case, they cover a linear scan of
buf_pool.flush_list, while the worst case of a critical section of
buf_pool.mutex covers a linear scan of the potentially much longer
buf_pool.LRU list.
mysql_mutex_is_owner(), safe_mutex_is_owner(): New predicate, usable
with SAFE_MUTEX. Some InnoDB debug assertions need this predicate
instead of mysql_mutex_assert_owner() or mysql_mutex_assert_not_owner().
buf_pool_t::n_flush_LRU, buf_pool_t::n_flush_list:
Replaces buf_pool_t::init_flush[] and buf_pool_t::n_flush[].
The number of active flush operations.
buf_pool_t::mutex, buf_pool_t::flush_list_mutex: Use mysql_mutex_t
instead of ib_mutex_t, to have native mutexes with PERFORMANCE_SCHEMA
and SAFE_MUTEX instrumentation.
buf_pool_t::done_flush_LRU: Condition variable for !n_flush_LRU.
buf_pool_t::done_flush_list: Condition variable for !n_flush_list.
buf_pool_t::do_flush_list: Condition variable to wake up the
buf_flush_page_cleaner when a log checkpoint needs to be written
or the server is being shut down. Replaces buf_flush_event.
We will keep using timed waits (the page cleaner thread will wake
_at least_ once per second), because the calculations for
innodb_adaptive_flushing depend on fixed time intervals.
buf_dblwr: Allocate statically, and move all code to member functions.
Use a native mutex and condition variable. Remove code to deal with
single-page flushing.
buf_dblwr_check_block(): Make the check debug-only. We were spending
a significant amount of execution time in page_simple_validate_new().
flush_counters_t::unzip_LRU_evicted: Remove.
IORequest: Make more members const. FIXME: m_fil_node should be removed.
buf_flush_sync_lsn: Protect by std::atomic, not page_cleaner.mutex
(which we are removing).
page_cleaner_slot_t, page_cleaner_t: Remove many redundant members.
pc_request_flush_slot(): Replaces pc_request() and pc_flush_slot().
recv_writer_thread: Remove. Recovery works just fine without it, if we
simply invoke buf_flush_sync() at the end of each batch in
recv_sys_t::apply().
recv_recovery_from_checkpoint_finish(): Remove. We can simply call
recv_sys.debug_free() directly.
srv_started_redo: Replaces srv_start_state.
SRV_SHUTDOWN_FLUSH_PHASE: Remove. logs_empty_and_mark_files_at_shutdown()
can communicate with the normal page cleaner loop via the new function
flush_buffer_pool().
buf_flush_remove(): Assert that the calling thread is holding
buf_pool.flush_list_mutex. This removes unnecessary mutex operations
from buf_flush_remove_pages() and buf_flush_dirty_pages(),
which replace buf_LRU_flush_or_remove_pages().
buf_flush_lists(): Renamed from buf_flush_batch(), with simplified
interface. Return the number of flushed pages. Clarified comments and
renamed min_n to max_n. Identify LRU batch by lsn=0. Merge all the functions
buf_flush_start(), buf_flush_batch(), buf_flush_end() directly to this
function, which was their only caller, and remove 2 unnecessary
buf_pool.mutex release/re-acquisition that we used to perform around
the buf_flush_batch() call. At the start, if not all log has been
durably written, wait for a background task to do it, or start a new
task to do it. This allows the log write to run concurrently with our
page flushing batch. Any pages that were skipped due to too recent
FIL_PAGE_LSN or due to them being latched by a writer should be flushed
during the next batch, unless there are further modifications to those
pages. It is possible that a page that we must flush due to small
oldest_modification also carries a recent FIL_PAGE_LSN or is being
constantly modified. In the worst case, all writers would then end up
waiting in log_free_check() to allow the flushing and the checkpoint
to complete.
buf_do_flush_list_batch(): Clarify comments, and rename min_n to max_n.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_flush_space(): Auxiliary function to look up a tablespace for
page flushing.
buf_flush_page(): Defer the computation of space->full_crc32(). Never
call log_write_up_to(), but instead skip persistent pages whose latest
modification (FIL_PAGE_LSN) is newer than the redo log. Also skip
pages on which we cannot acquire a shared latch without waiting.
buf_flush_try_neighbors(): Do not bother checking buf_fix_count
because buf_flush_page() will no longer wait for the page latch.
Take the tablespace as a parameter, and only execute this function
when innodb_flush_neighbors>0. Avoid repeated calls of page_id_t::fold().
buf_flush_relocate_on_flush_list(): Declare as cold, and push down
a condition from the callers.
buf_flush_check_neighbor(): Take id.fold() as a parameter.
buf_flush_sync(): Ensure that the buf_pool.flush_list is empty,
because the flushing batch will skip pages whose modifications have
not yet been written to the log or were latched for modification.
buf_free_from_unzip_LRU_list_batch(): Remove redundant local variables.
buf_flush_LRU_list_batch(): Let the caller buf_do_LRU_batch() initialize
the counters, and report n->evicted.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_do_LRU_batch(): Return the number of pages flushed.
buf_LRU_free_page(): Only release and re-acquire buf_pool.mutex if
adaptive hash index entries are pointing to the block.
buf_LRU_get_free_block(): Do not wake up the page cleaner, because it
will no longer perform any useful work for us, and we do not want it
to compete for I/O while buf_flush_lists(innodb_lru_flush_size, 0)
writes out and evicts at most innodb_lru_flush_size pages. (The
function buf_do_LRU_batch() may complete after writing fewer pages if
more than innodb_lru_scan_depth pages end up in buf_pool.free list.)
Eliminate some mutex release-acquire cycles, and wait for the LRU
flush batch to complete before rescanning.
buf_LRU_check_size_of_non_data_objects(): Simplify the code.
buf_page_write_complete(): Remove the parameter evict, and always
evict pages that were part of an LRU flush.
buf_page_create(): Take a pre-allocated page as a parameter.
buf_pool_t::free_block(): Free a pre-allocated block.
recv_sys_t::recover_low(), recv_sys_t::apply(): Preallocate the block
while not holding recv_sys.mutex. During page allocation, we may
initiate a page flush, which in turn may initiate a log flush, which
would require acquiring log_sys.mutex, which should always be acquired
before recv_sys.mutex in order to avoid deadlocks. Therefore, we must
not be holding recv_sys.mutex while allocating a buffer pool block.
BtrBulk::logFreeCheck(): Skip a redundant condition.
row_undo_step(): Do not invoke srv_inc_activity_count() for every row
that is being rolled back. It should suffice to invoke the function in
trx_flush_log_if_needed() during trx_t::commit_in_memory() when the
rollback completes.
sync_check_enable(): Remove. We will enable innodb_sync_debug from the
very beginning.
Reviewed by: Vladislav Vaintroub
5 years ago MDEV-15053 Reduce buf_pool_t::mutex contention
User-visible changes: The INFORMATION_SCHEMA views INNODB_BUFFER_PAGE
and INNODB_BUFFER_PAGE_LRU will report a dummy value FLUSH_TYPE=0
and will no longer report the PAGE_STATE value READY_FOR_USE.
We will remove some fields from buf_page_t and move much code to
member functions of buf_pool_t and buf_page_t, so that the access
rules of data members can be enforced consistently.
Evicting or adding pages in buf_pool.LRU will remain covered by
buf_pool.mutex.
Evicting or adding pages in buf_pool.page_hash will remain
covered by both buf_pool.mutex and the buf_pool.page_hash X-latch.
After this fix, buf_pool.page_hash lookups can entirely
avoid acquiring buf_pool.mutex, only relying on
buf_pool.hash_lock_get() S-latch.
Similarly, buf_flush_check_neighbors() can will rely solely on
buf_pool.mutex, no buf_pool.page_hash latch at all.
The buf_pool.mutex is rather contended in I/O heavy benchmarks,
especially when the workload does not fit in the buffer pool.
The first attempt to alleviate the contention was the
buf_pool_t::mutex split in
commit 4ed7082eefe56b3e97e0edefb3df76dd7ef5e858
which introduced buf_block_t::mutex, which we are now removing.
Later, multiple instances of buf_pool_t were introduced
in commit c18084f71b02ea707c6461353e6cfc15d7553bc6
and recently removed by us in
commit 1a6f708ec594ac0ae2dd30db926ab07b100fa24b (MDEV-15058).
UNIV_BUF_DEBUG: Remove. This option to enable some buffer pool
related debugging in otherwise non-debug builds has not been used
for years. Instead, we have been using UNIV_DEBUG, which is enabled
in CMAKE_BUILD_TYPE=Debug.
buf_block_t::mutex, buf_pool_t::zip_mutex: Remove. We can mainly rely on
std::atomic and the buf_pool.page_hash latches, and in some cases
depend on buf_pool.mutex or buf_pool.flush_list_mutex just like before.
We must always release buf_block_t::lock before invoking
unfix() or io_unfix(), to prevent a glitch where a block that was
added to the buf_pool.free list would apper X-latched. See
commit c5883debd6ef440a037011c11873b396923e93c5 how this glitch
was finally caught in a debug environment.
We move some buf_pool_t::page_hash specific code from the
ha and hash modules to buf_pool, for improved readability.
buf_pool_t::close(): Assert that all blocks are clean, except
on aborted startup or crash-like shutdown.
buf_pool_t::validate(): No longer attempt to validate
n_flush[] against the number of BUF_IO_WRITE fixed blocks,
because buf_page_t::flush_type no longer exists.
buf_pool_t::watch_set(): Replaces buf_pool_watch_set().
Reduce mutex contention by separating the buf_pool.watch[]
allocation and the insert into buf_pool.page_hash.
buf_pool_t::page_hash_lock<bool exclusive>(): Acquire a
buf_pool.page_hash latch.
Replaces and extends buf_page_hash_lock_s_confirm()
and buf_page_hash_lock_x_confirm().
buf_pool_t::READ_AHEAD_PAGES: Renamed from BUF_READ_AHEAD_PAGES.
buf_pool_t::curr_size, old_size, read_ahead_area, n_pend_reads:
Use Atomic_counter.
buf_pool_t::running_out(): Replaces buf_LRU_buf_pool_running_out().
buf_pool_t::LRU_remove(): Remove a block from the LRU list
and return its predecessor. Incorporates buf_LRU_adjust_hp(),
which was removed.
buf_page_get_gen(): Remove a redundant call of fsp_is_system_temporary(),
for mode == BUF_GET_IF_IN_POOL_OR_WATCH, which is only used by
BTR_DELETE_OP (purge), which is never invoked on temporary tables.
buf_free_from_unzip_LRU_list_batch(): Avoid redundant assignments.
buf_LRU_free_from_unzip_LRU_list(): Simplify the loop condition.
buf_LRU_free_page(): Clarify the function comment.
buf_flush_check_neighbor(), buf_flush_check_neighbors():
Rewrite the construction of the page hash range. We will hold
the buf_pool.mutex for up to buf_pool.read_ahead_area (at most 64)
consecutive lookups of buf_pool.page_hash.
buf_flush_page_and_try_neighbors(): Remove.
Merge to its only callers, and remove redundant operations in
buf_flush_LRU_list_batch().
buf_read_ahead_random(), buf_read_ahead_linear(): Rewrite.
Do not acquire buf_pool.mutex, and iterate directly with page_id_t.
ut_2_power_up(): Remove. my_round_up_to_next_power() is inlined
and avoids any loops.
fil_page_get_prev(), fil_page_get_next(), fil_addr_is_null(): Remove.
buf_flush_page(): Add a fil_space_t* parameter. Minimize the
buf_pool.mutex hold time. buf_pool.n_flush[] is no longer updated
atomically with the io_fix, and we will protect most buf_block_t
fields with buf_block_t::lock. The function
buf_flush_write_block_low() is removed and merged here.
buf_page_init_for_read(): Use static linkage. Initialize the newly
allocated block and acquire the exclusive buf_block_t::lock while not
holding any mutex.
IORequest::IORequest(): Remove the body. We only need to invoke
set_punch_hole() in buf_flush_page() and nowhere else.
buf_page_t::flush_type: Remove. Replaced by IORequest::flush_type.
This field is only used during a fil_io() call.
That function already takes IORequest as a parameter, so we had
better introduce for the rarely changing field.
buf_block_t::init(): Replaces buf_page_init().
buf_page_t::init(): Replaces buf_page_init_low().
buf_block_t::initialise(): Initialise many fields, but
keep the buf_page_t::state(). Both buf_pool_t::validate() and
buf_page_optimistic_get() requires that buf_page_t::in_file()
be protected atomically with buf_page_t::in_page_hash
and buf_page_t::in_LRU_list.
buf_page_optimistic_get(): Now that buf_block_t::mutex
no longer exists, we must check buf_page_t::io_fix()
after acquiring the buf_pool.page_hash lock, to detect
whether buf_page_init_for_read() has been initiated.
We will also check the io_fix() before acquiring hash_lock
in order to avoid unnecessary computation.
The field buf_block_t::modify_clock (protected by buf_block_t::lock)
allows buf_page_optimistic_get() to validate the block.
buf_page_t::real_size: Remove. It was only used while flushing
pages of page_compressed tables.
buf_page_encrypt(): Add an output parameter that allows us ot eliminate
buf_page_t::real_size. Replace a condition with debug assertion.
buf_page_should_punch_hole(): Remove.
buf_dblwr_t::add_to_batch(): Replaces buf_dblwr_add_to_batch().
Add the parameter size (to replace buf_page_t::real_size).
buf_dblwr_t::write_single_page(): Replaces buf_dblwr_write_single_page().
Add the parameter size (to replace buf_page_t::real_size).
fil_system_t::detach(): Replaces fil_space_detach().
Ensure that fil_validate() will not be violated even if
fil_system.mutex is released and reacquired.
fil_node_t::complete_io(): Renamed from fil_node_complete_io().
fil_node_t::close_to_free(): Replaces fil_node_close_to_free().
Avoid invoking fil_node_t::close() because fil_system.n_open
has already been decremented in fil_space_t::detach().
BUF_BLOCK_READY_FOR_USE: Remove. Directly use BUF_BLOCK_MEMORY.
BUF_BLOCK_ZIP_DIRTY: Remove. Directly use BUF_BLOCK_ZIP_PAGE,
and distinguish dirty pages by buf_page_t::oldest_modification().
BUF_BLOCK_POOL_WATCH: Remove. Use BUF_BLOCK_NOT_USED instead.
This state was only being used for buf_page_t that are in
buf_pool.watch.
buf_pool_t::watch[]: Remove pointer indirection.
buf_page_t::in_flush_list: Remove. It was set if and only if
buf_page_t::oldest_modification() is nonzero.
buf_page_decrypt_after_read(), buf_corrupt_page_release(),
buf_page_check_corrupt(): Change the const fil_space_t* parameter
to const fil_node_t& so that we can report the correct file name.
buf_page_monitor(): Declare as an ATTRIBUTE_COLD global function.
buf_page_io_complete(): Split to buf_page_read_complete() and
buf_page_write_complete().
buf_dblwr_t::in_use: Remove.
buf_dblwr_t::buf_block_array: Add IORequest::flush_t.
buf_dblwr_sync_datafiles(): Remove. It was a useless wrapper of
os_aio_wait_until_no_pending_writes().
buf_flush_write_complete(): Declare static, not global.
Add the parameter IORequest::flush_t.
buf_flush_freed_page(): Simplify the code.
recv_sys_t::flush_lru: Renamed from flush_type and changed to bool.
fil_read(), fil_write(): Replaced with direct use of fil_io().
fil_buffering_disabled(): Remove. Check srv_file_flush_method directly.
fil_mutex_enter_and_prepare_for_io(): Return the resolved
fil_space_t* to avoid a duplicated lookup in the caller.
fil_report_invalid_page_access(): Clean up the parameters.
fil_io(): Return fil_io_t, which comprises fil_node_t and error code.
Always invoke fil_space_t::acquire_for_io() and let either the
sync=true caller or fil_aio_callback() invoke
fil_space_t::release_for_io().
fil_aio_callback(): Rewrite to replace buf_page_io_complete().
fil_check_pending_operations(): Remove a parameter, and remove some
redundant lookups.
fil_node_close_to_free(): Wait for n_pending==0. Because we no longer
do an extra lookup of the tablespace between fil_io() and the
completion of the operation, we must give fil_node_t::complete_io() a
chance to decrement the counter.
fil_close_tablespace(): Remove unused parameter trx, and document
that this is only invoked during the error handling of IMPORT TABLESPACE.
row_import_discard_changes(): Merged with the only caller,
row_import_cleanup(). Do not lock up the data dictionary while
invoking fil_close_tablespace().
logs_empty_and_mark_files_at_shutdown(): Do not invoke
fil_close_all_files(), to avoid a !needs_flush assertion failure
on fil_node_t::close().
innodb_shutdown(): Invoke os_aio_free() before fil_close_all_files().
fil_close_all_files(): Invoke fil_flush_file_spaces()
to ensure proper durability.
thread_pool::unbind(): Fix a crash that would occur on Windows
after srv_thread_pool->disable_aio() and os_file_close().
This fix was submitted by Vladislav Vaintroub.
Thanks to Matthias Leich and Axel Schwenke for extensive testing,
Vladislav Vaintroub for helpful comments, and Eugene Kosov for a review.
5 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-27058: Reduce the size of buf_block_t and buf_page_t
buf_page_t::frame: Moved from buf_block_t::frame.
All 'thin' buf_page_t describing compressed-only ROW_FORMAT=COMPRESSED
pages will have frame=nullptr, while all 'fat' buf_block_t
will have a non-null frame pointing to aligned innodb_page_size bytes.
This eliminates the need for separate states for
BUF_BLOCK_FILE_PAGE and BUF_BLOCK_ZIP_PAGE.
buf_page_t::lock: Moved from buf_block_t::lock. That is, all block
descriptors will have a page latch. The IO_PIN state that was used
for discarding or creating the uncompressed page frame of a
ROW_FORMAT=COMPRESSED block is replaced by a combination of read-fix
and page X-latch.
page_zip_des_t::fix: Replaces state_, buf_fix_count_, io_fix_, status
of buf_page_t with a single std::atomic<uint32_t>. All modifications
will use store(), fetch_add(), fetch_sub(). This space was previously
wasted to alignment on 64-bit systems. We will use the following encoding
that combines a state (partly read-fix or write-fix) and a buffer-fix
count:
buf_page_t::NOT_USED=0 (previously BUF_BLOCK_NOT_USED)
buf_page_t::MEMORY=1 (previously BUF_BLOCK_MEMORY)
buf_page_t::REMOVE_HASH=2 (previously BUF_BLOCK_REMOVE_HASH)
buf_page_t::FREED=3 + fix: pages marked as freed in the file
buf_page_t::UNFIXED=1U<<29 + fix: normal pages
buf_page_t::IBUF_EXIST=2U<<29 + fix: normal pages; may need ibuf merge
buf_page_t::REINIT=3U<<29 + fix: reinitialized pages (skip doublewrite)
buf_page_t::READ_FIX=4U<<29 + fix: read-fixed pages (also X-latched)
buf_page_t::WRITE_FIX=5U<<29 + fix: write-fixed pages (also U-latched)
buf_page_t::WRITE_FIX_IBUF=6U<<29 + fix: write-fixed; may have ibuf
buf_page_t::WRITE_FIX_REINIT=7U<<29 + fix: write-fixed (no doublewrite)
buf_page_t::write_complete(): Change WRITE_FIX or WRITE_FIX_REINIT to
UNFIXED, and WRITE_FIX_IBUF to IBUF_EXIST, before releasing the U-latch.
buf_page_t::read_complete(): Renamed from buf_page_read_complete().
Change READ_FIX to UNFIXED or IBUF_EXIST, before releasing the X-latch.
buf_page_t::can_relocate(): If the page latch is being held or waited for,
or the block is buffer-fixed or io-fixed, return false. (The condition
on the page latch is new.)
Outside buf_page_get_gen(), buf_page_get_low() and buf_page_free(), we
will acquire the page latch before fix(), and unfix() before unlocking.
buf_page_t::flush(): Replaces buf_flush_page(). Optimize the
handling of FREED pages.
buf_pool_t::release_freed_page(): Assume that buf_pool.mutex is held
by the caller.
buf_page_t::is_read_fixed(), buf_page_t::is_write_fixed(): New predicates.
buf_page_get_low(): Ignore guesses that are read-fixed because they
may not yet be registered in buf_pool.page_hash and buf_pool.LRU.
buf_page_optimistic_get(): Acquire latch before buffer-fixing.
buf_page_make_young(): Leave read-fixed blocks alone, because they
might not be registered in buf_pool.LRU yet.
recv_sys_t::recover_deferred(), recv_sys_t::recover_low():
Possibly fix MDEV-26326, by holding a page X-latch instead of
only buffer-fixing the page.
4 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-23399: Performance regression with write workloads
The buffer pool refactoring in MDEV-15053 and MDEV-22871 shifted
the performance bottleneck to the page flushing.
The configuration parameters will be changed as follows:
innodb_lru_flush_size=32 (new: how many pages to flush on LRU eviction)
innodb_lru_scan_depth=1536 (old: 1024)
innodb_max_dirty_pages_pct=90 (old: 75)
innodb_max_dirty_pages_pct_lwm=75 (old: 0)
Note: The parameter innodb_lru_scan_depth will only affect LRU
eviction of buffer pool pages when a new page is being allocated. The
page cleaner thread will no longer evict any pages. It used to
guarantee that some pages will remain free in the buffer pool. Now, we
perform that eviction 'on demand' in buf_LRU_get_free_block().
The parameter innodb_lru_scan_depth(srv_LRU_scan_depth) is used as follows:
* When the buffer pool is being shrunk in buf_pool_t::withdraw_blocks()
* As a buf_pool.free limit in buf_LRU_list_batch() for terminating
the flushing that is initiated e.g., by buf_LRU_get_free_block()
The parameter also used to serve as an initial limit for unzip_LRU
eviction (evicting uncompressed page frames while retaining
ROW_FORMAT=COMPRESSED pages), but now we will use a hard-coded limit
of 100 or unlimited for invoking buf_LRU_scan_and_free_block().
The status variables will be changed as follows:
innodb_buffer_pool_pages_flushed: This includes also the count of
innodb_buffer_pool_pages_LRU_flushed and should work reliably,
updated one by one in buf_flush_page() to give more real-time
statistics. The function buf_flush_stats(), which we are removing,
was not called in every code path. For both counters, we will use
regular variables that are incremented in a critical section of
buf_pool.mutex. Note that show_innodb_vars() directly links to the
variables, and reads of the counters will *not* be protected by
buf_pool.mutex, so you cannot get a consistent snapshot of both variables.
The following INFORMATION_SCHEMA.INNODB_METRICS counters will be
removed, because the page cleaner no longer deals with writing or
evicting least recently used pages, and because the single-page writes
have been removed:
* buffer_LRU_batch_flush_avg_time_slot
* buffer_LRU_batch_flush_avg_time_thread
* buffer_LRU_batch_flush_avg_time_est
* buffer_LRU_batch_flush_avg_pass
* buffer_LRU_single_flush_scanned
* buffer_LRU_single_flush_num_scan
* buffer_LRU_single_flush_scanned_per_call
When moving to a single buffer pool instance in MDEV-15058, we missed
some opportunity to simplify the buf_flush_page_cleaner thread. It was
unnecessarily using a mutex and some complex data structures, even
though we always have a single page cleaner thread.
Furthermore, the buf_flush_page_cleaner thread had separate 'recovery'
and 'shutdown' modes where it was waiting to be triggered by some
other thread, adding unnecessary latency and potential for hangs in
relatively rarely executed startup or shutdown code.
The page cleaner was also running two kinds of batches in an
interleaved fashion: "LRU flush" (writing out some least recently used
pages and evicting them on write completion) and the normal batches
that aim to increase the MIN(oldest_modification) in the buffer pool,
to help the log checkpoint advance.
The buf_pool.flush_list flushing was being blocked by
buf_block_t::lock for no good reason. Furthermore, if the FIL_PAGE_LSN
of a page is ahead of log_sys.get_flushed_lsn(), that is, what has
been persistently written to the redo log, we would trigger a log
flush and then resume the page flushing. This would unnecessarily
limit the performance of the page cleaner thread and trigger the
infamous messages "InnoDB: page_cleaner: 1000ms intended loop took 4450ms.
The settings might not be optimal" that were suppressed in
commit d1ab89037a518fcffbc50c24e4bd94e4ec33aed0 unless log_warnings>2.
Our revised algorithm will make log_sys.get_flushed_lsn() advance at
the start of buf_flush_lists(), and then execute a 'best effort' to
write out all pages. The flush batches will skip pages that were modified
since the log was written, or are are currently exclusively locked.
The MDEV-13670 message "page_cleaner: 1000ms intended loop took" message
will be removed, because by design, the buf_flush_page_cleaner() should
not be blocked during a batch for extended periods of time.
We will remove the single-page flushing altogether. Related to this,
the debug parameter innodb_doublewrite_batch_size will be removed,
because all of the doublewrite buffer will be used for flushing
batches. If a page needs to be evicted from the buffer pool and all
100 least recently used pages in the buffer pool have unflushed
changes, buf_LRU_get_free_block() will execute buf_flush_lists() to
write out and evict innodb_lru_flush_size pages. At most one thread
will execute buf_flush_lists() in buf_LRU_get_free_block(); other
threads will wait for that LRU flushing batch to finish.
To improve concurrency, we will replace the InnoDB ib_mutex_t and
os_event_t native mutexes and condition variables in this area of code.
Most notably, this means that the buffer pool mutex (buf_pool.mutex)
is no longer instrumented via any InnoDB interfaces. It will continue
to be instrumented via PERFORMANCE_SCHEMA.
For now, both buf_pool.flush_list_mutex and buf_pool.mutex will be
declared with MY_MUTEX_INIT_FAST (PTHREAD_MUTEX_ADAPTIVE_NP). The critical
sections of buf_pool.flush_list_mutex should be shorter than those for
buf_pool.mutex, because in the worst case, they cover a linear scan of
buf_pool.flush_list, while the worst case of a critical section of
buf_pool.mutex covers a linear scan of the potentially much longer
buf_pool.LRU list.
mysql_mutex_is_owner(), safe_mutex_is_owner(): New predicate, usable
with SAFE_MUTEX. Some InnoDB debug assertions need this predicate
instead of mysql_mutex_assert_owner() or mysql_mutex_assert_not_owner().
buf_pool_t::n_flush_LRU, buf_pool_t::n_flush_list:
Replaces buf_pool_t::init_flush[] and buf_pool_t::n_flush[].
The number of active flush operations.
buf_pool_t::mutex, buf_pool_t::flush_list_mutex: Use mysql_mutex_t
instead of ib_mutex_t, to have native mutexes with PERFORMANCE_SCHEMA
and SAFE_MUTEX instrumentation.
buf_pool_t::done_flush_LRU: Condition variable for !n_flush_LRU.
buf_pool_t::done_flush_list: Condition variable for !n_flush_list.
buf_pool_t::do_flush_list: Condition variable to wake up the
buf_flush_page_cleaner when a log checkpoint needs to be written
or the server is being shut down. Replaces buf_flush_event.
We will keep using timed waits (the page cleaner thread will wake
_at least_ once per second), because the calculations for
innodb_adaptive_flushing depend on fixed time intervals.
buf_dblwr: Allocate statically, and move all code to member functions.
Use a native mutex and condition variable. Remove code to deal with
single-page flushing.
buf_dblwr_check_block(): Make the check debug-only. We were spending
a significant amount of execution time in page_simple_validate_new().
flush_counters_t::unzip_LRU_evicted: Remove.
IORequest: Make more members const. FIXME: m_fil_node should be removed.
buf_flush_sync_lsn: Protect by std::atomic, not page_cleaner.mutex
(which we are removing).
page_cleaner_slot_t, page_cleaner_t: Remove many redundant members.
pc_request_flush_slot(): Replaces pc_request() and pc_flush_slot().
recv_writer_thread: Remove. Recovery works just fine without it, if we
simply invoke buf_flush_sync() at the end of each batch in
recv_sys_t::apply().
recv_recovery_from_checkpoint_finish(): Remove. We can simply call
recv_sys.debug_free() directly.
srv_started_redo: Replaces srv_start_state.
SRV_SHUTDOWN_FLUSH_PHASE: Remove. logs_empty_and_mark_files_at_shutdown()
can communicate with the normal page cleaner loop via the new function
flush_buffer_pool().
buf_flush_remove(): Assert that the calling thread is holding
buf_pool.flush_list_mutex. This removes unnecessary mutex operations
from buf_flush_remove_pages() and buf_flush_dirty_pages(),
which replace buf_LRU_flush_or_remove_pages().
buf_flush_lists(): Renamed from buf_flush_batch(), with simplified
interface. Return the number of flushed pages. Clarified comments and
renamed min_n to max_n. Identify LRU batch by lsn=0. Merge all the functions
buf_flush_start(), buf_flush_batch(), buf_flush_end() directly to this
function, which was their only caller, and remove 2 unnecessary
buf_pool.mutex release/re-acquisition that we used to perform around
the buf_flush_batch() call. At the start, if not all log has been
durably written, wait for a background task to do it, or start a new
task to do it. This allows the log write to run concurrently with our
page flushing batch. Any pages that were skipped due to too recent
FIL_PAGE_LSN or due to them being latched by a writer should be flushed
during the next batch, unless there are further modifications to those
pages. It is possible that a page that we must flush due to small
oldest_modification also carries a recent FIL_PAGE_LSN or is being
constantly modified. In the worst case, all writers would then end up
waiting in log_free_check() to allow the flushing and the checkpoint
to complete.
buf_do_flush_list_batch(): Clarify comments, and rename min_n to max_n.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_flush_space(): Auxiliary function to look up a tablespace for
page flushing.
buf_flush_page(): Defer the computation of space->full_crc32(). Never
call log_write_up_to(), but instead skip persistent pages whose latest
modification (FIL_PAGE_LSN) is newer than the redo log. Also skip
pages on which we cannot acquire a shared latch without waiting.
buf_flush_try_neighbors(): Do not bother checking buf_fix_count
because buf_flush_page() will no longer wait for the page latch.
Take the tablespace as a parameter, and only execute this function
when innodb_flush_neighbors>0. Avoid repeated calls of page_id_t::fold().
buf_flush_relocate_on_flush_list(): Declare as cold, and push down
a condition from the callers.
buf_flush_check_neighbor(): Take id.fold() as a parameter.
buf_flush_sync(): Ensure that the buf_pool.flush_list is empty,
because the flushing batch will skip pages whose modifications have
not yet been written to the log or were latched for modification.
buf_free_from_unzip_LRU_list_batch(): Remove redundant local variables.
buf_flush_LRU_list_batch(): Let the caller buf_do_LRU_batch() initialize
the counters, and report n->evicted.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_do_LRU_batch(): Return the number of pages flushed.
buf_LRU_free_page(): Only release and re-acquire buf_pool.mutex if
adaptive hash index entries are pointing to the block.
buf_LRU_get_free_block(): Do not wake up the page cleaner, because it
will no longer perform any useful work for us, and we do not want it
to compete for I/O while buf_flush_lists(innodb_lru_flush_size, 0)
writes out and evicts at most innodb_lru_flush_size pages. (The
function buf_do_LRU_batch() may complete after writing fewer pages if
more than innodb_lru_scan_depth pages end up in buf_pool.free list.)
Eliminate some mutex release-acquire cycles, and wait for the LRU
flush batch to complete before rescanning.
buf_LRU_check_size_of_non_data_objects(): Simplify the code.
buf_page_write_complete(): Remove the parameter evict, and always
evict pages that were part of an LRU flush.
buf_page_create(): Take a pre-allocated page as a parameter.
buf_pool_t::free_block(): Free a pre-allocated block.
recv_sys_t::recover_low(), recv_sys_t::apply(): Preallocate the block
while not holding recv_sys.mutex. During page allocation, we may
initiate a page flush, which in turn may initiate a log flush, which
would require acquiring log_sys.mutex, which should always be acquired
before recv_sys.mutex in order to avoid deadlocks. Therefore, we must
not be holding recv_sys.mutex while allocating a buffer pool block.
BtrBulk::logFreeCheck(): Skip a redundant condition.
row_undo_step(): Do not invoke srv_inc_activity_count() for every row
that is being rolled back. It should suffice to invoke the function in
trx_flush_log_if_needed() during trx_t::commit_in_memory() when the
rollback completes.
sync_check_enable(): Remove. We will enable innodb_sync_debug from the
very beginning.
Reviewed by: Vladislav Vaintroub
5 years ago MDEV-23399: Performance regression with write workloads
The buffer pool refactoring in MDEV-15053 and MDEV-22871 shifted
the performance bottleneck to the page flushing.
The configuration parameters will be changed as follows:
innodb_lru_flush_size=32 (new: how many pages to flush on LRU eviction)
innodb_lru_scan_depth=1536 (old: 1024)
innodb_max_dirty_pages_pct=90 (old: 75)
innodb_max_dirty_pages_pct_lwm=75 (old: 0)
Note: The parameter innodb_lru_scan_depth will only affect LRU
eviction of buffer pool pages when a new page is being allocated. The
page cleaner thread will no longer evict any pages. It used to
guarantee that some pages will remain free in the buffer pool. Now, we
perform that eviction 'on demand' in buf_LRU_get_free_block().
The parameter innodb_lru_scan_depth(srv_LRU_scan_depth) is used as follows:
* When the buffer pool is being shrunk in buf_pool_t::withdraw_blocks()
* As a buf_pool.free limit in buf_LRU_list_batch() for terminating
the flushing that is initiated e.g., by buf_LRU_get_free_block()
The parameter also used to serve as an initial limit for unzip_LRU
eviction (evicting uncompressed page frames while retaining
ROW_FORMAT=COMPRESSED pages), but now we will use a hard-coded limit
of 100 or unlimited for invoking buf_LRU_scan_and_free_block().
The status variables will be changed as follows:
innodb_buffer_pool_pages_flushed: This includes also the count of
innodb_buffer_pool_pages_LRU_flushed and should work reliably,
updated one by one in buf_flush_page() to give more real-time
statistics. The function buf_flush_stats(), which we are removing,
was not called in every code path. For both counters, we will use
regular variables that are incremented in a critical section of
buf_pool.mutex. Note that show_innodb_vars() directly links to the
variables, and reads of the counters will *not* be protected by
buf_pool.mutex, so you cannot get a consistent snapshot of both variables.
The following INFORMATION_SCHEMA.INNODB_METRICS counters will be
removed, because the page cleaner no longer deals with writing or
evicting least recently used pages, and because the single-page writes
have been removed:
* buffer_LRU_batch_flush_avg_time_slot
* buffer_LRU_batch_flush_avg_time_thread
* buffer_LRU_batch_flush_avg_time_est
* buffer_LRU_batch_flush_avg_pass
* buffer_LRU_single_flush_scanned
* buffer_LRU_single_flush_num_scan
* buffer_LRU_single_flush_scanned_per_call
When moving to a single buffer pool instance in MDEV-15058, we missed
some opportunity to simplify the buf_flush_page_cleaner thread. It was
unnecessarily using a mutex and some complex data structures, even
though we always have a single page cleaner thread.
Furthermore, the buf_flush_page_cleaner thread had separate 'recovery'
and 'shutdown' modes where it was waiting to be triggered by some
other thread, adding unnecessary latency and potential for hangs in
relatively rarely executed startup or shutdown code.
The page cleaner was also running two kinds of batches in an
interleaved fashion: "LRU flush" (writing out some least recently used
pages and evicting them on write completion) and the normal batches
that aim to increase the MIN(oldest_modification) in the buffer pool,
to help the log checkpoint advance.
The buf_pool.flush_list flushing was being blocked by
buf_block_t::lock for no good reason. Furthermore, if the FIL_PAGE_LSN
of a page is ahead of log_sys.get_flushed_lsn(), that is, what has
been persistently written to the redo log, we would trigger a log
flush and then resume the page flushing. This would unnecessarily
limit the performance of the page cleaner thread and trigger the
infamous messages "InnoDB: page_cleaner: 1000ms intended loop took 4450ms.
The settings might not be optimal" that were suppressed in
commit d1ab89037a518fcffbc50c24e4bd94e4ec33aed0 unless log_warnings>2.
Our revised algorithm will make log_sys.get_flushed_lsn() advance at
the start of buf_flush_lists(), and then execute a 'best effort' to
write out all pages. The flush batches will skip pages that were modified
since the log was written, or are are currently exclusively locked.
The MDEV-13670 message "page_cleaner: 1000ms intended loop took" message
will be removed, because by design, the buf_flush_page_cleaner() should
not be blocked during a batch for extended periods of time.
We will remove the single-page flushing altogether. Related to this,
the debug parameter innodb_doublewrite_batch_size will be removed,
because all of the doublewrite buffer will be used for flushing
batches. If a page needs to be evicted from the buffer pool and all
100 least recently used pages in the buffer pool have unflushed
changes, buf_LRU_get_free_block() will execute buf_flush_lists() to
write out and evict innodb_lru_flush_size pages. At most one thread
will execute buf_flush_lists() in buf_LRU_get_free_block(); other
threads will wait for that LRU flushing batch to finish.
To improve concurrency, we will replace the InnoDB ib_mutex_t and
os_event_t native mutexes and condition variables in this area of code.
Most notably, this means that the buffer pool mutex (buf_pool.mutex)
is no longer instrumented via any InnoDB interfaces. It will continue
to be instrumented via PERFORMANCE_SCHEMA.
For now, both buf_pool.flush_list_mutex and buf_pool.mutex will be
declared with MY_MUTEX_INIT_FAST (PTHREAD_MUTEX_ADAPTIVE_NP). The critical
sections of buf_pool.flush_list_mutex should be shorter than those for
buf_pool.mutex, because in the worst case, they cover a linear scan of
buf_pool.flush_list, while the worst case of a critical section of
buf_pool.mutex covers a linear scan of the potentially much longer
buf_pool.LRU list.
mysql_mutex_is_owner(), safe_mutex_is_owner(): New predicate, usable
with SAFE_MUTEX. Some InnoDB debug assertions need this predicate
instead of mysql_mutex_assert_owner() or mysql_mutex_assert_not_owner().
buf_pool_t::n_flush_LRU, buf_pool_t::n_flush_list:
Replaces buf_pool_t::init_flush[] and buf_pool_t::n_flush[].
The number of active flush operations.
buf_pool_t::mutex, buf_pool_t::flush_list_mutex: Use mysql_mutex_t
instead of ib_mutex_t, to have native mutexes with PERFORMANCE_SCHEMA
and SAFE_MUTEX instrumentation.
buf_pool_t::done_flush_LRU: Condition variable for !n_flush_LRU.
buf_pool_t::done_flush_list: Condition variable for !n_flush_list.
buf_pool_t::do_flush_list: Condition variable to wake up the
buf_flush_page_cleaner when a log checkpoint needs to be written
or the server is being shut down. Replaces buf_flush_event.
We will keep using timed waits (the page cleaner thread will wake
_at least_ once per second), because the calculations for
innodb_adaptive_flushing depend on fixed time intervals.
buf_dblwr: Allocate statically, and move all code to member functions.
Use a native mutex and condition variable. Remove code to deal with
single-page flushing.
buf_dblwr_check_block(): Make the check debug-only. We were spending
a significant amount of execution time in page_simple_validate_new().
flush_counters_t::unzip_LRU_evicted: Remove.
IORequest: Make more members const. FIXME: m_fil_node should be removed.
buf_flush_sync_lsn: Protect by std::atomic, not page_cleaner.mutex
(which we are removing).
page_cleaner_slot_t, page_cleaner_t: Remove many redundant members.
pc_request_flush_slot(): Replaces pc_request() and pc_flush_slot().
recv_writer_thread: Remove. Recovery works just fine without it, if we
simply invoke buf_flush_sync() at the end of each batch in
recv_sys_t::apply().
recv_recovery_from_checkpoint_finish(): Remove. We can simply call
recv_sys.debug_free() directly.
srv_started_redo: Replaces srv_start_state.
SRV_SHUTDOWN_FLUSH_PHASE: Remove. logs_empty_and_mark_files_at_shutdown()
can communicate with the normal page cleaner loop via the new function
flush_buffer_pool().
buf_flush_remove(): Assert that the calling thread is holding
buf_pool.flush_list_mutex. This removes unnecessary mutex operations
from buf_flush_remove_pages() and buf_flush_dirty_pages(),
which replace buf_LRU_flush_or_remove_pages().
buf_flush_lists(): Renamed from buf_flush_batch(), with simplified
interface. Return the number of flushed pages. Clarified comments and
renamed min_n to max_n. Identify LRU batch by lsn=0. Merge all the functions
buf_flush_start(), buf_flush_batch(), buf_flush_end() directly to this
function, which was their only caller, and remove 2 unnecessary
buf_pool.mutex release/re-acquisition that we used to perform around
the buf_flush_batch() call. At the start, if not all log has been
durably written, wait for a background task to do it, or start a new
task to do it. This allows the log write to run concurrently with our
page flushing batch. Any pages that were skipped due to too recent
FIL_PAGE_LSN or due to them being latched by a writer should be flushed
during the next batch, unless there are further modifications to those
pages. It is possible that a page that we must flush due to small
oldest_modification also carries a recent FIL_PAGE_LSN or is being
constantly modified. In the worst case, all writers would then end up
waiting in log_free_check() to allow the flushing and the checkpoint
to complete.
buf_do_flush_list_batch(): Clarify comments, and rename min_n to max_n.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_flush_space(): Auxiliary function to look up a tablespace for
page flushing.
buf_flush_page(): Defer the computation of space->full_crc32(). Never
call log_write_up_to(), but instead skip persistent pages whose latest
modification (FIL_PAGE_LSN) is newer than the redo log. Also skip
pages on which we cannot acquire a shared latch without waiting.
buf_flush_try_neighbors(): Do not bother checking buf_fix_count
because buf_flush_page() will no longer wait for the page latch.
Take the tablespace as a parameter, and only execute this function
when innodb_flush_neighbors>0. Avoid repeated calls of page_id_t::fold().
buf_flush_relocate_on_flush_list(): Declare as cold, and push down
a condition from the callers.
buf_flush_check_neighbor(): Take id.fold() as a parameter.
buf_flush_sync(): Ensure that the buf_pool.flush_list is empty,
because the flushing batch will skip pages whose modifications have
not yet been written to the log or were latched for modification.
buf_free_from_unzip_LRU_list_batch(): Remove redundant local variables.
buf_flush_LRU_list_batch(): Let the caller buf_do_LRU_batch() initialize
the counters, and report n->evicted.
Cache the last looked up tablespace. If neighbor flushing is not applicable,
invoke buf_flush_page() directly, avoiding a page lookup in between.
buf_do_LRU_batch(): Return the number of pages flushed.
buf_LRU_free_page(): Only release and re-acquire buf_pool.mutex if
adaptive hash index entries are pointing to the block.
buf_LRU_get_free_block(): Do not wake up the page cleaner, because it
will no longer perform any useful work for us, and we do not want it
to compete for I/O while buf_flush_lists(innodb_lru_flush_size, 0)
writes out and evicts at most innodb_lru_flush_size pages. (The
function buf_do_LRU_batch() may complete after writing fewer pages if
more than innodb_lru_scan_depth pages end up in buf_pool.free list.)
Eliminate some mutex release-acquire cycles, and wait for the LRU
flush batch to complete before rescanning.
buf_LRU_check_size_of_non_data_objects(): Simplify the code.
buf_page_write_complete(): Remove the parameter evict, and always
evict pages that were part of an LRU flush.
buf_page_create(): Take a pre-allocated page as a parameter.
buf_pool_t::free_block(): Free a pre-allocated block.
recv_sys_t::recover_low(), recv_sys_t::apply(): Preallocate the block
while not holding recv_sys.mutex. During page allocation, we may
initiate a page flush, which in turn may initiate a log flush, which
would require acquiring log_sys.mutex, which should always be acquired
before recv_sys.mutex in order to avoid deadlocks. Therefore, we must
not be holding recv_sys.mutex while allocating a buffer pool block.
BtrBulk::logFreeCheck(): Skip a redundant condition.
row_undo_step(): Do not invoke srv_inc_activity_count() for every row
that is being rolled back. It should suffice to invoke the function in
trx_flush_log_if_needed() during trx_t::commit_in_memory() when the
rollback completes.
sync_check_enable(): Remove. We will enable innodb_sync_debug from the
very beginning.
Reviewed by: Vladislav Vaintroub
5 years ago MDEV-29445: Reimplement SET GLOBAL innodb_buffer_pool_size
We deprecate and ignore the parameter innodb_buffer_pool_chunk_size
and let the buffer pool size to be changed in arbitrary 1-megabyte
increments.
innodb_buffer_pool_size_max: A new read-only startup parameter
that specifies the maximum innodb_buffer_pool_size. If 0 or
unspecified, it will default to the specified innodb_buffer_pool_size
rounded up to the allocation unit (2 MiB or 8 MiB). The maximum value
is 4GiB-2MiB on 32-bit systems and 16EiB-8MiB on 64-bit systems.
This maximum is very likely to be limited further by the operating system.
The status variable Innodb_buffer_pool_resize_status will reflect
the status of shrinking the buffer pool. When no shrinking is in
progress, the string will be empty.
Unlike before, the execution of SET GLOBAL innodb_buffer_pool_size
will block until the requested buffer pool size change has been
implemented, or the execution is interrupted by a KILL statement
a client disconnect, or server shutdown. If the
buf_flush_page_cleaner() thread notices that we are running out of
memory, the operation may fail with ER_WRONG_USAGE.
SET GLOBAL innodb_buffer_pool_size will be refused
if the server was started with --large-pages (even if
no HugeTLB pages were successfully allocated). This functionality
is somewhat exercised by the test main.large_pages, which now runs
also on Microsoft Windows. On Linux, explicit HugeTLB mappings are
apparently excluded from the reported Redident Set Size (RSS), and
apparently unshrinkable between mmap(2) and munmap(2).
The buffer pool will be mapped to a contiguous virtual memory area
that will be aligned and partitioned into extents of 8 MiB on
64-bit systems and 2 MiB on 32-bit systems.
Within an extent, the first few innodb_page_size blocks contain
buf_block_t objects that will cover the page frames in the rest
of the extent. The number of such frames is precomputed in the
array first_page_in_extent[] for each innodb_page_size.
In this way, there is a trivial mapping between
page frames and block descriptors and we do not need any
lookup tables like buf_pool.zip_hash or buf_pool_t::chunk_t::map.
We will always allocate the same number of block descriptors for
an extent, even if we do not need all the buf_block_t in the last
extent in case the innodb_buffer_pool_size is not an integer multiple
of the of extents size.
The minimum innodb_buffer_pool_size is 256*5/4 pages. At the default
innodb_page_size=16k this corresponds to 5 MiB. However, now that the
innodb_buffer_pool_size includes the memory allocated for the block
descriptors, the minimum would be innodb_buffer_pool_size=6m.
my_large_virtual_alloc(): A new function, similar to my_large_malloc().
my_virtual_mem_reserve(), my_virtual_mem_commit(),
my_virtual_mem_decommit(), my_virtual_mem_release():
New interface mostly by Vladislav Vaintroub, to separately
reserve and release virtual address space, as well as to
commit and decommit memory within it.
After my_virtual_mem_decommit(), the virtual memory range will be
read-only or unaccessible, depending on whether the build option
cmake -DHAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT=1
has been specified. This option is hard-coded on Microsoft Windows,
where VirtualMemory(MEM_DECOMMIT) will make the memory unaccessible.
On IBM AIX, Linux, Illumos and possibly Apple macOS, the virtual memory
will be zeroed out immediately. On other POSIX-like systems,
madvise(MADV_FREE) will be used if available, to give the operating
system kernel a permission to zero out the virtual memory range.
We prefer immediate freeing so that the reported
resident set size (RSS) of the process will reflect the current
innodb_buffer_pool_size. Shrinking the buffer pool is a rarely
executed resource intensive operation, and the immediate configuration
of the MMU mappings should not incur significant additional penalty.
opt_super_large_pages: Declare only on Solaris. Actually, this is
specific to the SPARC implementation of Solaris, but because we
lack access to a Solaris development environment, we will not revise
this for other MMU and ISA.
buf_pool_t::chunk_t::create(): Remove.
buf_pool_t::create(): Initialize all n_blocks of the buf_pool.free list.
buf_pool_t::allocate(): Renamed from buf_LRU_get_free_only().
buf_pool_t::LRU_warned: Changed to Atomic_relaxed<bool>,
only to be modified by the buf_flush_page_cleaner() thread.
buf_pool_t::shrink(): Attempt to shrink the buffer pool.
There are 3 possible outcomes: SHRINK_DONE (success),
SHRINK_IN_PROGRESS (the caller may keep trying),
and SHRINK_ABORT (we seem to be running out of buffer pool).
While traversing buf_pool.LRU, release the contended
buf_pool.mutex once in every 32 iterations in order to
reduce starvation. Use lru_scan_itr for efficient traversal,
similar to buf_LRU_free_from_common_LRU_list().
buf_pool_t::shrunk(): Update the reduced size of the buffer pool
in a way that is compatible with buf_pool_t::page_guess(),
and invoke my_virtual_mem_decommit().
buf_pool_t::resize(): Before invoking shrink(), run one batch of
buf_flush_page_cleaner() in order to prevent LRU_warn().
Abort if shrink() recommends it, or no blocks were withdrawn in
the past 15 seconds, or the execution of the statement
SET GLOBAL innodb_buffer_pool_size was interrupted.
buf_pool_t::first_to_withdraw: The first block descriptor that is
out of the bounds of the shrunk buffer pool.
buf_pool_t::withdrawn: The list of withdrawn blocks.
If buf_pool_t::resize() is aborted before shrink() completes,
we must be able to resurrect the withdrawn blocks in the free list.
buf_pool_t::contains_zip(): Added a parameter for the
number of least significant pointer bits to disregard,
so that we can find any pointers to within a block
that is supposed to be free.
buf_pool_t::is_shrinking(): Return the total number or blocks that
were withdrawn or are to be withdrawn.
buf_pool_t::to_withdraw(): Return the number of blocks that will need to
be withdrawn.
buf_pool_t::usable_size(): Number of usable pages, considering possible
in-progress attempt at shrinking the buffer pool.
buf_pool_t::page_guess(): Try to buffer-fix a guessed block pointer.
If HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT is set, the pointer will
be validated before being dereferenced.
buf_pool_t::get_info(): Replaces buf_stats_get_pool_info().
innodb_init_param(): Refactored. We must first compute
srv_page_size_shift and then determine the valid bounds of
innodb_buffer_pool_size.
buf_buddy_shrink(): Replaces buf_buddy_realloc().
Part of the work is deferred to buf_buddy_condense_free(),
which is being executed when we are not holding any
buf_pool.page_hash latch.
buf_buddy_condense_free(): Do not relocate blocks.
buf_buddy_free_low(): Do not care about buffer pool shrinking.
This will be handled by buf_buddy_shrink() and
buf_buddy_condense_free().
buf_buddy_alloc_zip(): Assert !buf_pool.contains_zip()
when we are allocating from the binary buddy system.
Previously we were asserting this on multiple recursion levels.
buf_buddy_block_free(), buf_buddy_free_low():
Assert !buf_pool.contains_zip().
buf_buddy_alloc_from(): Remove the redundant parameter j.
buf_flush_LRU_list_batch(): Add the parameter to_withdraw
to keep track of buf_pool.n_blocks_to_withdraw.
buf_do_LRU_batch(): Skip buf_free_from_unzip_LRU_list_batch()
if we are shrinking the buffer pool. In that case, we want
to minimize the page relocations and just finish as quickly
as possible.
trx_purge_attach_undo_recs(): Limit purge_sys.n_pages_handled()
in every iteration, in case the buffer pool is being shrunk
in the middle of a purge batch.
Reviewed by: Debarun Banerjee
7 months ago MDEV-12026: Implement innodb_checksum_algorithm=full_crc32
MariaDB data-at-rest encryption (innodb_encrypt_tables)
had repurposed the same unused data field that was repurposed
in MySQL 5.7 (and MariaDB 10.2) for the Split Sequence Number (SSN)
field of SPATIAL INDEX. Because of this, MariaDB was unable to
support encryption on SPATIAL INDEX pages.
Furthermore, InnoDB page checksums skipped some bytes, and there
are multiple variations and checksum algorithms. By default,
InnoDB accepts all variations of all algorithms that ever existed.
This unnecessarily weakens the page checksums.
We hereby introduce two more innodb_checksum_algorithm variants
(full_crc32, strict_full_crc32) that are special in a way:
When either setting is active, newly created data files will
carry a flag (fil_space_t::full_crc32()) that indicates that
all pages of the file will use a full CRC-32C checksum over the
entire page contents (excluding the bytes where the checksum
is stored, at the very end of the page). Such files will always
use that checksum, no matter what the parameter
innodb_checksum_algorithm is assigned to.
For old files, the old checksum algorithms will continue to be
used. The value strict_full_crc32 will be equivalent to strict_crc32
and the value full_crc32 will be equivalent to crc32.
ROW_FORMAT=COMPRESSED tables will only use the old format.
These tables do not support new features, such as larger
innodb_page_size or instant ADD/DROP COLUMN. They may be
deprecated in the future. We do not want an unnecessary
file format change for them.
The new full_crc32() format also cleans up the MariaDB tablespace
flags. We will reserve flags to store the page_compressed
compression algorithm, and to store the compressed payload length,
so that checksum can be computed over the compressed (and
possibly encrypted) stream and can be validated without
decrypting or decompressing the page.
In the full_crc32 format, there no longer are separate before-encryption
and after-encryption checksums for pages. The single checksum is
computed on the page contents that is written to the file.
We do not make the new algorithm the default for two reasons.
First, MariaDB 10.4.2 was a beta release, and the default values
of parameters should not change after beta. Second, we did not
yet implement the full_crc32 format for page_compressed pages.
This will be fixed in MDEV-18644.
This is joint work with Marko Mäkelä.
7 years ago MDEV-12026: Implement innodb_checksum_algorithm=full_crc32
MariaDB data-at-rest encryption (innodb_encrypt_tables)
had repurposed the same unused data field that was repurposed
in MySQL 5.7 (and MariaDB 10.2) for the Split Sequence Number (SSN)
field of SPATIAL INDEX. Because of this, MariaDB was unable to
support encryption on SPATIAL INDEX pages.
Furthermore, InnoDB page checksums skipped some bytes, and there
are multiple variations and checksum algorithms. By default,
InnoDB accepts all variations of all algorithms that ever existed.
This unnecessarily weakens the page checksums.
We hereby introduce two more innodb_checksum_algorithm variants
(full_crc32, strict_full_crc32) that are special in a way:
When either setting is active, newly created data files will
carry a flag (fil_space_t::full_crc32()) that indicates that
all pages of the file will use a full CRC-32C checksum over the
entire page contents (excluding the bytes where the checksum
is stored, at the very end of the page). Such files will always
use that checksum, no matter what the parameter
innodb_checksum_algorithm is assigned to.
For old files, the old checksum algorithms will continue to be
used. The value strict_full_crc32 will be equivalent to strict_crc32
and the value full_crc32 will be equivalent to crc32.
ROW_FORMAT=COMPRESSED tables will only use the old format.
These tables do not support new features, such as larger
innodb_page_size or instant ADD/DROP COLUMN. They may be
deprecated in the future. We do not want an unnecessary
file format change for them.
The new full_crc32() format also cleans up the MariaDB tablespace
flags. We will reserve flags to store the page_compressed
compression algorithm, and to store the compressed payload length,
so that checksum can be computed over the compressed (and
possibly encrypted) stream and can be validated without
decrypting or decompressing the page.
In the full_crc32 format, there no longer are separate before-encryption
and after-encryption checksums for pages. The single checksum is
computed on the page contents that is written to the file.
We do not make the new algorithm the default for two reasons.
First, MariaDB 10.4.2 was a beta release, and the default values
of parameters should not change after beta. Second, we did not
yet implement the full_crc32 format for page_compressed pages.
This will be fixed in MDEV-18644.
This is joint work with Marko Mäkelä.
7 years ago |
|
/*****************************************************************************
Copyright (c) 1995, 2018, Oracle and/or its affiliates. All Rights Reserved.Copyright (c) 2013, 2022, MariaDB Corporation.
This program is free software; you can redistribute it and/or modify it underthe terms of the GNU General Public License as published by the Free SoftwareFoundation; version 2 of the License.
This program is distributed in the hope that it will be useful, but WITHOUTANY WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESSFOR A PARTICULAR PURPOSE. See the GNU General Public License for more details.
You should have received a copy of the GNU General Public License along withthis program; if not, write to the Free Software Foundation, Inc.,51 Franklin Street, Fifth Floor, Boston, MA 02110-1335 USA
*****************************************************************************/
/**************************************************//**
@file buf/buf0buf.ccThe database buffer buf_pool
Created 11/5/1995 Heikki Tuuri*******************************************************/
#include "assume_aligned.h"
#include "mtr0types.h"
#include "mach0data.h"
#include "buf0checksum.h"
#include "mariadb_stats.h"
#include <string.h>
#ifdef UNIV_INNOCHECKSUM
# include "my_sys.h"
# include "buf0buf.h"
#else
#include "my_cpu.h"
#include "mem0mem.h"
#include "btr0btr.h"
#include "fil0fil.h"
#include "fil0crypt.h"
#include "buf0rea.h"
#include "buf0flu.h"
#include "buf0buddy.h"
#include "buf0dblwr.h"
#include "lock0lock.h"
#include "btr0sea.h"
#include "ibuf0ibuf.h"
#include "log0log.h"
#include "dict0stats_bg.h"
#include "srv0srv.h"
#include "srv0start.h"
#include "dict0dict.h"
#include "log0recv.h"
#include "srv0mon.h"
#include "log0crypt.h"
#include "fil0pagecompress.h"
#endif /* !UNIV_INNOCHECKSUM */
#include "page0zip.h"
#include "buf0dump.h"
#include <map>
#include <sstream>
#include "log.h"
#include "my_virtual_mem.h"
using st_::span;
#ifdef HAVE_LIBNUMA
#include <numa.h>
#include <numaif.h>
struct set_numa_interleave_t{ set_numa_interleave_t() { if (srv_numa_interleave) {
struct bitmask *numa_mems_allowed = numa_get_mems_allowed(); MEM_MAKE_DEFINED(numa_mems_allowed, sizeof *numa_mems_allowed); ib::info() << "Setting NUMA memory policy to" " MPOL_INTERLEAVE"; if (set_mempolicy(MPOL_INTERLEAVE, numa_mems_allowed->maskp, numa_mems_allowed->size) != 0) {
ib::warn() << "Failed to set NUMA memory" " policy to MPOL_INTERLEAVE: " << strerror(errno); } numa_bitmask_free(numa_mems_allowed); } }
~set_numa_interleave_t() { if (srv_numa_interleave) {
ib::info() << "Setting NUMA memory policy to" " MPOL_DEFAULT"; if (set_mempolicy(MPOL_DEFAULT, NULL, 0) != 0) { ib::warn() << "Failed to set NUMA memory" " policy to MPOL_DEFAULT: " << strerror(errno); } } }};
#define NUMA_MEMPOLICY_INTERLEAVE_IN_SCOPE set_numa_interleave_t scoped_numa
#else
#define NUMA_MEMPOLICY_INTERLEAVE_IN_SCOPE
#endif /* HAVE_LIBNUMA */
/*
IMPLEMENTATION OF THE BUFFER POOL =================================
Buffer frames and blocks ------------------------Following the terminology of Gray and Reuter, we call the memoryblocks where file pages are loaded buffer frames. For each bufferframe there is a control block, or shortly, a block, in the buffercontrol array. The control info which does not need to be storedin the file along with the file page, resides in the control block.
Buffer pool struct ------------------The buffer buf_pool contains a single mutex which protects all thecontrol data structures of the buf_pool. The content of a buffer frame isprotected by a separate read-write lock in its control block, though.These locks can be locked and unlocked without owning the buf_pool.mutex.The OS events in the buf_pool struct can be waited for without owning thebuf_pool.mutex.
The buf_pool.mutex is a hot-spot in main memory, causing a lot ofmemory bus traffic on multiprocessor systems when processorsalternately access the mutex. On our Pentium, the mutex is accessedmaybe every 10 microseconds. We gave up the solution to have mutexesfor each control block, for instance, because it seemed to becomplicated.
A solution to reduce mutex contention of the buf_pool.mutex is tocreate a separate mutex for the page hash table. On Pentium,accessing the hash table takes 2 microseconds, about halfof the total buf_pool.mutex hold time.
Control blocks --------------
The control block contains, for instance, the bufferfix countwhich is incremented when a thread wants a file page to be fixedin a buffer frame. The bufferfix operation does not lock thecontents of the frame, however. For this purpose, the controlblock contains a read-write lock.
The buffer frames have to be aligned so that the start memoryaddress of a frame is divisible by the universal page size, whichis a power of two.
The control blocks containing file pages are put to a hash tableaccording to the file address of the page.We could speed up the access to an individual page by using"pointer swizzling": we could replace the page references onnon-leaf index pages by direct pointers to the page, if it existsin the buf_pool. We could make a separate hash table where we couldchain all the page references in non-leaf pages residing in the buf_pool,using the page reference as the hash key,and at the time of reading of a page update the pointers accordingly.Drawbacks of this solution are added complexity and,possibly, extra space required on non-leaf pages for memory pointers.A simpler solution is just to speed up the hash table mechanismin the database, using tables whose size is a power of 2.
Lists of blocks ---------------
There are several lists of control blocks.
The free list (buf_pool.free) contains blocks which are currently notused.
The common LRU list contains all the blocks holding a file pageexcept those for which the bufferfix count is non-zero.The pages are in the LRU list roughly in the order of the lastaccess to the page, so that the oldest pages are at the end of thelist. We also keep a pointer to near the end of the LRU list,which we can use when we want to artificially age a page in thebuf_pool. This is used if we know that some page is not neededagain for some time: we insert the block right after the pointer,causing it to be replaced sooner than would normally be the case.Currently this aging mechanism is used for read-ahead mechanismof pages, and it can also be used when there is a scan of a fulltable which cannot fit in the memory. Putting the pages near theend of the LRU list, we make sure that most of the buf_pool staysin the main memory, undisturbed.
The unzip_LRU list contains a subset of the common LRU list. Theblocks on the unzip_LRU list hold a compressed file page and thecorresponding uncompressed page frame. A block is in unzip_LRU if andonly if the predicate block->page.belongs_to_unzip_LRU()holds. The blocks in unzip_LRU will be in same order as they are inthe common LRU list. That is, each manipulation of the common LRUlist will result in the same manipulation of the unzip_LRU list.
The chain of modified blocks (buf_pool.flush_list) contains the blocksholding persistent file pages that have been modified in the memorybut not written to disk yet. The block with the oldest modificationwhich has not yet been written to disk is at the end of the chain.The access to this list is protected by buf_pool.flush_list_mutex.
The control blocks for uncompressed pages are accessible viabuf_block_t objects that are reachable via buf_pool.chunks[].The control blocks (buf_page_t) of those ROW_FORMAT=COMPRESSED pagesthat are not in buf_pool.flush_list and for which no uncompressedpage has been allocated in buf_pool are only accessible viabuf_pool.LRU.
The chains of free memory blocks (buf_pool.zip_free[]) are used bythe buddy allocator (buf0buddy.cc) to keep track of currently unusedmemory blocks of size 1024..innodb_page_size / 2. Theseblocks are inside the memory blocks of size innodb_page_size and typeBUF_BLOCK_MEMORY that the buddy allocator requests from the bufferpool. The buddy allocator is solely used for allocatingROW_FORMAT=COMPRESSED page frames.
Loading a file page -------------------
First, a victim block for replacement has to be found in thebuf_pool. It is taken from the free list or searched for from theend of the LRU-list. An exclusive lock is reserved for the frame,the io_fix is set in the block fixing the block in buf_pool,and the io-operation for loading the page is queued. The io-handler threadreleases the X-lock on the frame and releases the io_fixwhen the io operation completes.
A thread may request the above operation using the functionbuf_page_get(). It may then continue to request a lock on the frame.The lock is granted when the io-handler releases the x-lock.
Read-ahead ----------
The read-ahead mechanism is intended to be intelligent andisolated from the semantically higher levels of the databaseindex management. From the higher level we only need theinformation if a file page has a natural successor orpredecessor page. On the leaf level of a B-tree index,these are the next and previous pages in the naturalorder of the pages.
Let us first explain the read-ahead mechanism when the leafsof a B-tree are scanned in an ascending or descending order.When a read page is the first time referenced in the buf_pool,the buffer manager checks if it is at the border of a so-calledlinear read-ahead area. The tablespace is divided into theseareas of size 64 blocks, for example. So if the page is at theborder of such an area, the read-ahead mechanism checks ifall the other blocks in the area have been accessed in anascending or descending order. If this is the case, the systemlooks at the natural successor or predecessor of the page,checks if that is at the border of another area, and in this caseissues read-requests for all the pages in that area. Maybewe could relax the condition that all the pages in the areahave to be accessed: if data is deleted from a table, there mayappear holes of unused pages in the area.
A different read-ahead mechanism is used when there appearsto be a random access pattern to a file.If a new page is referenced in the buf_pool, and several pagesof its random access area (for instance, 32 consecutive pagesin a tablespace) have recently been referenced, we may predictthat the whole area may be needed in the near future, and issuethe read requests for the whole area.*/
#ifndef UNIV_INNOCHECKSUM
/** Compute the number of page frames needed for buf_block_t,
per innodb_buffer_pool_extent_size.@param ps innodb_page_size@return number of buf_block_t frames per extent */static constexpr uint8_t first_page(size_t ps){ return uint8_t(innodb_buffer_pool_extent_size / ps - innodb_buffer_pool_extent_size / (ps + sizeof(buf_block_t)));}
/** Compute the number of bytes needed for buf_block_t,
per innodb_buffer_pool_extent_size.@param ps innodb_page_size@return number of buf_block_t frames per extent */static constexpr size_t first_frame(size_t ps){ return first_page(ps) * ps;}
/** Compute the number of pages per innodb_buffer_pool_extent_size.
@param ps innodb_page_size@return number of buf_block_t frames per extent */static constexpr uint16_t pages(size_t ps){ return uint16_t(innodb_buffer_pool_extent_size / ps - first_page(ps));}
/** The byte offset of the first page frame in a buffer pool extent
of innodb_buffer_pool_extent_size bytes */static constexpr size_t first_frame_in_extent[]={ first_frame(4096), first_frame(8192), first_frame(16384), first_frame(32768), first_frame(65536)};
/** The position offset of the first page frame in a buffer pool extent
of innodb_buffer_pool_extent_size bytes */static constexpr uint8_t first_page_in_extent[]={ first_page(4096), first_page(8192), first_page(16384), first_page(32768), first_page(65536)};
/** Number of pages per buffer pool extent
of innodb_buffer_pool_extent_size bytes */static constexpr size_t pages_in_extent[]={ pages(4096), pages(8192), pages(16384), pages(32768), pages(65536)};
# ifdef SUX_LOCK_GENERIC
void page_hash_latch::read_lock_wait() noexcept{ /* First, try busy spinning for a while. */ for (auto spin= srv_n_spin_wait_rounds; spin--; ) { LF_BACKOFF(); if (read_trylock()) return; } /* Fall back to yielding to other threads. */ do std::this_thread::yield(); while (!read_trylock());}
void page_hash_latch::write_lock_wait() noexcept{ write_lock_wait_start();
/* First, try busy spinning for a while. */ for (auto spin= srv_n_spin_wait_rounds; spin--; ) { if (write_lock_poll()) return; LF_BACKOFF(); }
/* Fall back to yielding to other threads. */ do std::this_thread::yield(); while (!write_lock_poll());}# endif
/** Number of attempts made to read in a page in the buffer pool */constexpr ulint BUF_PAGE_READ_MAX_RETRIES= 100;/** The maximum portion of the buffer pool that can be used for the
read-ahead buffer. (Divide buf_pool size by this amount) */constexpr uint32_t BUF_READ_AHEAD_PORTION= 32;
/** A 64KiB buffer of NUL bytes, for use in assertions and checks,
and dummy default values of instantly dropped columns.Initially, BLOB field references are set to NUL bytes, indtuple_convert_big_rec(). */const byte *field_ref_zero;
/** The InnoDB buffer pool */buf_pool_t buf_pool;
#ifdef UNIV_DEBUG
/** This is used to insert validation operations in execution
in the debug version */static Atomic_counter<size_t> buf_dbg_counter;#endif /* UNIV_DEBUG */
/** Macro to determine whether the read of write counter is used depending
on the io_type */#define MONITOR_RW_COUNTER(read, counter) \
(read ? (counter##_READ) : (counter##_WRITTEN))
/** Decrypt a page for temporary tablespace.
@param[in,out] tmp_frame Temporary buffer@param[in] src_frame Page to decrypt@return true if temporary tablespace decrypted, false if not */static bool buf_tmp_page_decrypt(byte* tmp_frame, byte* src_frame){ if (buf_is_zeroes(span<const byte>(src_frame, srv_page_size))) { return true; }
/* read space & lsn */ uint header_len = FIL_PAGE_FILE_FLUSH_LSN_OR_KEY_VERSION;
/* Copy FIL page header, it is not encrypted */ memcpy(tmp_frame, src_frame, header_len);
/* Calculate the offset where decryption starts */ const byte* src = src_frame + header_len; byte* dst = tmp_frame + header_len; uint srclen = uint(srv_page_size) - (header_len + FIL_PAGE_FCRC32_CHECKSUM); ulint offset = mach_read_from_4(src_frame + FIL_PAGE_OFFSET);
if (!log_tmp_block_decrypt(src, srclen, dst, (offset * srv_page_size))) { return false; }
static_assert(FIL_PAGE_FCRC32_CHECKSUM == 4, "alignment"); memcpy_aligned<4>(tmp_frame + srv_page_size - FIL_PAGE_FCRC32_CHECKSUM, src_frame + srv_page_size - FIL_PAGE_FCRC32_CHECKSUM, FIL_PAGE_FCRC32_CHECKSUM);
memcpy_aligned<UNIV_PAGE_SIZE_MIN>(src_frame, tmp_frame, srv_page_size); srv_stats.pages_decrypted.inc(); srv_stats.n_temp_blocks_decrypted.inc();
return true; /* page was decrypted */}
/** Decrypt a page.
@param[in,out] bpage Page control block@param[in] node data file@return whether the operation was successful */static bool buf_page_decrypt_after_read(buf_page_t *bpage, const fil_node_t &node){ ut_ad(node.space->referenced()); ut_ad(node.space->id == bpage->id().space()); const auto flags = node.space->flags;
byte* dst_frame = bpage->zip.data ? bpage->zip.data : bpage->frame; bool page_compressed = node.space->is_compressed() && buf_page_is_compressed(dst_frame, flags); const page_id_t id(bpage->id());
if (id.page_no() == 0) { /* File header pages are not encrypted/compressed */ return (true); }
buf_tmp_buffer_t* slot;
if (id.space() == SRV_TMP_SPACE_ID && innodb_encrypt_temporary_tables) { slot = buf_pool.io_buf_reserve(false); slot->allocate(); bool ok = buf_tmp_page_decrypt(slot->crypt_buf, dst_frame); slot->release(); return ok; }
/* Page is encrypted if encryption information is found from
tablespace and page contains used key_version. This is true also for pages first compressed and then encrypted. */
uint key_version = buf_page_get_key_version(dst_frame, flags);
if (page_compressed && !key_version) { /* the page we read is unencrypted */ /* Find free slot from temporary memory array */decompress: if (fil_space_t::full_crc32(flags) && buf_page_is_corrupted(true, dst_frame, flags)) { return false; }
slot = buf_pool.io_buf_reserve(false); slot->allocate();
decompress_with_slot: ulint write_size = fil_page_decompress( slot->crypt_buf, dst_frame, flags); slot->release(); ut_ad(node.space->referenced()); return write_size != 0; }
if (key_version && node.space->crypt_data) { /* Verify encryption checksum before we even try to
decrypt. */ if (!buf_page_verify_crypt_checksum(dst_frame, flags)) {decrypt_failed: ib::error() << "Encrypted page " << id << " in file " << node.name << " looks corrupted; key_version=" << key_version; return false; }
slot = buf_pool.io_buf_reserve(false); slot->allocate();
/* decrypt using crypt_buf to dst_frame */ if (!fil_space_decrypt(node.space, slot->crypt_buf, dst_frame)) { slot->release(); goto decrypt_failed; }
if ((fil_space_t::full_crc32(flags) && page_compressed) || fil_page_get_type(dst_frame) == FIL_PAGE_PAGE_COMPRESSED_ENCRYPTED) { goto decompress_with_slot; }
slot->release(); } else if (fil_page_get_type(dst_frame) == FIL_PAGE_PAGE_COMPRESSED_ENCRYPTED) { goto decompress; }
ut_ad(node.space->referenced()); return true;}#endif /* !UNIV_INNOCHECKSUM */
/** Checks if the page is in crc32 checksum format.
@param[in] read_buf database page@param[in] checksum_field1 new checksum field@param[in] checksum_field2 old checksum field@return true if the page is in crc32 checksum format. */staticboolbuf_page_is_checksum_valid_crc32( const byte* read_buf, ulint checksum_field1, ulint checksum_field2) noexcept{ const uint32_t crc32 = buf_calc_page_crc32(read_buf);
#ifdef UNIV_INNOCHECKSUM
extern FILE* log_file; extern uint32_t cur_page_num; if (log_file) { fprintf(log_file, "page::" UINT32PF ";" " crc32 calculated = " UINT32PF ";" " recorded checksum field1 = " ULINTPF " recorded" " checksum field2 =" ULINTPF "\n", cur_page_num, crc32, checksum_field1, checksum_field2); }#endif /* UNIV_INNOCHECKSUM */
if (checksum_field1 != checksum_field2) { return false; }
return checksum_field1 == crc32;}
#ifndef UNIV_INNOCHECKSUM
/** Check whether a page is newer than the durable LSN.
@param check_lsn whether to check the LSN@param read_buf page frame@return whether the FIL_PAGE_LSN is invalid (ahead of the durable LSN) */static bool buf_page_check_lsn(bool check_lsn, const byte *read_buf) noexcept{ if (!check_lsn) return false; /* A page may not be read before it is written, and it may not be
written before the corresponding log has been durably written. Hence, we refer to the current durable LSN here */ lsn_t current_lsn= log_sys.get_flushed_lsn(std::memory_order_relaxed); if (UNIV_UNLIKELY(current_lsn == log_sys.FIRST_LSN) && srv_force_recovery == SRV_FORCE_NO_LOG_REDO) return false; const lsn_t page_lsn= mach_read_from_8(read_buf + FIL_PAGE_LSN);
if (UNIV_LIKELY(current_lsn >= page_lsn)) return false;
const uint32_t space_id= mach_read_from_4(read_buf + FIL_PAGE_SPACE_ID); const uint32_t page_no= mach_read_from_4(read_buf + FIL_PAGE_OFFSET);
sql_print_error("InnoDB: Page " "[page id: space=" UINT32PF ", page number=" UINT32PF "]" " log sequence number " LSN_PF " is in the future! Current system log sequence number " LSN_PF ".", space_id, page_no, page_lsn, current_lsn);
if (srv_force_recovery) return false;
sql_print_error("InnoDB: Your database may be corrupt or" " you may have copied the InnoDB" " tablespace but not the ib_logfile0. %s", FORCE_RECOVERY_MSG);
return true;}#endif
/** Check if a buffer is all zeroes.
@param[in] buf data to check@return whether the buffer is all zeroes */bool buf_is_zeroes(span<const byte> buf) noexcept{ ut_ad(buf.size() <= UNIV_PAGE_SIZE_MAX); return memcmp(buf.data(), field_ref_zero, buf.size()) == 0;}
/** Check if a page is corrupt.
@param check_lsn whether FIL_PAGE_LSN should be checked@param read_buf database page@param fsp_flags contents of FIL_SPACE_FLAGS@return whether the page is corrupted */buf_page_is_corrupted_reasonbuf_page_is_corrupted(bool check_lsn, const byte *read_buf, uint32_t fsp_flags) noexcept{ if (fil_space_t::full_crc32(fsp_flags)) { bool compressed = false, corrupted = false; const uint size = buf_page_full_crc32_size( read_buf, &compressed, &corrupted); if (corrupted) { return CORRUPTED_OTHER; } const byte* end = read_buf + (size - FIL_PAGE_FCRC32_CHECKSUM); uint crc32 = mach_read_from_4(end);
if (!crc32 && size == srv_page_size && buf_is_zeroes(span<const byte>(read_buf, size))) { return NOT_CORRUPTED; }
DBUG_EXECUTE_IF( "page_intermittent_checksum_mismatch", { static int page_counter; if (mach_read_from_4(FIL_PAGE_OFFSET + read_buf) && page_counter++ == 3) { crc32++; } });
if (crc32 != my_crc32c(0, read_buf, size - FIL_PAGE_FCRC32_CHECKSUM)) { return CORRUPTED_OTHER; } static_assert(FIL_PAGE_FCRC32_KEY_VERSION == 0, "alignment"); static_assert(FIL_PAGE_LSN % 4 == 0, "alignment"); static_assert(FIL_PAGE_FCRC32_END_LSN % 4 == 0, "alignment"); if (!compressed && !mach_read_from_4(FIL_PAGE_FCRC32_KEY_VERSION + read_buf) && memcmp_aligned<4>(read_buf + (FIL_PAGE_LSN + 4), end - (FIL_PAGE_FCRC32_END_LSN - FIL_PAGE_FCRC32_CHECKSUM), 4)) { return CORRUPTED_OTHER; }
return#ifndef UNIV_INNOCHECKSUM
buf_page_check_lsn(check_lsn, read_buf) ? CORRUPTED_FUTURE_LSN :#endif
NOT_CORRUPTED; }
const ulint zip_size = fil_space_t::zip_size(fsp_flags); const uint16_t page_type = fil_page_get_type(read_buf);
/* We can trust page type if page compression is set on tablespace
flags because page compression flag means file must have been created with 10.1 (later than 5.5 code base). In 10.1 page compressed tables do not contain post compression checksum and FIL_PAGE_END_LSN_OLD_CHKSUM field stored. Note that space can be null if we are in fil_check_first_page() and first page is not compressed or encrypted. Page checksum is verified after decompression (i.e. normally pages are already decompressed at this stage). */ if ((page_type == FIL_PAGE_PAGE_COMPRESSED || page_type == FIL_PAGE_PAGE_COMPRESSED_ENCRYPTED)#ifndef UNIV_INNOCHECKSUM
&& FSP_FLAGS_HAS_PAGE_COMPRESSION(fsp_flags)#endif
) { check_lsn: return#ifndef UNIV_INNOCHECKSUM
buf_page_check_lsn(check_lsn, read_buf) ? CORRUPTED_FUTURE_LSN :#endif
NOT_CORRUPTED; }
static_assert(FIL_PAGE_LSN % 4 == 0, "alignment"); static_assert(FIL_PAGE_END_LSN_OLD_CHKSUM % 4 == 0, "alignment");
if (!zip_size && memcmp_aligned<4>(read_buf + FIL_PAGE_LSN + 4, read_buf + srv_page_size - FIL_PAGE_END_LSN_OLD_CHKSUM + 4, 4)) { /* Stored log sequence numbers at the start and the end
of page do not match */
return CORRUPTED_OTHER; }
/* Check whether the checksum fields have correct values */
if (zip_size) { if (!page_zip_verify_checksum(read_buf, zip_size)) { return CORRUPTED_OTHER; } goto check_lsn; }
const uint32_t checksum_field1 = mach_read_from_4( read_buf + FIL_PAGE_SPACE_OR_CHKSUM);
const uint32_t checksum_field2 = mach_read_from_4( read_buf + srv_page_size - FIL_PAGE_END_LSN_OLD_CHKSUM);
static_assert(FIL_PAGE_LSN % 8 == 0, "alignment");
/* A page filled with NUL bytes is considered not corrupted.
Before MariaDB Server 10.1.25 (MDEV-12113) or 10.2.2 (or MySQL 5.7), the FIL_PAGE_FILE_FLUSH_LSN field may have been written nonzero for the first page of each file of the system tablespace. We want to ignore it for the system tablespace, but because we do not know the expected tablespace here, we ignore the field for all data files, except for innodb_checksum_algorithm=full_crc32 which we handled above. */ if (!checksum_field1 && !checksum_field2) { /* Checksum fields can have valid value as zero.
If the page is not empty then do the checksum calculation for the page. */ bool all_zeroes = true; for (size_t i = 0; i < srv_page_size; i++) {#ifndef UNIV_INNOCHECKSUM
if (i == FIL_PAGE_FILE_FLUSH_LSN_OR_KEY_VERSION) { i += 8; }#endif
if (read_buf[i]) { all_zeroes = false; break; } }
if (all_zeroes) { return NOT_CORRUPTED; } }
#ifndef UNIV_INNOCHECKSUM
switch (srv_checksum_algorithm) { case SRV_CHECKSUM_ALGORITHM_STRICT_FULL_CRC32: case SRV_CHECKSUM_ALGORITHM_STRICT_CRC32:#endif /* !UNIV_INNOCHECKSUM */
if (!buf_page_is_checksum_valid_crc32(read_buf, checksum_field1, checksum_field2)) { return CORRUPTED_OTHER; } goto check_lsn;#ifndef UNIV_INNOCHECKSUM
default: if (checksum_field1 == BUF_NO_CHECKSUM_MAGIC && checksum_field2 == BUF_NO_CHECKSUM_MAGIC) { goto check_lsn; }
const uint32_t crc32 = buf_calc_page_crc32(read_buf);
/* Very old versions of InnoDB only stored 8 byte lsn to the
start and the end of the page. */
/* Since innodb_checksum_algorithm is not strict_* allow
any of the algos to match for the old field */
if (checksum_field2 != mach_read_from_4(read_buf + FIL_PAGE_LSN) && checksum_field2 != BUF_NO_CHECKSUM_MAGIC) {
DBUG_EXECUTE_IF( "page_intermittent_checksum_mismatch", { static int page_counter; if (mach_read_from_4(FIL_PAGE_OFFSET + read_buf) && page_counter++ == 3) return CORRUPTED_OTHER; });
if ((checksum_field1 != crc32 || checksum_field2 != crc32) && checksum_field2 != buf_calc_page_old_checksum(read_buf)) { return CORRUPTED_OTHER; } }
switch (checksum_field1) { case 0: case BUF_NO_CHECKSUM_MAGIC: break; default: if ((checksum_field1 != crc32 || checksum_field2 != crc32) && checksum_field1 != buf_calc_page_new_checksum(read_buf)) { return CORRUPTED_OTHER; } } }#endif /* !UNIV_INNOCHECKSUM */
goto check_lsn;}
#ifndef UNIV_INNOCHECKSUM
#ifdef __linux__
#include <poll.h>
#include <sys/eventfd.h>
#include <fstream>
/** Memory Pressure
based off https://www.kernel.org/doc/html/latest/accounting/psi.html#pressure-interface
and https://www.kernel.org/doc/html/latest/admin-guide/cgroup-v2.html#memory */
class mem_pressure{ /* triggers + eventfd */ struct pollfd m_fds[3]; nfds_t m_num_fds; int m_event_fd= -1; Atomic_relaxed<bool> m_abort= false;
std::thread m_thd; /* mem pressure garbage collection restricted to interval */ static constexpr ulonglong max_interval_us= 60*1000000;
public: mem_pressure() : m_num_fds(0) {}
bool setup() { m_num_fds= 0;
if (my_use_large_pages) return false;
static_assert(array_elements(m_fds) == (array_elements(m_triggers) + 1), "insufficient fds"); std::string memcgroup{"/sys/fs/cgroup"}; std::string cgroup; { std::ifstream selfcgroup("/proc/self/cgroup"); std::getline(selfcgroup, cgroup, '\n'); }
cgroup.erase(0, 3); // Remove "0::"
memcgroup+= cgroup + "/memory.pressure";
for (auto trig= std::begin(m_triggers); trig!= std::end(m_triggers); ++trig) { if ((m_fds[m_num_fds].fd= open(memcgroup.c_str(), O_RDWR | O_NONBLOCK | O_CLOEXEC)) < 0) { /* User can't do anything about it, no point giving warning */ shutdown(); return false; } my_register_filename(m_fds[m_num_fds].fd, memcgroup.c_str(), FILE_BY_OPEN, 0, MYF(0)); ssize_t slen= strlen(*trig); if (write(m_fds[m_num_fds].fd, *trig, slen) < slen) { /* we may fail this one, but continue to the next */ my_close(m_fds[m_num_fds].fd, MYF(MY_WME)); continue; } m_fds[m_num_fds].events= POLLPRI; m_num_fds++; } if (m_num_fds < 1) return false;
if ((m_event_fd= eventfd(0, EFD_CLOEXEC|EFD_NONBLOCK)) == -1) { /* User can't do anything about it, no point giving warning */ shutdown(); return false; } my_register_filename(m_event_fd, "mem_pressure_eventfd", FILE_BY_DUP, 0, MYF(0)); m_fds[m_num_fds].fd= m_event_fd; m_fds[m_num_fds].events= POLLIN; m_num_fds++; m_thd= std::thread(pressure_routine, this); sql_print_information("InnoDB: Initialized memory pressure event listener"); return true; }
void shutdown() { /* m_event_fd is in this list */ while (m_num_fds) { m_num_fds--; my_close(m_fds[m_num_fds].fd, MYF(MY_WME)); m_fds[m_num_fds].fd= -1; } m_event_fd= -1; }
static void pressure_routine(mem_pressure *m);
#ifdef UNIV_DEBUG
void trigger_collection() { uint64_t u= 1; if (m_event_fd < 0 || write(m_event_fd, &u, sizeof(uint64_t)) != sizeof(uint64_t)) sql_print_information("InnoDB: (Debug) Failed to trigger memory pressure"); }#endif
void quit() { uint64_t u= 1; m_abort= true;#pragma GCC diagnostic push
#pragma GCC diagnostic ignored "-Wunused-result"
/* return result ignored, cannot do anything with it */ write(m_event_fd, &u, sizeof(uint64_t));#pragma GCC diagnostic pop
}
void join() { if (m_thd.joinable()) { quit(); m_thd.join(); } }
static const char* const m_triggers[2];};
/*
ref: https://docs.kernel.org/accounting/psi.html
maximum window size (second number) 10 seconds. window size in multiples of 2 second interval required (for Unprivileged) Time is in usec.*/const char* const mem_pressure::m_triggers[]= {"some 5000000 10000000", /* 5s out of 10s */ "full 10000 2000000"}; /* 10ms out of 2s */
static mem_pressure mem_pressure_obj;
void mem_pressure::pressure_routine(mem_pressure *m){ DBUG_ASSERT(m == &mem_pressure_obj); if (my_thread_init()) { m->shutdown(); return; }
ulonglong last= microsecond_interval_timer() - max_interval_us; while (!m->m_abort) { if (poll(&m->m_fds[0], m->m_num_fds, -1) < 0) { if (errno == EINTR) continue; else break; } if (m->m_abort) break;
for (pollfd &p : st_::span<pollfd>(m->m_fds, m->m_num_fds)) { if (p.revents & POLLPRI) { ulonglong now= microsecond_interval_timer(); if ((now - last) > max_interval_us) { last= now; buf_pool.garbage_collect(); } }
#ifdef UNIV_DEBUG
if (p.revents & POLLIN) { uint64_t u; /* we haven't aborted, so this must be a debug trigger */ if (read(p.fd, &u, sizeof(u)) >=0) buf_pool.garbage_collect(); }#endif
} } m->shutdown();
my_thread_end();}
/** Initialize mem pressure. */ATTRIBUTE_COLD static void buf_mem_pressure_detect_init() noexcept{ mem_pressure_obj.setup();}
ATTRIBUTE_COLD void buf_mem_pressure_shutdown() noexcept{ mem_pressure_obj.join();}#endif
#if defined __linux__ || !defined DBUG_OFF
inline void buf_pool_t::garbage_collect() noexcept{ mysql_mutex_lock(&mutex); const size_t old_size{size_in_bytes}, min_size{size_in_bytes_auto_min}; const size_t reduce_size= std::max(innodb_buffer_pool_extent_size, ut_calc_align((old_size - min_size) / 2, innodb_buffer_pool_extent_size)); if (old_size < min_size + reduce_size || first_to_withdraw || old_size != size_in_bytes_requested) { mysql_mutex_unlock(&mutex); sql_print_information("InnoDB: Memory pressure event disregarded;" " innodb_buffer_pool_size=%zum," " innodb_buffer_pool_size_min=%zum", old_size >> 20, min_size >> 20); return; }
size_t size= old_size - reduce_size; size_t n_blocks_new= get_n_blocks(size);
ut_ad(UT_LIST_GET_LEN(withdrawn) == 0); ut_ad(n_blocks_to_withdraw == 0);
n_blocks_to_withdraw= n_blocks - n_blocks_new; first_to_withdraw= &get_nth_page(n_blocks_new)->page;
size_in_bytes_requested= size; mysql_mutex_unlock(&mutex); mysql_mutex_lock(&flush_list_mutex); page_cleaner_wakeup(true); my_cond_wait(&done_flush_list, &flush_list_mutex.m_mutex); mysql_mutex_unlock(&flush_list_mutex);# ifdef BTR_CUR_HASH_ADAPT
bool ahi_disabled= btr_search_disable();# endif /* BTR_CUR_HASH_ADAPT */
time_t start= time(nullptr); mysql_mutex_lock(&mutex);
do { if (shrink(size)) { const size_t old_blocks{n_blocks}; n_blocks= n_blocks_new;
size_t s= n_blocks_new / BUF_READ_AHEAD_PORTION; read_ahead_area= s >= READ_AHEAD_PAGES ? READ_AHEAD_PAGES : my_round_up_to_next_power(uint32(s));
os_total_large_mem_allocated-= reduce_size; shrunk(size, reduce_size); ibuf_max_size_update(srv_change_buffer_max_size);# ifdef BTR_CUR_HASH_ADAPT
if (ahi_disabled) btr_search_enable(true);# endif
mysql_mutex_unlock(&mutex); sql_print_information("InnoDB: Memory pressure event shrunk" " innodb_buffer_pool_size=%zum (%zu pages)" " from %zum (%zu pages)", size >> 20, n_blocks_new, old_size >> 20, old_blocks); ut_d(validate()); return; } } while (time(nullptr) - start < 15);
ut_ad(size_in_bytes > size_in_bytes_requested); n_blocks_to_withdraw= 0; first_to_withdraw= nullptr; size_in_bytes_requested= size_in_bytes;
while (buf_page_t *b= UT_LIST_GET_FIRST(withdrawn)) { UT_LIST_REMOVE(withdrawn, b); UT_LIST_ADD_LAST(free, b); ut_d(b->in_free_list= true); ut_ad(b->state() == buf_page_t::NOT_USED); b->lock.init(); }
mysql_mutex_unlock(&mutex); sql_print_information("InnoDB: Memory pressure event failed to shrink" " innodb_buffer_pool_size=%zum", old_size); ut_d(validate());}#endif
#if defined(DBUG_OFF) && defined(HAVE_MADVISE) && defined(MADV_DODUMP)
/** Enable buffers to be dumped to core files.
A convenience function, not called anyhwere directly howeverit is left available for gdb or any debugger to callin the event that you want all of the memory to be dumpedto a core file.
@return number of errors found in madvise() calls */MY_ATTRIBUTE((used))int buf_pool_t::madvise_do_dump() noexcept{ int ret= 0;
/* mirrors allocation in log_t::create() */ if (log_sys.buf) { ret += madvise(log_sys.buf, log_sys.buf_size, MADV_DODUMP); ret += madvise(log_sys.flush_buf, log_sys.buf_size, MADV_DODUMP); }
ret+= madvise(buf_pool.memory, buf_pool.size_in_bytes, MADV_DODUMP); return ret;}#endif
#ifndef UNIV_DEBUG
static inline byte hex_to_ascii(byte hex_digit) noexcept{ const int offset= hex_digit <= 9 ? '0' : 'a' - 10; return byte(hex_digit + offset);}#endif
/** Dump a page to stderr.
@param[in] read_buf database page@param[in] zip_size compressed page size, or 0 */ATTRIBUTE_COLDvoid buf_page_print(const byte *read_buf, ulint zip_size) noexcept{#ifndef UNIV_DEBUG
const size_t size = zip_size ? zip_size : srv_page_size; const byte * const end= read_buf + size; sql_print_information("InnoDB: Page dump (%zu bytes):", size);
do { byte row[64];
for (byte *r= row; r != &row[64]; r+= 2, read_buf++) { r[0]= hex_to_ascii(byte(*read_buf >> 4)); r[1]= hex_to_ascii(*read_buf & 15); }
sql_print_information("InnoDB: %.*s", 64, row); } while (read_buf != end);
sql_print_information("InnoDB: End of page dump");#endif
}
IF_DBUG(,inline) byte *buf_block_t::frame_address() const noexcept{ static_assert(ut_is_2pow(innodb_buffer_pool_extent_size), "");
byte *frame_= reinterpret_cast<byte*> ((reinterpret_cast<size_t>(this) & ~(innodb_buffer_pool_extent_size - 1)) | first_frame_in_extent[srv_page_size_shift - UNIV_PAGE_SIZE_SHIFT_MIN]); ut_ad(reinterpret_cast<const byte*>(this) + sizeof(*this) <= frame_); frame_+= (((reinterpret_cast<size_t>(this) & (innodb_buffer_pool_extent_size - 1)) / sizeof(*this)) << srv_page_size_shift); return frame_;}
buf_block_t *buf_pool_t::block_from(const void *ptr) noexcept{ static_assert(ut_is_2pow(innodb_buffer_pool_extent_size), ""); ut_ad(static_cast<const char*>(ptr) >= buf_pool.memory);
byte *first_block= reinterpret_cast<byte*> (reinterpret_cast<size_t>(ptr) & ~(innodb_buffer_pool_extent_size - 1)); const size_t first_frame= first_frame_in_extent[srv_page_size_shift - UNIV_PAGE_SIZE_SHIFT_MIN];
ut_ad(static_cast<const byte*>(ptr) >= first_block + first_frame); return reinterpret_cast<buf_block_t*>(first_block) + (((size_t(ptr) & (innodb_buffer_pool_extent_size - 1)) - first_frame) >> srv_page_size_shift);}
/** Determine the address of the first invalid block descriptor
@param n_blocks buf_pool.n_blocks@return offset of the first invalid buf_block_t, relative to buf_pool.memory */static size_t block_descriptors_in_bytes(size_t n_blocks) noexcept{ const size_t ssize= srv_page_size_shift - UNIV_PAGE_SIZE_SHIFT_MIN; const size_t extent_size= pages_in_extent[ssize]; return n_blocks / extent_size * innodb_buffer_pool_extent_size + (n_blocks % extent_size) * sizeof(buf_block_t);}
buf_block_t *buf_pool_t::get_nth_page(size_t pos) const noexcept{ mysql_mutex_assert_owner(&mutex); ut_ad(pos < n_blocks); return reinterpret_cast<buf_block_t*> (memory + block_descriptors_in_bytes(pos));}
buf_block_t *buf_pool_t::allocate() noexcept{ mysql_mutex_assert_owner(&mutex);
while (buf_page_t *b= UT_LIST_GET_FIRST(free)) { ut_ad(b->in_free_list); ut_d(b->in_free_list = FALSE); ut_ad(!b->oldest_modification()); ut_ad(!b->in_LRU_list); ut_a(!b->in_file()); UT_LIST_REMOVE(free, b);
if (UNIV_LIKELY(!n_blocks_to_withdraw) || !withdraw(*b)) { /* No adaptive hash index entries may point to a free block. */ assert_block_ahi_empty(reinterpret_cast<buf_block_t*>(b)); b->set_state(buf_page_t::MEMORY); b->set_os_used(); return reinterpret_cast<buf_block_t*>(b); } }
return nullptr;}
/** Create the hash table.
@param n the lower bound of n_cells */void buf_pool_t::page_hash_table::create(ulint n) noexcept{ n_cells= ut_find_prime(n); const size_t size= MY_ALIGN(pad(n_cells) * sizeof *array, CPU_LEVEL1_DCACHE_LINESIZE); void *v= aligned_malloc(size, CPU_LEVEL1_DCACHE_LINESIZE); memset_aligned<CPU_LEVEL1_DCACHE_LINESIZE>(v, 0, size); array= static_cast<hash_chain*>(v);}
size_t buf_pool_t::get_n_blocks(size_t size_in_bytes) noexcept{ const size_t ssize= srv_page_size_shift - UNIV_PAGE_SIZE_SHIFT_MIN; size_t n_blocks_alloc= size_in_bytes / innodb_buffer_pool_extent_size * pages_in_extent[ssize];
if (const size_t incomplete_extent_pages= (size_in_bytes & (innodb_buffer_pool_extent_size - 1)) >> srv_page_size_shift) { ssize_t d= incomplete_extent_pages - first_page_in_extent[ssize]; ut_ad(d > 0); n_blocks_alloc+= d; }
return n_blocks_alloc;}
size_t buf_pool_t::blocks_in_bytes(size_t n_blocks) noexcept{ const size_t shift{srv_page_size_shift}; const size_t ssize{shift - UNIV_PAGE_SIZE_SHIFT_MIN}; const size_t extent_size= pages_in_extent[ssize]; size_t size_in_bytes= n_blocks / extent_size * innodb_buffer_pool_extent_size; if (size_t remainder= n_blocks % extent_size) size_in_bytes+= (remainder + first_page_in_extent[ssize]) << shift; ut_ad(get_n_blocks(size_in_bytes) == n_blocks); return size_in_bytes;}
/** Create the buffer pool.
@return whether the creation failed */bool buf_pool_t::create() noexcept{ ut_ad(this == &buf_pool); ut_ad(!is_initialised()); ut_ad(size_in_bytes_requested > 0); ut_ad(!(size_in_bytes_max & (innodb_buffer_pool_extent_size - 1))); ut_ad(!(size_in_bytes_requested & ((1U << 20) - 1))); ut_ad(size_in_bytes_requested <= size_in_bytes_max); /* mariabackup loads tablespaces, and it requires field_ref_zero to be
allocated before innodb initialization */ ut_ad(srv_operation >= SRV_OPERATION_RESTORE || !field_ref_zero);
if (!field_ref_zero) { if (auto b= aligned_malloc(UNIV_PAGE_SIZE_MAX, 4096)) { field_ref_zero= static_cast<const byte*> (memset_aligned<4096>(b, 0, UNIV_PAGE_SIZE_MAX)); goto init; }
oom: ut_ad(!is_initialised()); sql_print_error("InnoDB: Cannot map innodb_buffer_pool_size_max=%zum", size_in_bytes_max >> 20); return true; }
init: DBUG_EXECUTE_IF("ib_buf_chunk_init_fails", goto oom;); size_t size= size_in_bytes_max; sql_print_information("InnoDB: innodb_buffer_pool_size_max=%zum," " innodb_buffer_pool_size=%zum", size >> 20, size_in_bytes_requested >> 20);
retry: { NUMA_MEMPOLICY_INTERLEAVE_IN_SCOPE;#ifdef _WIN32
memory_unaligned= my_virtual_mem_reserve(&size);#else
memory_unaligned= my_large_virtual_alloc(&size);#endif
}
if (!memory_unaligned) goto oom;
const size_t alignment_waste= ((~size_t(memory_unaligned) & (innodb_buffer_pool_extent_size - 1)) + 1) & (innodb_buffer_pool_extent_size - 1);
if (size < size_in_bytes_max + alignment_waste) { my_virtual_mem_release(memory_unaligned, size); size+= 1 + (~size_t(memory_unaligned) & (innodb_buffer_pool_extent_size - 1)); goto retry; }
MEM_UNDEFINED(memory_unaligned, size); ut_dontdump(memory_unaligned, size, true); memory= memory_unaligned + alignment_waste; size_unaligned= size; size-= alignment_waste; size&= ~(innodb_buffer_pool_extent_size - 1);
const size_t actual_size= size_in_bytes_requested; ut_ad(actual_size <= size);
size_in_bytes= actual_size; os_total_large_mem_allocated+= actual_size;
#ifdef UNIV_PFS_MEMORY
PSI_MEMORY_CALL(memory_alloc)(mem_key_buf_buf_pool, actual_size, &owner);#endif
#ifdef _WIN32
if (!my_virtual_mem_commit(memory, actual_size)) { my_virtual_mem_release(memory_unaligned, size_unaligned); memory= nullptr; memory_unaligned= nullptr; goto oom; }#else
update_malloc_size(actual_size, 0);#endif
#ifdef HAVE_LIBNUMA
if (srv_numa_interleave) { struct bitmask *numa_mems_allowed= numa_get_mems_allowed(); MEM_MAKE_DEFINED(numa_mems_allowed, sizeof *numa_mems_allowed); if (mbind(memory_unaligned, size_unaligned, MPOL_INTERLEAVE, numa_mems_allowed->maskp, numa_mems_allowed->size, MPOL_MF_MOVE)) sql_print_warning("InnoDB: Failed to set NUMA memory policy of" " buffer pool page frames to MPOL_INTERLEAVE" " (error: %s).", strerror(errno)); numa_bitmask_free(numa_mems_allowed); }#endif /* HAVE_LIBNUMA */
n_blocks= get_n_blocks(actual_size); n_blocks_to_withdraw= 0; UT_LIST_INIT(free, &buf_page_t::list); const size_t ssize= srv_page_size_shift - UNIV_PAGE_SIZE_SHIFT_MIN;
for (char *extent= memory, *end= memory + block_descriptors_in_bytes(n_blocks); extent < end; extent+= innodb_buffer_pool_extent_size) { buf_block_t *block= reinterpret_cast<buf_block_t*>(extent); const buf_block_t *extent_end= block + pages_in_extent[ssize]; if (reinterpret_cast<const char*>(extent_end) > end) extent_end= reinterpret_cast<buf_block_t*>(end); MEM_MAKE_DEFINED(block, (extent_end - block) * sizeof *block); for (byte *frame= reinterpret_cast<byte*>(extent) + first_frame_in_extent[ssize]; block < extent_end; block++, frame+= srv_page_size) { ut_ad(!memcmp(block, field_ref_zero, sizeof *block)); block->page.frame= frame; block->page.lock.init(); UT_LIST_ADD_LAST(free, &block->page); ut_d(block->page.in_free_list= true); } }
#if defined(__aarch64__)
mysql_mutex_init(buf_pool_mutex_key, &mutex, MY_MUTEX_INIT_FAST);#else
mysql_mutex_init(buf_pool_mutex_key, &mutex, nullptr);#endif
UT_LIST_INIT(withdrawn, &buf_page_t::list); UT_LIST_INIT(LRU, &buf_page_t::LRU); UT_LIST_INIT(flush_list, &buf_page_t::list); UT_LIST_INIT(unzip_LRU, &buf_block_t::unzip_LRU);
for (size_t i= 0; i < UT_ARR_SIZE(zip_free); ++i) UT_LIST_INIT(zip_free[i], &buf_buddy_free_t::list); ulint s= n_blocks; s/= BUF_READ_AHEAD_PORTION; read_ahead_area= s >= READ_AHEAD_PAGES ? READ_AHEAD_PAGES : my_round_up_to_next_power(static_cast<uint32_t>(s));
page_hash.create(2 * n_blocks); last_printout_time= time(nullptr);
mysql_mutex_init(flush_list_mutex_key, &flush_list_mutex, MY_MUTEX_INIT_FAST);
pthread_cond_init(&done_flush_LRU, nullptr); pthread_cond_init(&done_flush_list, nullptr); pthread_cond_init(&do_flush_list, nullptr); pthread_cond_init(&done_free, nullptr);
try_LRU_scan= true;
ut_d(flush_hp.m_mutex= &flush_list_mutex;); ut_d(lru_hp.m_mutex= &mutex); ut_d(lru_scan_itr.m_mutex= &mutex);
io_buf.create((srv_n_read_io_threads + srv_n_write_io_threads) * OS_AIO_N_PENDING_IOS_PER_THREAD);
last_activity_count= srv_get_activity_count();
buf_LRU_old_ratio_update(100 * 3 / 8, false); btr_search_sys_create();
#ifdef __linux__
if (srv_operation == SRV_OPERATION_NORMAL) buf_mem_pressure_detect_init();#endif
ut_ad(is_initialised()); sql_print_information("InnoDB: Completed initialization of buffer pool"); return false;}
/** Clean up after successful create() */void buf_pool_t::close() noexcept{ ut_ad(this == &buf_pool); if (!is_initialised()) return;
mysql_mutex_destroy(&mutex); mysql_mutex_destroy(&flush_list_mutex);
for (buf_page_t *bpage= UT_LIST_GET_LAST(LRU), *prev_bpage= nullptr; bpage; bpage= prev_bpage) { prev_bpage= UT_LIST_GET_PREV(LRU, bpage); ut_ad(bpage->in_file()); ut_ad(bpage->in_LRU_list); /* The buffer pool must be clean during normal shutdown.
Only on aborted startup (with recovery) or with innodb_fast_shutdown=2 we may discard changes. */ ut_d(const lsn_t oldest= bpage->oldest_modification();) ut_ad(fsp_is_system_temporary(bpage->id().space()) ? (oldest == 0 || oldest == 2) : oldest <= 1 || srv_is_being_started || srv_fast_shutdown == 2);
if (UNIV_UNLIKELY(!bpage->frame)) { bpage->lock.free(); ut_free(bpage); } }
{ const size_t size{size_in_bytes};
for (char *extent= memory, *end= memory + block_descriptors_in_bytes(n_blocks); extent < end; extent+= innodb_buffer_pool_extent_size) for (buf_block_t *block= reinterpret_cast<buf_block_t*>(extent), *extent_end= block + pages_in_extent[srv_page_size_shift - UNIV_PAGE_SIZE_SHIFT_MIN]; block < extent_end && reinterpret_cast<char*>(block) < end; block++) { MEM_MAKE_DEFINED(&block->page.lock, sizeof &block->page.lock); block->page.lock.free(); }
ut_dodump(memory_unaligned, size_unaligned);#ifdef UNIV_PFS_MEMORY
PSI_MEMORY_CALL(memory_free)(mem_key_buf_buf_pool, size, owner); owner= nullptr;#endif
os_total_large_mem_allocated-= size; my_virtual_mem_decommit(memory, size); my_virtual_mem_release(memory_unaligned, size_unaligned); memory= nullptr; memory_unaligned= nullptr; }
pthread_cond_destroy(&done_flush_LRU); pthread_cond_destroy(&done_flush_list); pthread_cond_destroy(&do_flush_list); pthread_cond_destroy(&done_free);
page_hash.free();
io_buf.close(); aligned_free(const_cast<byte*>(field_ref_zero)); field_ref_zero= nullptr;}
void buf_pool_t::io_buf_t::create(ulint n_slots) noexcept{ this->n_slots= n_slots; slots= static_cast<buf_tmp_buffer_t*> (ut_malloc_nokey(n_slots * sizeof *slots)); memset((void*) slots, 0, n_slots * sizeof *slots);}
void buf_pool_t::io_buf_t::close() noexcept{ for (buf_tmp_buffer_t *s= slots, *e= slots + n_slots; s != e; s++) { aligned_free(s->crypt_buf); aligned_free(s->comp_buf); } ut_free(slots); slots= nullptr; n_slots= 0;}
buf_tmp_buffer_t *buf_pool_t::io_buf_t::reserve(bool wait_for_reads) noexcept{ for (;;) { for (buf_tmp_buffer_t *s= slots, *e= slots + n_slots; s != e; s++) if (s->acquire()) return s; buf_dblwr.flush_buffered_writes(); os_aio_wait_until_no_pending_writes(true); if (!wait_for_reads) continue; for (buf_tmp_buffer_t *s= slots, *e= slots + n_slots; s != e; s++) if (s->acquire()) return s; os_aio_wait_until_no_pending_reads(true); }}
ATTRIBUTE_COLD bool buf_pool_t::withdraw(buf_page_t &bpage) noexcept{ mysql_mutex_assert_owner(&mutex); ut_ad(n_blocks_to_withdraw); ut_ad(first_to_withdraw); ut_ad(!bpage.zip.data); if (&bpage < first_to_withdraw) return false; n_blocks_to_withdraw--; bpage.lock.free(); UT_LIST_ADD_LAST(withdrawn, &bpage); return true;}
ATTRIBUTE_COLD buf_pool_t::shrink_status buf_pool_t::shrink(size_t size) noexcept{ mysql_mutex_assert_owner(&mutex); buf_load_abort();
if (!n_blocks_to_withdraw) { withdraw_done: first_to_withdraw= nullptr; while (buf_page_t *b= UT_LIST_GET_FIRST(withdrawn)) { UT_LIST_REMOVE(withdrawn, b); /* satisfy the check in lazy_allocate() */ ut_d(memset((void*) b, 0, sizeof(buf_block_t))); } return SHRINK_DONE; }
buf_buddy_condense_free(size);
for (buf_page_t *b= UT_LIST_GET_FIRST(free), *next; b; b= next) { ut_ad(b->in_free_list); ut_ad(!b->in_LRU_list); ut_ad(!b->zip.data); ut_ad(!b->oldest_modification()); ut_a(b->state() == buf_page_t::NOT_USED);
next= UT_LIST_GET_NEXT(list, b);
if (b >= first_to_withdraw) { UT_LIST_REMOVE(free, b); b->lock.free(); UT_LIST_ADD_LAST(withdrawn, b); if (!--n_blocks_to_withdraw) goto withdraw_done; } }
buf_block_t *block= allocate(); size_t scanned= 0; for (buf_page_t *b= lru_scan_itr.start(), *prev; block && b; b= prev) { ut_ad(b->in_LRU_list); ut_a(b->in_file());
prev= UT_LIST_GET_PREV(LRU, b);
if (!b->can_relocate()) { next: if (++scanned & 31) continue; /* Avoid starvation by periodically releasing buf_pool.mutex. */ lru_scan_itr.set(prev); mysql_mutex_unlock(&mutex); mysql_mutex_lock(&mutex); prev= lru_scan_itr.get(); continue; }
const page_id_t id{b->id()}; hash_chain &chain= page_hash.cell_get(id.fold()); page_hash_latch &hash_lock= page_hash.lock_get(chain); hash_lock.lock();
{ /* relocate flush_list and b->page.zip */ bool have_flush_list_mutex= false;
switch (b->oldest_modification()) { case 2: ut_ad(fsp_is_system_temporary(id.space())); /* fall through */ case 0: break; default: mysql_mutex_lock(&flush_list_mutex); switch (ut_d(lsn_t om=) b->oldest_modification()) { case 1: delete_from_flush_list(b); /* fall through */ case 0: mysql_mutex_unlock(&flush_list_mutex); break; default: ut_ad(om != 2); have_flush_list_mutex= true; } }
if (!b->can_relocate()) { next_quick: if (have_flush_list_mutex) mysql_mutex_unlock(&flush_list_mutex); hash_lock.unlock(); continue; }
if (UNIV_UNLIKELY(will_be_withdrawn(b->zip.data, size))) { block= buf_buddy_shrink(b, block); ut_ad(mach_read_from_4(b->zip.data + FIL_PAGE_OFFSET) == id.page_no()); if (UNIV_UNLIKELY(!n_blocks_to_withdraw)) { if (have_flush_list_mutex) mysql_mutex_unlock(&flush_list_mutex); hash_lock.unlock(); if (block) buf_LRU_block_free_non_file_page(block); goto withdraw_done; } if (!block && !(block= allocate())) goto next_quick; }
if (!b->frame || b < first_to_withdraw) goto next_quick;
ut_ad(is_uncompressed_current(b));
byte *const frame= block->page.frame; memcpy_aligned<4096>(frame, b->frame, srv_page_size); b->lock.free(); block->page.lock.free(); new(&block->page) buf_page_t(*b); block->page.frame= frame;
if (have_flush_list_mutex) { buf_flush_relocate_on_flush_list(b, &block->page); mysql_mutex_unlock(&flush_list_mutex); } }
/* relocate LRU list */ if (buf_page_t *prev_b= LRU_remove(b)) UT_LIST_INSERT_AFTER(LRU, prev_b, &block->page); else UT_LIST_ADD_FIRST(LRU, &block->page);
if (LRU_old == b) LRU_old= &block->page;
ut_ad(block->page.in_LRU_list);
/* relocate page_hash */ ut_ad(b == page_hash.get(id, chain)); page_hash.replace(chain, b, &block->page);
if (b->zip.data) { ut_ad(mach_read_from_4(b->zip.data + FIL_PAGE_OFFSET) == id.page_no()); b->zip.data= nullptr; /* relocate unzip_LRU list */ buf_block_t *old_block= reinterpret_cast<buf_block_t*>(b); ut_ad(old_block->in_unzip_LRU_list); ut_d(old_block->in_unzip_LRU_list= false); ut_d(block->in_unzip_LRU_list= true);
buf_block_t *prev= UT_LIST_GET_PREV(unzip_LRU, old_block); UT_LIST_REMOVE(unzip_LRU, old_block);
if (prev) UT_LIST_INSERT_AFTER(unzip_LRU, prev, block); else UT_LIST_ADD_FIRST(unzip_LRU, block); }
buf_block_modify_clock_inc(block);
#ifdef BTR_CUR_HASH_ADAPT
assert_block_ahi_empty_on_init(block); block->index= nullptr; block->n_hash_helps= 0; block->n_fields= 1; block->left_side= true;#endif /* BTR_CUR_HASH_ADAPT */
hash_lock.unlock();
ut_d(b->in_LRU_list= false);
b->set_state(buf_page_t::NOT_USED); UT_LIST_ADD_LAST(withdrawn, b); if (!--n_blocks_to_withdraw) goto withdraw_done;
block= allocate(); goto next; }
mysql_mutex_lock(&flush_list_mutex);
if (LRU_warned && !UT_LIST_GET_FIRST(free)) { LRU_warned_clear(); mysql_mutex_unlock(&flush_list_mutex); return SHRINK_ABORT; }
try_LRU_scan= false; mysql_mutex_unlock(&mutex); page_cleaner_wakeup(true); my_cond_wait(&done_flush_list, &flush_list_mutex.m_mutex); mysql_mutex_unlock(&flush_list_mutex); mysql_mutex_lock(&mutex);
if (!n_blocks_to_withdraw) goto withdraw_done;
return SHRINK_IN_PROGRESS;}
inline void buf_pool_t::shrunk(size_t size, size_t reduced) noexcept{ ut_ad(size + reduced == size_in_bytes); size_in_bytes_requested= size; size_in_bytes= size;# ifndef HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT
/* Only page_guess() may read this memory, which after
my_virtual_mem_decommit() may be zeroed out or preserve its original contents. Try to catch any unintended reads outside page_guess(). */ MEM_UNDEFINED(memory + size, size_in_bytes_max - size);# else
for (size_t n= page_hash.pad(page_hash.n_cells), i= 0; i < n; i+= page_hash.ELEMENTS_PER_LATCH + 1) { auto &latch= reinterpret_cast<page_hash_latch&>(page_hash.array[i]); latch.lock(); /* We already shrunk size_in_bytes. The exclusive lock here
ensures that any page_guess() will detect an out-of-bounds guess before we invoke my_virtual_mem_decommit() below. */ latch.unlock(); }# endif
my_virtual_mem_decommit(memory + size, reduced);#ifdef UNIV_PFS_MEMORY
PSI_MEMORY_CALL(memory_free)(mem_key_buf_buf_pool, reduced, owner);#endif
}
ATTRIBUTE_COLD void buf_pool_t::resize(size_t size, THD *thd) noexcept{ ut_ad(this == &buf_pool); mysql_mutex_assert_owner(&LOCK_global_system_variables); ut_ad(size <= size_in_bytes_max); if (my_use_large_pages) { my_error(ER_VARIABLE_IS_READONLY, MYF(0), "InnoDB", "innodb_buffer_pool_size", "large_pages=0"); return; }
size_t n_blocks_new= get_n_blocks(size);
mysql_mutex_lock(&mutex);
const size_t old_size= size_in_bytes; if (first_to_withdraw || old_size != size_in_bytes_requested) { mysql_mutex_unlock(&mutex); my_printf_error(ER_WRONG_USAGE, "innodb_buffer_pool_size change is already in progress", MYF(0)); return; }
ut_ad(UT_LIST_GET_LEN(withdrawn) == 0); ut_ad(n_blocks_to_withdraw == 0);#ifdef __linux__
DBUG_EXECUTE_IF("trigger_garbage_collection", mem_pressure_obj.trigger_collection(););#endif
if (size == old_size) { mysql_mutex_unlock(&mutex); DBUG_EXECUTE_IF("trigger_garbage_collection", std::this_thread::sleep_for(std::chrono::milliseconds(50)); garbage_collect();); return; }
#ifdef BTR_CUR_HASH_ADAPT
bool ahi_disabled= false;#endif
const bool significant_change= n_blocks_new > n_blocks * 2 || n_blocks > n_blocks_new * 2; const ssize_t n_blocks_removed= n_blocks - n_blocks_new;
if (n_blocks_removed <= 0) { if (!my_virtual_mem_commit(memory + old_size, size - old_size)) { mysql_mutex_unlock(&mutex); sql_print_error("InnoDB: Cannot commit innodb_buffer_pool_size=%zum;" " retaining innodb_buffer_pool_size=%zum", size >> 20, old_size >> 20); my_error(ER_OUT_OF_RESOURCES, MYF(0)); return; }
size_in_bytes_requested= size; size_in_bytes= size;
{ const size_t ssize= srv_page_size_shift - UNIV_PAGE_SIZE_SHIFT_MIN; const size_t pages= pages_in_extent[ssize]; const size_t first_extent= n_blocks / pages;
char *extent= memory + first_extent * innodb_buffer_pool_extent_size;
buf_block_t *block= reinterpret_cast<buf_block_t*>(extent); if (const size_t first_blocks= n_blocks % pages) { /* Extend the last (partial) extent until its end */ const buf_block_t *extent_end= block + (first_extent == (n_blocks_new / pages) ? (n_blocks_new % pages) : pages); block+= first_blocks; memset((void*) block, 0, (extent_end - block) * sizeof *block);
for (byte *frame= reinterpret_cast<byte*>(extent) + first_frame_in_extent[ssize] + (first_blocks << srv_page_size_shift); block < extent_end; block++, frame+= srv_page_size) { block->page.frame= frame; block->page.lock.init(); UT_LIST_ADD_LAST(free, &block->page); ut_d(block->page.in_free_list= true); } extent+= innodb_buffer_pool_extent_size; }
/* Fill in further extents; @see buf_pool_t::create() */ for (const char *const end_new= memory + block_descriptors_in_bytes(n_blocks_new); extent < end_new; extent+= innodb_buffer_pool_extent_size) { block= reinterpret_cast<buf_block_t*>(extent); const buf_block_t *extent_end= block + pages; if (reinterpret_cast<const char*>(extent_end) > end_new) extent_end= reinterpret_cast<const buf_block_t*>(end_new);
memset((void*) block, 0, (extent_end - block) * sizeof *block); for (byte *frame= reinterpret_cast<byte*>(extent) + first_frame_in_extent[ssize]; block < extent_end; block++, frame+= srv_page_size) { block->page.frame= frame; block->page.lock.init(); UT_LIST_ADD_LAST(free, &block->page); ut_d(block->page.in_free_list= true); } } }
mysql_mutex_unlock(&LOCK_global_system_variables); resized: ut_ad(UT_LIST_GET_LEN(withdrawn) == 0); ut_ad(n_blocks_to_withdraw == 0); ut_ad(!first_to_withdraw); const size_t old_blocks{n_blocks}; n_blocks= n_blocks_new;
size_t s= n_blocks_new / BUF_READ_AHEAD_PORTION; read_ahead_area= s >= READ_AHEAD_PAGES ? READ_AHEAD_PAGES : my_round_up_to_next_power(uint32(s));
if (ssize_t d= size - old_size) { os_total_large_mem_allocated+= d; if (d > 0) { /* Already committed memory earlier */ ut_ad(n_blocks_removed <= 0);#ifdef UNIV_PFS_MEMORY
PSI_MEMORY_CALL(memory_alloc)(mem_key_buf_buf_pool, d, &owner);#endif
} else shrunk(size, size_t(-d)); }
mysql_mutex_unlock(&mutex);
if (significant_change) { sql_print_information("InnoDB: Resizing hash tables"); srv_lock_table_size= 5 * n_blocks_new; lock_sys.resize(srv_lock_table_size); dict_sys.resize(); }
ibuf_max_size_update(srv_change_buffer_max_size);#ifdef BTR_CUR_HASH_ADAPT
if (ahi_disabled) btr_search_enable(true);#endif
mysql_mutex_lock(&LOCK_global_system_variables); bool resized= n_blocks_removed < 0; if (n_blocks_removed > 0) { mysql_mutex_lock(&mutex); resized= size_in_bytes == old_size; if (resized) { size_in_bytes_requested= size; size_in_bytes= size; } mysql_mutex_unlock(&mutex); }
if (resized) sql_print_information("InnoDB: innodb_buffer_pool_size=%zum (%zu pages)" " resized from %zum (%zu pages)", size >> 20, n_blocks_new, old_size >> 20, old_blocks); } else { size_t to_withdraw= size_t(n_blocks_removed); n_blocks_to_withdraw= to_withdraw; first_to_withdraw= &get_nth_page(n_blocks_new)->page; size_in_bytes_requested= size; mysql_mutex_unlock(&LOCK_global_system_variables);
mysql_mutex_unlock(&mutex); DEBUG_SYNC_C("buf_pool_shrink_before_wakeup"); mysql_mutex_lock(&flush_list_mutex); page_cleaner_wakeup(true); my_cond_wait(&done_flush_list, &flush_list_mutex.m_mutex); mysql_mutex_unlock(&flush_list_mutex);#ifdef BTR_CUR_HASH_ADAPT
ahi_disabled= btr_search_disable();#endif /* BTR_CUR_HASH_ADAPT */
mysql_mutex_lock(&mutex);
time_t last_message= 0;
do { time_t now= time(nullptr); if (now - last_message > 15) { if (last_message != 0 && to_withdraw == n_blocks_to_withdraw) break; to_withdraw= n_blocks_to_withdraw; last_message= now; sql_print_information("InnoDB: Trying to shrink" " innodb_buffer_pool_size=%zum (%zu pages)" " from %zum (%zu pages, to withdraw %zu)", size >> 20, n_blocks_new, old_size >> 20, n_blocks, to_withdraw); } shrink_status s{shrink(size)}; if (s == SHRINK_DONE) goto resized; if (s != SHRINK_IN_PROGRESS) break; } while (!thd_kill_level(thd));
ut_ad(size_in_bytes > size_in_bytes_requested); n_blocks_to_withdraw= 0; first_to_withdraw= nullptr; size_in_bytes_requested= size_in_bytes;
while (buf_page_t *b= UT_LIST_GET_FIRST(withdrawn)) { UT_LIST_REMOVE(withdrawn, b); UT_LIST_ADD_LAST(free, b); ut_d(b->in_free_list= true); ut_ad(b->state() == buf_page_t::NOT_USED); b->lock.init(); }
mysql_mutex_unlock(&mutex); my_printf_error(ER_WRONG_USAGE, "innodb_buffer_pool_size change aborted", MYF(ME_ERROR_LOG)); mysql_mutex_lock(&LOCK_global_system_variables); }
ut_d(validate());}
/** Relocate a ROW_FORMAT=COMPRESSED block in the LRU list and
buf_pool.page_hash.The caller must relocate bpage->list.@param bpage ROW_FORMAT=COMPRESSED only block@param dpage destination control block */static void buf_relocate(buf_page_t *bpage, buf_page_t *dpage) noexcept{ const page_id_t id{bpage->id()}; buf_pool_t::hash_chain &chain= buf_pool.page_hash.cell_get(id.fold()); ut_ad(!bpage->frame); mysql_mutex_assert_owner(&buf_pool.mutex); ut_ad(mach_read_from_4(bpage->zip.data + FIL_PAGE_OFFSET) == id.page_no()); ut_ad(buf_pool.page_hash.lock_get(chain).is_write_locked()); ut_ad(bpage == buf_pool.page_hash.get(id, chain)); ut_ad(!buf_pool.watch_is_sentinel(*bpage)); ut_d(const auto state= bpage->state()); ut_ad(state >= buf_page_t::FREED); ut_ad(state <= buf_page_t::READ_FIX); ut_ad(bpage->lock.is_write_locked()); const auto frame= dpage->frame; ut_ad(frame == reinterpret_cast<buf_block_t*>(dpage)->frame_address());
dpage->lock.free(); new (dpage) buf_page_t(*bpage);
dpage->frame= frame;
/* Important that we adjust the hazard pointer before
removing bpage from LRU list. */ if (buf_page_t *b= buf_pool.LRU_remove(bpage)) UT_LIST_INSERT_AFTER(buf_pool.LRU, b, dpage); else UT_LIST_ADD_FIRST(buf_pool.LRU, dpage);
if (UNIV_UNLIKELY(buf_pool.LRU_old == bpage)) { buf_pool.LRU_old= dpage;#ifdef UNIV_LRU_DEBUG
/* buf_pool.LRU_old must be the first item in the LRU list
whose "old" flag is set. */ ut_a(buf_pool.LRU_old->old); ut_a(!UT_LIST_GET_PREV(LRU, buf_pool.LRU_old) || !UT_LIST_GET_PREV(LRU, buf_pool.LRU_old)->old); ut_a(!UT_LIST_GET_NEXT(LRU, buf_pool.LRU_old) || UT_LIST_GET_NEXT(LRU, buf_pool.LRU_old)->old); } else { /* Check that the "old" flag is consistent in
the block and its neighbours. */ dpage->set_old(dpage->is_old());#endif /* UNIV_LRU_DEBUG */
}
ut_d(CheckInLRUList::validate());
buf_pool.page_hash.replace(chain, bpage, dpage);}
buf_page_t *buf_pool_t::watch_set(const page_id_t id, buf_pool_t::hash_chain &chain) noexcept{ ut_ad(&chain == &page_hash.cell_get(id.fold())); page_hash.lock_get(chain).lock();
buf_page_t *bpage= page_hash.get(id, chain);
if (bpage) {got_block: bpage->fix(); if (watch_is_sentinel(*bpage)) { ut_ad(!bpage->oldest_modification()); bpage= nullptr; } page_hash.lock_get(chain).unlock(); return bpage; }
page_hash.lock_get(chain).unlock(); /* Allocate a watch[] and then try to insert it into the page_hash. */ mysql_mutex_lock(&mutex);
/* The maximum number of purge tasks should never exceed
the UT_ARR_SIZE(watch) - 1, and there is no way for a purge task to hold a watch when setting another watch. */ for (buf_page_t *w= &watch[UT_ARR_SIZE(watch)]; w-- >= watch; ) { ut_ad(w->access_time == 0); ut_ad(!w->oldest_modification()); ut_ad(!w->zip.data); static_assert(buf_page_t::NOT_USED == 0, "efficiency"); if (ut_d(auto s=) w->state()) { /* This watch may be in use for some other page. */ ut_ad(s >= buf_page_t::UNFIXED); continue; } /* w is pointing to watch[], which is protected by mutex.
Normally, buf_page_t::id for objects that are reachable by page_hash.get(id, chain) are protected by hash_lock. */ w->set_state(buf_page_t::UNFIXED + 1); w->id_= id;
page_hash.lock_get(chain).lock(); bpage= page_hash.get(id, chain); if (UNIV_LIKELY_NULL(bpage)) { w->set_state(buf_page_t::NOT_USED); mysql_mutex_unlock(&mutex); goto got_block; }
ut_ad(w->state() == buf_page_t::UNFIXED + 1); buf_pool.page_hash.append(chain, w); mysql_mutex_unlock(&mutex); page_hash.lock_get(chain).unlock(); return nullptr; }
ut_error;}
/** Stop watching whether a page has been read in.
watch_set(id) must have returned nullptr before.@param id page identifier@param chain unlocked hash table chain */TRANSACTIONAL_TARGETvoid buf_pool_t::watch_unset(const page_id_t id, buf_pool_t::hash_chain &chain) noexcept{ mysql_mutex_assert_not_owner(&mutex); buf_page_t *w; { transactional_lock_guard<page_hash_latch> g{page_hash.lock_get(chain)}; /* The page must exist because watch_set() did fix(). */ w= page_hash.get(id, chain); ut_ad(w->in_page_hash); if (!watch_is_sentinel(*w)) { no_watch: w->unfix(); w= nullptr; } else { ut_ad(!w->oldest_modification()); const auto state= w->state(); ut_ad(~buf_page_t::LRU_MASK & state); ut_ad(state >= buf_page_t::UNFIXED + 1); if (state != buf_page_t::UNFIXED + 1) goto no_watch; } }
if (!w) return;
const auto old= w; /* The following is based on buf_pool_t::watch_remove(). */ mysql_mutex_lock(&mutex); w= page_hash.get(id, chain);
{ transactional_lock_guard<page_hash_latch> g {buf_pool.page_hash.lock_get(chain)}; auto f= w->unfix(); ut_ad(f < buf_page_t::READ_FIX || w != old);
if (f == buf_page_t::UNFIXED && w == old) { page_hash.remove(chain, w); // Now that w is detached from page_hash, release it to watch[].
ut_ad(w->id_ == id); ut_ad(!w->frame); ut_ad(!w->zip.data); w->set_state(buf_page_t::NOT_USED); } }
mysql_mutex_unlock(&mutex);}
/** Mark the page status as FREED for the given tablespace and page number.
@param[in,out] space tablespace@param[in] page page number@param[in,out] mtr mini-transaction */TRANSACTIONAL_TARGETvoid buf_page_free(fil_space_t *space, uint32_t page, mtr_t *mtr){ ut_ad(mtr); ut_ad(mtr->is_active());
if (srv_immediate_scrub_data_uncompressed#if defined HAVE_FALLOC_PUNCH_HOLE_AND_KEEP_SIZE || defined _WIN32
|| space->is_compressed()#endif
) mtr->add_freed_offset(space, page);
++buf_pool.stat.n_page_gets; const page_id_t page_id(space->id, page); buf_pool_t::hash_chain &chain= buf_pool.page_hash.cell_get(page_id.fold()); uint32_t fix; buf_block_t *block; { transactional_shared_lock_guard<page_hash_latch> g {buf_pool.page_hash.lock_get(chain)}; block= reinterpret_cast<buf_block_t*> (buf_pool.page_hash.get(page_id, chain)); if (!block || !block->page.frame) /* FIXME: convert ROW_FORMAT=COMPRESSED, without buf_zip_decompress() */ return; /* To avoid a deadlock with buf_LRU_free_page() of some other page
and buf_page_write_complete() of this page, we must not wait for a page latch while holding a page_hash latch. */ fix= block->page.fix(); }
if (UNIV_UNLIKELY(fix < buf_page_t::UNFIXED)) { block->page.unfix(); return; }
block->page.lock.x_lock(); if (block->page.is_ibuf_exist()) ibuf_merge_or_delete_for_page(nullptr, page_id, block->page.zip_size());#ifdef BTR_CUR_HASH_ADAPT
if (block->index) btr_search_drop_page_hash_index(block, false);#endif /* BTR_CUR_HASH_ADAPT */
block->page.set_freed(block->page.state()); mtr->memo_push(block, MTR_MEMO_PAGE_X_MODIFY);}
static void buf_inc_get(ha_handler_stats *stats){ mariadb_increment_pages_accessed(stats); ++buf_pool.stat.n_page_gets;}
TRANSACTIONAL_TARGETbuf_page_t *buf_page_get_zip(const page_id_t page_id) noexcept{ ha_handler_stats *const stats= mariadb_stats; buf_inc_get(stats);
buf_pool_t::hash_chain &chain= buf_pool.page_hash.cell_get(page_id.fold()); page_hash_latch &hash_lock= buf_pool.page_hash.lock_get(chain); buf_page_t *bpage;
for (;;) {#ifndef NO_ELISION
if (xbegin()) { if (hash_lock.is_locked()) xend(); else { bpage= buf_pool.page_hash.get(page_id, chain); const bool got_s_latch= bpage && !buf_pool.watch_is_sentinel(*bpage) && bpage->lock.s_lock_try(); xend(); if (got_s_latch) break; } }#endif
hash_lock.lock_shared(); bpage= buf_pool.page_hash.get(page_id, chain); if (!bpage || buf_pool.watch_is_sentinel(*bpage)) { hash_lock.unlock_shared(); switch (dberr_t err= buf_read_page(page_id, false)) { case DB_SUCCESS: case DB_SUCCESS_LOCKED_REC: mariadb_increment_pages_read(stats); continue; case DB_TABLESPACE_DELETED: return nullptr; default: sql_print_error("InnoDB: Reading compressed page " "[page id: space=" UINT32PF ", page number=" UINT32PF "] failed with error: %s", page_id.space(), page_id.page_no(), ut_strerr(err)); return nullptr; } }
ut_ad(bpage->in_file()); ut_ad(page_id == bpage->id());
const bool got_s_latch= bpage->lock.s_lock_try(); hash_lock.unlock_shared(); if (UNIV_LIKELY(got_s_latch)) break; /* We may fail to acquire bpage->lock because a read is holding an
exclusive latch on this block and either in progress or invoking buf_pool_t::corrupted_evict().
Let us aqcuire and release buf_pool.mutex to ensure that any buf_pool_t::corrupted_evict() will proceed before we reacquire the hash_lock that it could be waiting for.
While we are at it, let us also try to discard any uncompressed page frame of the compressed BLOB page, in case one had been allocated for writing the BLOB. */ mysql_mutex_lock(&buf_pool.mutex); bpage= buf_pool.page_hash.get(page_id, chain); if (bpage) buf_LRU_free_page(bpage, false); mysql_mutex_unlock(&buf_pool.mutex); }
if (UNIV_UNLIKELY(!bpage->zip.data)) { ut_ad("no ROW_FORMAT=COMPRESSED page!" == 0); bpage->lock.s_unlock(); bpage= nullptr; } else buf_page_make_young_if_needed(bpage);
#ifdef UNIV_DEBUG
if (!(++buf_dbg_counter % 5771)) buf_pool.validate();#endif /* UNIV_DEBUG */
return bpage;}
/********************************************************************//**
Initialize some fields of a control block. */UNIV_INLINEvoidbuf_block_init_low(/*===============*/ buf_block_t* block) /*!< in: block to init */{#ifdef BTR_CUR_HASH_ADAPT
/* No adaptive hash index entries may point to a previously
unused (and now freshly allocated) block. */ assert_block_ahi_empty_on_init(block); block->index = NULL;
block->n_hash_helps = 0; block->n_fields = 1; block->n_bytes = 0; block->left_side = TRUE;#endif /* BTR_CUR_HASH_ADAPT */
}
bool buf_zip_decompress(buf_block_t *block, bool check) noexcept{ const byte* frame = block->page.zip.data; ulint size = page_zip_get_size(&block->page.zip); /* The tablespace will not be found if this function is called
during IMPORT. */ fil_space_t* space= fil_space_t::get(block->page.id().space()); const unsigned key_version = mach_read_from_4( frame + FIL_PAGE_FILE_FLUSH_LSN_OR_KEY_VERSION); fil_space_crypt_t* crypt_data = space ? space->crypt_data : NULL; const bool encrypted = crypt_data && crypt_data->type != CRYPT_SCHEME_UNENCRYPTED && (!crypt_data->is_default_encryption() || srv_encrypt_tables);
ut_ad(block->zip_size()); ut_a(block->page.id().space() != 0); ut_ad(mach_read_from_4(frame + FIL_PAGE_OFFSET) == block->page.id().page_no());
if (UNIV_UNLIKELY(check && !page_zip_verify_checksum(frame, size))) {
ib::error() << "Compressed page checksum mismatch for " << (space ? space->chain.start->name : "") << block->page.id() << ": stored: " << mach_read_from_4(frame + FIL_PAGE_SPACE_OR_CHKSUM) << ", crc32: " << page_zip_calc_checksum(frame, size, false) << " adler32: " << page_zip_calc_checksum(frame, size, true); goto err_exit; }
switch (fil_page_get_type(frame)) { case FIL_PAGE_INDEX: case FIL_PAGE_RTREE: if (page_zip_decompress(&block->page.zip, block->page.frame, TRUE)) {func_exit: if (space) { space->release(); } return true; }
ib::error() << "Unable to decompress " << (space ? space->chain.start->name : "") << block->page.id(); goto err_exit; case FIL_PAGE_TYPE_ALLOCATED: case FIL_PAGE_INODE: case FIL_PAGE_IBUF_BITMAP: case FIL_PAGE_TYPE_FSP_HDR: case FIL_PAGE_TYPE_XDES: case FIL_PAGE_TYPE_ZBLOB: case FIL_PAGE_TYPE_ZBLOB2: /* Copy to uncompressed storage. */ memcpy(block->page.frame, frame, block->zip_size()); goto func_exit; }
ib::error() << "Unknown compressed page type " << fil_page_get_type(frame) << " in " << (space ? space->chain.start->name : "") << block->page.id();
err_exit: if (encrypted) { ib::info() << "Row compressed page could be encrypted" " with key_version " << key_version; }
if (space) { space->release(); }
return false;}
ATTRIBUTE_COLD/** Try to merge buffered changes to a buffer pool page.
@param block buffer-fixed and latched block@param rw_latch RW_X_LATCH, RW_SX_LATCH, RW_S_LATCH held on block@param err error code@return whether the page is invalid (corrupted) */static bool buf_page_ibuf_merge_try(buf_block_t *block, ulint rw_latch, dberr_t *err){ ut_ad(block->page.lock.have_any()); ut_ad(block->page.buf_fix_count());
if (fil_page_get_type(block->page.frame) != FIL_PAGE_INDEX || !page_is_leaf(block->page.frame)) return false;
if (rw_latch != RW_X_LATCH) { if (rw_latch == RW_S_LATCH) { if (!block->page.lock.s_x_upgrade()) { uint32_t state; state= block->page.state(); if (state < buf_page_t::UNFIXED) { fail: block->page.lock.x_unlock(); return true; } ut_ad(state & ~buf_page_t::LRU_MASK); ut_ad(state < buf_page_t::READ_FIX); if (state < buf_page_t::IBUF_EXIST || state >= buf_page_t::REINIT) /* ibuf_merge_or_delete_for_page() was already invoked in
another thread. */ goto downgrade_to_s; } } else { ut_ad(rw_latch == RW_SX_LATCH); block->page.lock.u_x_upgrade(); } }
ut_ad(block->page.lock.have_x()); block->page.clear_ibuf_exist(); if (dberr_t e= ibuf_merge_or_delete_for_page(block, block->page.id(), block->zip_size())) { if (err) *err= e; goto fail; }
switch (rw_latch) { default: ut_ad(rw_latch == RW_X_LATCH); break; case RW_SX_LATCH: block->page.lock.x_u_downgrade(); break; case RW_S_LATCH: downgrade_to_s: block->page.lock.x_u_downgrade(); block->page.lock.u_s_downgrade(); break; }
return false;}
ATTRIBUTE_COLDbuf_block_t *buf_pool_t::unzip(buf_page_t *b, buf_pool_t::hash_chain &chain) noexcept{ buf_block_t *block= buf_LRU_get_free_block(false); buf_block_init_low(block); page_hash_latch &hash_lock= page_hash.lock_get(chain); wait_for_unfix: mysql_mutex_lock(&mutex); hash_lock.lock();
/* b->lock implies !b->can_relocate() */ ut_ad(b->lock.have_x()); ut_ad(b == page_hash.get(b->id(), chain));
/* Wait for b->unfix() in any other threads. */ uint32_t state= b->state(); ut_ad(buf_page_t::buf_fix_count(state)); ut_ad(!buf_page_t::is_freed(state));
switch (state) { case buf_page_t::UNFIXED + 1: case buf_page_t::IBUF_EXIST + 1: case buf_page_t::REINIT + 1: break; default: ut_ad(state < buf_page_t::READ_FIX);
if (state < buf_page_t::UNFIXED + 1) { ut_ad(state > buf_page_t::FREED); b->lock.x_unlock(); hash_lock.unlock(); buf_LRU_block_free_non_file_page(block); mysql_mutex_unlock(&mutex); b->unfix(); return nullptr; }
mysql_mutex_unlock(&mutex); hash_lock.unlock(); std::this_thread::sleep_for(std::chrono::microseconds(100)); goto wait_for_unfix; }
/* Ensure that another buf_page_get_low() or buf_page_t::page_fix()
will wait for block->page.lock.x_unlock(). buf_relocate() will copy the state from b to block and replace b with block in page_hash. */ b->set_state(buf_page_t::READ_FIX);
mysql_mutex_lock(&flush_list_mutex); buf_relocate(b, &block->page);
/* X-latch the block for the duration of the decompression. */ block->page.lock.x_lock();
buf_flush_relocate_on_flush_list(b, &block->page); mysql_mutex_unlock(&flush_list_mutex);
/* Insert at the front of unzip_LRU list */ buf_unzip_LRU_add_block(block, false);
mysql_mutex_unlock(&mutex); hash_lock.unlock();
#if defined SUX_LOCK_GENERIC || defined UNIV_DEBUG
b->lock.x_unlock(); b->lock.free();#endif
ut_free(b);
n_pend_unzip++; const bool ok{buf_zip_decompress(block, false)}; n_pend_unzip--;
if (UNIV_UNLIKELY(!ok)) { mysql_mutex_lock(&mutex); block->page.read_unfix(state); block->page.lock.x_unlock(); if (!buf_LRU_free_page(&block->page, true)) ut_ad(0); mysql_mutex_unlock(&mutex); return nullptr; } else block->page.read_unfix(state);
return block;}
buf_block_t *buf_pool_t::page_fix(const page_id_t id, dberr_t *err, buf_pool_t::page_fix_conflicts c) noexcept{ ha_handler_stats *const stats= mariadb_stats; buf_inc_get(stats); auto& chain= page_hash.cell_get(id.fold()); page_hash_latch &hash_lock= page_hash.lock_get(chain); for (;;) { hash_lock.lock_shared(); buf_page_t *b= page_hash.get(id, chain); if (b && !watch_is_sentinel(*b)) { uint32_t state= b->fix() + 1; hash_lock.unlock_shared();
if (UNIV_UNLIKELY(state < buf_page_t::UNFIXED)) { ut_ad(state > buf_page_t::FREED); if (c == FIX_ALSO_FREED && b->id() == id) { ut_ad(state == buf_page_t::FREED + 1); return reinterpret_cast<buf_block_t*>(b); } /* The page was marked as freed or corrupted. */ unfix_corrupted: b->unfix(); corrupted: if (err) *err= DB_CORRUPTION; return nullptr; }
if ((state >= buf_page_t::READ_FIX && state < buf_page_t::WRITE_FIX) || (state >= buf_page_t::IBUF_EXIST && state < buf_page_t::REINIT)) { if (c == FIX_NOWAIT) { would_block: b->unfix(); return reinterpret_cast<buf_block_t*>(-1); }
if (UNIV_LIKELY(b->frame != nullptr)) ut_ad(b->frame==reinterpret_cast<buf_block_t*>(b)->frame_address()); else if (state < buf_page_t::READ_FIX) goto unzip; else { wait_for_unzip: b->unfix(); std::this_thread::sleep_for(std::chrono::microseconds(100)); continue; } b->lock.s_lock(); state= b->state(); ut_ad(state < buf_page_t::READ_FIX || state >= buf_page_t::WRITE_FIX);
if (state >= buf_page_t::IBUF_EXIST && state < buf_page_t::REINIT && buf_page_ibuf_merge_try(reinterpret_cast<buf_block_t*>(b), RW_S_LATCH, err)) goto unfix_corrupted;
b->lock.s_unlock(); }
if (UNIV_UNLIKELY(!b->frame)) { unzip: if (b->lock.x_lock_try()); else if (c == FIX_NOWAIT) goto would_block; else goto wait_for_unzip;
buf_block_t *block= unzip(b, chain); if (!block) goto corrupted;
b= &block->page; state= b->state();
if (state >= buf_page_t::IBUF_EXIST && state < buf_page_t::REINIT && buf_page_ibuf_merge_try(block, RW_X_LATCH, err)) goto unfix_corrupted;
b->lock.x_unlock(); }
return reinterpret_cast<buf_block_t*>(b); }
hash_lock.unlock_shared();
if (c == FIX_NOWAIT) return reinterpret_cast<buf_block_t*>(-1);
switch (dberr_t local_err= buf_read_page(id)) { default: if (err) *err= local_err; return nullptr; case DB_SUCCESS: case DB_SUCCESS_LOCKED_REC: mariadb_increment_pages_read(stats); buf_read_ahead_random(id, false); } }}
TRANSACTIONAL_TARGETuint32_t buf_pool_t::page_guess(buf_block_t *b, page_hash_latch &latch, const page_id_t id) noexcept{ transactional_shared_lock_guard<page_hash_latch> g{latch};#ifndef HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT
/* shrunk() and my_virtual_mem_decommit() could retain the original
contents of the virtual memory range or zero it out immediately or with a delay. Any zeroing out may lead to a false positive for b->page.id() == id but never for b->page.state(). At the time of the shrunk() call, shrink() and buf_LRU_block_free_non_file_page() should guarantee that b->page.state() is equal to buf_page_t::NOT_USED (0) for all to-be-freed blocks. */#else
/* shrunk() made the memory inaccessible. */ if (UNIV_UNLIKELY(reinterpret_cast<char*>(b) >= memory + size_in_bytes)) return 0;#endif
const page_id_t block_id{b->page.id()};#ifndef HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT
/* shrunk() may have invoked MEM_UNDEFINED() on this memory to be able
to catch any unintended access elsewhere in our code. */ MEM_MAKE_DEFINED(&block_id, sizeof block_id);#endif
if (id == block_id) { uint32_t state= b->page.state();#ifndef HAVE_UNACCESSIBLE_AFTER_MEM_DECOMMIT
/* shrunk() may have invoked MEM_UNDEFINED() on this memory to be able
to catch any unintended access elsewhere in our code. */ MEM_MAKE_DEFINED(&state, sizeof state);#endif
/* Ignore guesses that point to read-fixed blocks. We can only
avoid a race condition by looking up the block via page_hash. */ if ((state >= buf_page_t::FREED && state < buf_page_t::READ_FIX) || state >= buf_page_t::WRITE_FIX) return b->page.fix(); ut_ad(b->page.frame); } return 0;}
/** Low level function used to get access to a database page.
@param[in] page_id page id@param[in] zip_size ROW_FORMAT=COMPRESSED page size, or 0@param[in] rw_latch latch mode@param[in] guess guessed block or NULL@param[in] mode BUF_GET, BUF_GET_IF_IN_POOL,BUF_PEEK_IF_IN_POOL, or BUF_GET_IF_IN_POOL_OR_WATCH@param[in] mtr mini-transaction@param[out] err DB_SUCCESS or error code@param[in] allow_ibuf_merge Allow change buffer merge to happen@return pointer to the block@retval nullptr if the block is corrupted or unavailable */TRANSACTIONAL_TARGETbuf_block_t*buf_page_get_low( const page_id_t page_id, ulint zip_size, rw_lock_type_t rw_latch, buf_block_t* guess, ulint mode, mtr_t* mtr, dberr_t* err, bool allow_ibuf_merge) noexcept{ ulint retries = 0;
ut_ad(mtr->is_active()); ut_ad(rw_latch != RW_NO_LATCH || !allow_ibuf_merge);
if (err) { *err = DB_SUCCESS; }
#ifdef UNIV_DEBUG
switch (mode) { default: ut_ad(!allow_ibuf_merge); ut_ad(mode == BUF_PEEK_IF_IN_POOL); break; case BUF_GET_POSSIBLY_FREED: case BUF_GET_IF_IN_POOL: /* The caller may pass a dummy page size,
because it does not really matter. */ break; case BUF_GET: case BUF_GET_IF_IN_POOL_OR_WATCH: ut_ad(!mtr->is_freeing_tree()); fil_space_t* s = fil_space_get(page_id.space()); ut_ad(s); ut_ad(s->zip_size() == zip_size); }#endif /* UNIV_DEBUG */
ut_ad(!ibuf_inside(mtr) || ibuf_page_low(page_id, zip_size, FALSE, NULL));
ha_handler_stats* const stats = mariadb_stats; buf_inc_get(stats); auto& chain= buf_pool.page_hash.cell_get(page_id.fold()); page_hash_latch& hash_lock = buf_pool.page_hash.lock_get(chain);loop: buf_block_t* block = guess; uint32_t state;
if (block && (state = buf_pool.page_guess(block, hash_lock, page_id))) { goto got_block; }
guess = nullptr;
/* A memory transaction would frequently be aborted here. */ hash_lock.lock_shared(); block = reinterpret_cast<buf_block_t*>( buf_pool.page_hash.get(page_id, chain)); if (UNIV_LIKELY(block && !buf_pool.watch_is_sentinel(block->page))) { state = block->page.fix(); hash_lock.unlock_shared(); goto got_block; } hash_lock.unlock_shared();
/* Page not in buf_pool: needs to be read from file */ switch (mode) { case BUF_GET_IF_IN_POOL: case BUF_PEEK_IF_IN_POOL: return nullptr; case BUF_GET_IF_IN_POOL_OR_WATCH: /* Buffer-fixing inside watch_set() will prevent eviction */ block = reinterpret_cast<buf_block_t*> (buf_pool.watch_set(page_id, chain));
if (block) { state = block->page.state(); goto got_block_fixed; }
return nullptr; }
/* The call path is buf_read_page() ->
buf_read_page_low() (fil_space_t::io()) -> buf_page_t::read_complete() -> buf_decrypt_after_read(). Here fil_space_t* is used and we decrypt -> buf_page_check_corrupt() where page checksums are compared. Decryption, decompression as well as error handling takes place at a lower level. Here we only need to know whether the page really is corrupted, or if an encrypted page with a valid checksum cannot be decypted. */
switch (dberr_t local_err = buf_read_page(page_id)) { case DB_SUCCESS: case DB_SUCCESS_LOCKED_REC: mariadb_increment_pages_read(stats); buf_read_ahead_random(page_id, ibuf_inside(mtr)); break; default: if (mode != BUF_GET_POSSIBLY_FREED && retries++ < BUF_PAGE_READ_MAX_RETRIES) { DBUG_EXECUTE_IF("intermittent_read_failure", retries = BUF_PAGE_READ_MAX_RETRIES;); } /* fall through */ case DB_PAGE_CORRUPTED: if (err) { *err = local_err; } return nullptr; }
ut_d(if (!(++buf_dbg_counter % 5771)) buf_pool.validate()); goto loop;
got_block: state++;got_block_fixed: ut_ad(state > buf_page_t::FREED);
if (state > buf_page_t::READ_FIX && state < buf_page_t::WRITE_FIX) { if (mode == BUF_PEEK_IF_IN_POOL) {ignore_block: block->unfix();ignore_unfixed: ut_ad(mode == BUF_GET_POSSIBLY_FREED || mode == BUF_PEEK_IF_IN_POOL); if (err) { *err = DB_CORRUPTION; } return nullptr; }
if (UNIV_UNLIKELY(!block->page.frame)) { goto wait_for_unzip; } /* A read-fix is released after block->page.lock
in buf_page_t::read_complete() or buf_pool_t::corrupted_evict(), or after buf_zip_decompress() in this function. */ block->page.lock.s_lock(); state = block->page.state(); ut_ad(state < buf_page_t::READ_FIX || state >= buf_page_t::WRITE_FIX); const page_id_t id{block->page.id()}; block->page.lock.s_unlock();
if (UNIV_UNLIKELY(state < buf_page_t::UNFIXED)) { if (UNIV_UNLIKELY(id == page_id)) { /* The page read was completed, and
another thread marked the page as free while we were waiting. */ goto ignore_block; }
ut_ad(id == page_id_t{~0ULL}); block->page.unfix();
if (++retries < BUF_PAGE_READ_MAX_RETRIES) { goto loop; }
if (err) { *err = DB_PAGE_CORRUPTED; }
return nullptr; } ut_ad(id == page_id); } else if (mode != BUF_PEEK_IF_IN_POOL) { } else if (UNIV_UNLIKELY(!block->page.frame)) { /* The BUF_PEEK_IF_IN_POOL mode is mainly used for dropping an
adaptive hash index. There cannot be an adaptive hash index for a compressed-only page. */ goto ignore_block; }
ut_ad(mode == BUF_GET_IF_IN_POOL || mode == BUF_PEEK_IF_IN_POOL || block->zip_size() == zip_size);
#if defined UNIV_DEBUG || defined UNIV_IBUF_DEBUG
re_evict: if (mode != BUF_GET_IF_IN_POOL && mode != BUF_GET_IF_IN_POOL_OR_WATCH) { } else if (!ibuf_debug || recv_recovery_is_on()) { } else if (fil_space_t* space = fil_space_t::get(page_id.space())) { for (ulint i = 0; i < mtr->get_savepoint(); i++) { if (buf_block_t* b = mtr->block_at_savepoint(i)) { if (b->page.oldest_modification() > 2 && b->page.lock.have_any()) { /* We are holding a dirty page latch
that would hang buf_flush_sync(). */ space->release(); goto re_evict_fail; } } }
/* Try to evict the block from the buffer pool, to use the
insert buffer (change buffer) as much as possible. */
mysql_mutex_lock(&buf_pool.mutex);
block->unfix();
/* Blocks cannot be relocated or enter or exit the
buf_pool while we are holding the buf_pool.mutex. */ const bool evicted = buf_LRU_free_page(&block->page, true); space->release();
if (!evicted) { block->fix(); }
mysql_mutex_unlock(&buf_pool.mutex);
if (evicted) { if (mode == BUF_GET_IF_IN_POOL_OR_WATCH) { buf_pool.watch_set(page_id, chain); } return(NULL); }
buf_flush_sync();
state = block->page.state();
if (state == buf_page_t::UNFIXED + 1 && !block->page.oldest_modification()) { goto re_evict; }
/* Failed to evict the page; change it directly */ }re_evict_fail:#endif /* UNIV_DEBUG || UNIV_IBUF_DEBUG */
if (UNIV_UNLIKELY(state < buf_page_t::UNFIXED)) { goto ignore_block; } ut_ad((~buf_page_t::LRU_MASK) & state); ut_ad(state > buf_page_t::WRITE_FIX || state < buf_page_t::READ_FIX);
if (UNIV_UNLIKELY(!block->page.frame)) { if (!block->page.lock.x_lock_try()) {wait_for_unzip: /* The page is being read or written, or
another thread is executing buf_pool.unzip() on it. */ block->page.unfix(); std::this_thread::sleep_for( std::chrono::microseconds(100)); goto loop; }
block = buf_pool.unzip(&block->page, chain);
if (!block) { goto ignore_unfixed; }
block->page.lock.x_unlock(); }
#ifdef UNIV_DEBUG
if (!(++buf_dbg_counter % 5771)) buf_pool.validate();#endif /* UNIV_DEBUG */
/* The state = block->page.state() may be stale at this point,
and in fact, at any point of time if we consider its buffer-fix component. If the block is being read into the buffer pool, it is possible that buf_page_t::read_complete() will invoke buf_pool_t::corrupted_evict() and therefore invalidate it (invoke buf_page_t::set_corrupt_id() and set the state to FREED). Therefore, after acquiring the page latch we must recheck the state. */
switch (rw_latch) { case RW_NO_LATCH: ut_ad(!allow_ibuf_merge); mtr->memo_push(block, MTR_MEMO_BUF_FIX); return block; case RW_S_LATCH: block->page.lock.s_lock(); break; case RW_SX_LATCH: block->page.lock.u_lock(); ut_ad(!block->page.is_io_fixed()); break; default: ut_ad(rw_latch == RW_X_LATCH); if (block->page.lock.x_lock_upgraded()) { ut_ad(block->page.id() == page_id); block->unfix(); mtr->page_lock_upgrade(*block); return block; } }
mtr->memo_push(block, mtr_memo_type_t(rw_latch)); state = block->page.state();
if (UNIV_UNLIKELY(state < buf_page_t::UNFIXED)) { corrupted: mtr->release_last_page(); goto ignore_unfixed; }
ut_ad(state < buf_page_t::READ_FIX || state > buf_page_t::WRITE_FIX); if (state >= buf_page_t::IBUF_EXIST && state < buf_page_t::REINIT && allow_ibuf_merge && buf_page_ibuf_merge_try(block, rw_latch, err)) { ut_ad(block == mtr->at_savepoint(mtr->get_savepoint() - 1)); mtr->lock_register(mtr->get_savepoint() - 1, MTR_MEMO_BUF_FIX); goto corrupted; }#ifdef BTR_CUR_HASH_ADAPT
btr_search_drop_page_hash_index(block, true);#endif /* BTR_CUR_HASH_ADAPT */
ut_ad(block->page.frame == block->frame_address()); ut_ad(page_id_t(page_get_space_id(block->page.frame), page_get_page_no(block->page.frame)) == page_id); return block;}
/** Get access to a database page. Buffered redo log may be applied.
@param[in] page_id page id@param[in] zip_size ROW_FORMAT=COMPRESSED page size, or 0@param[in] rw_latch latch mode@param[in] guess guessed block or NULL@param[in] mode BUF_GET, BUF_GET_IF_IN_POOL,BUF_PEEK_IF_IN_POOL, or BUF_GET_IF_IN_POOL_OR_WATCH@param[in,out] mtr mini-transaction, or NULL@param[out] err DB_SUCCESS or error code@param[in] allow_ibuf_merge Allow change buffer merge to happen@return pointer to the block@retval nullptr if the block is corrupted or unavailable */buf_block_t*buf_page_get_gen( const page_id_t page_id, ulint zip_size, rw_lock_type_t rw_latch, buf_block_t* guess, ulint mode, mtr_t* mtr, dberr_t* err, bool allow_ibuf_merge) noexcept{ buf_block_t *block= recv_sys.recover(page_id); if (UNIV_LIKELY(!block)) return buf_page_get_low(page_id, zip_size, rw_latch, guess, mode, mtr, err, allow_ibuf_merge); else if (UNIV_UNLIKELY(block == reinterpret_cast<buf_block_t*>(-1))) { corrupted: if (err) *err= DB_CORRUPTION; return nullptr; } auto s= block->page.fix(); ut_ad(s >= buf_page_t::FREED); /* The block may be write-fixed at this point because we are not
holding a lock, but it must not be read-fixed. */ ut_ad(s < buf_page_t::READ_FIX || s >= buf_page_t::WRITE_FIX); if (err) *err= DB_SUCCESS; const bool must_merge= allow_ibuf_merge && ibuf_page_exists(page_id, block->zip_size()); if (s < buf_page_t::UNFIXED) { got_freed_page: ut_ad(mode == BUF_GET_POSSIBLY_FREED || mode == BUF_PEEK_IF_IN_POOL); mysql_mutex_lock(&buf_pool.mutex); block->page.unfix(); buf_LRU_free_page(&block->page, true); mysql_mutex_unlock(&buf_pool.mutex); goto corrupted; } else if (must_merge && fil_page_get_type(block->page.frame) == FIL_PAGE_INDEX && page_is_leaf(block->page.frame)) { block->page.lock.x_lock(); s= block->page.state(); ut_ad(s > buf_page_t::FREED); ut_ad(s < buf_page_t::READ_FIX); if (s < buf_page_t::UNFIXED) { block->page.lock.x_unlock(); goto got_freed_page; } else { if (block->page.is_ibuf_exist()) block->page.clear_ibuf_exist(); if (dberr_t e= ibuf_merge_or_delete_for_page(block, page_id, block->zip_size())) { if (err) *err= e; buf_pool.corrupted_evict(&block->page, s); return nullptr; } }
switch (rw_latch) { case RW_NO_LATCH: block->page.lock.x_unlock(); case RW_X_LATCH: break; case RW_SX_LATCH: block->page.lock.x_u_downgrade(); break; case RW_S_LATCH: block->page.lock.x_u_downgrade(); block->page.lock.u_s_downgrade(); }
mtr->memo_push(block, mtr_memo_type_t(rw_latch)); return block; } mtr->page_lock(block, rw_latch); return block;}
buf_block_t *buf_page_optimistic_fix(buf_block_t *block, page_id_t id) noexcept{ buf_pool_t::hash_chain &chain= buf_pool.page_hash.cell_get(id.fold()); page_hash_latch &hash_lock= buf_pool.page_hash.lock_get(chain); if (uint32_t state= buf_pool.page_guess(block, hash_lock, id)) { if (UNIV_LIKELY(state >= buf_page_t::UNFIXED)) return block; else /* Refuse access to pages that are marked as freed in the data file. */ block->page.unfix(); } return nullptr;}
buf_block_t *buf_page_optimistic_get(buf_block_t *block, rw_lock_type_t rw_latch, uint64_t modify_clock, mtr_t *mtr) noexcept{ ut_ad(mtr->is_active()); ut_ad(rw_latch == RW_S_LATCH || rw_latch == RW_X_LATCH); ut_ad(block->page.buf_fix_count());
if (rw_latch == RW_S_LATCH) { if (!block->page.lock.s_lock_try()) { fail: block->page.unfix(); return nullptr; }
ut_ad(!ibuf_inside(mtr) || ibuf_page(block->page.id(), block->zip_size(), nullptr));
if (modify_clock != block->modify_clock || block->page.is_freed()) { block->page.lock.s_unlock(); goto fail; }
ut_ad(!block->page.is_read_fixed()); buf_page_make_young_if_needed(&block->page); mtr->memo_push(block, MTR_MEMO_PAGE_S_FIX); } else if (block->page.lock.have_u_not_x()) { block->page.lock.u_x_upgrade(); block->page.unfix(); mtr->page_lock_upgrade(*block); ut_ad(modify_clock == block->modify_clock); } else if (!block->page.lock.x_lock_try()) goto fail; else { ut_ad(!block->page.is_io_fixed()); ut_ad(!ibuf_inside(mtr) || ibuf_page(block->page.id(), block->zip_size(), nullptr));
if (modify_clock != block->modify_clock || block->page.is_freed()) { block->page.lock.x_unlock(); goto fail; }
buf_page_make_young_if_needed(&block->page); mtr->memo_push(block, MTR_MEMO_PAGE_X_FIX); }
ut_d(if (!(++buf_dbg_counter % 5771)) buf_pool.validate()); ut_d(const auto state = block->page.state()); ut_ad(state > buf_page_t::UNFIXED); ut_ad(state < buf_page_t::READ_FIX || state > buf_page_t::WRITE_FIX); ut_ad(~buf_page_t::LRU_MASK & state); ut_ad(block->page.frame);
return block;}
/** Try to S-latch a page.
Suitable for using when holding the lock_sys latches (as it avoids deadlock).@param[in] page_id page identifier@param[in,out] mtr mini-transaction@return the block@retval nullptr if an S-latch cannot be granted immediately */TRANSACTIONAL_TARGETbuf_block_t *buf_page_try_get(const page_id_t page_id, mtr_t *mtr) noexcept{ ut_ad(mtr); ut_ad(mtr->is_active()); buf_pool_t::hash_chain &chain= buf_pool.page_hash.cell_get(page_id.fold()); buf_block_t *block;
{ transactional_shared_lock_guard<page_hash_latch> g {buf_pool.page_hash.lock_get(chain)}; block= reinterpret_cast<buf_block_t*> (buf_pool.page_hash.get(page_id, chain)); if (!block || !block->page.frame || !block->page.lock.s_lock_try()) return nullptr; }
block->page.fix(); ut_ad(!block->page.is_read_fixed()); mtr->memo_push(block, MTR_MEMO_PAGE_S_FIX);
#ifdef UNIV_DEBUG
if (!(++buf_dbg_counter % 5771)) buf_pool.validate();#endif /* UNIV_DEBUG */
ut_ad(block->page.buf_fix_count()); ut_ad(block->page.id() == page_id);
buf_inc_get(mariadb_stats); return block;}
/** Initialize the block.
@param page_id page identifier@param zip_size ROW_FORMAT=COMPRESSED page size, or 0@param fix initial buf_fix_count() */void buf_block_t::initialise(const page_id_t page_id, ulint zip_size, uint32_t fix) noexcept{ ut_ad(!page.in_file()); buf_block_init_low(this); page.init(fix, page_id); page.set_os_used(); page_zip_set_size(&page.zip, zip_size);}
TRANSACTIONAL_TARGETstatic buf_block_t *buf_page_create_low(page_id_t page_id, ulint zip_size, mtr_t *mtr, buf_block_t *free_block) noexcept{ ut_ad(mtr->is_active()); ut_ad(page_id.space() != 0 || !zip_size);
free_block->initialise(page_id, zip_size, buf_page_t::MEMORY);
buf_pool_t::hash_chain &chain= buf_pool.page_hash.cell_get(page_id.fold());retry: mysql_mutex_lock(&buf_pool.mutex);
buf_page_t *bpage= buf_pool.page_hash.get(page_id, chain);
if (bpage && !buf_pool.watch_is_sentinel(*bpage)) {#ifdef BTR_CUR_HASH_ADAPT
const dict_index_t *drop_hash_entry= nullptr;#endif
bool ibuf_exist= false;
if (!mtr->have_x_latch(reinterpret_cast<const buf_block_t&>(*bpage))) { /* Buffer-fix the block to prevent the block being concurrently freed
after we release the buffer pool mutex. It should work fine with concurrent load of the page (free on disk) to buffer pool due to possible read ahead. After we find a zero filled page during load, we call buf_pool_t::corrupted_evict, where we try to wait for all buffer fixes to go away only after resetting the page ID and releasing the page latch. */ auto state= bpage->fix(); DBUG_EXECUTE_IF("ib_buf_create_intermittent_wait", { static bool need_to_wait = false; need_to_wait = !need_to_wait; /* Simulate try lock failure in every alternate call. */ if (need_to_wait) { goto must_wait; } });
if (!bpage->lock.x_lock_try()) {#ifndef DBUG_OFF
must_wait:#endif
mysql_mutex_unlock(&buf_pool.mutex);
bpage->lock.x_lock(); const page_id_t id{bpage->id()}; if (UNIV_UNLIKELY(id != page_id)) { ut_ad(id.is_corrupted()); ut_ad(bpage->is_freed()); bpage->unfix(); bpage->lock.x_unlock(); goto retry; } mysql_mutex_lock(&buf_pool.mutex); state= bpage->state(); ut_ad(!bpage->is_io_fixed(state)); ut_ad(bpage->buf_fix_count(state)); } else state= bpage->state();
ut_ad(state >= buf_page_t::FREED); ut_ad(state < buf_page_t::READ_FIX);
if (state < buf_page_t::UNFIXED) bpage->set_reinit(buf_page_t::FREED); else { bpage->set_reinit(state & buf_page_t::LRU_MASK); ibuf_exist= (state & buf_page_t::LRU_MASK) == buf_page_t::IBUF_EXIST; }
if (UNIV_LIKELY(bpage->frame != nullptr)) { mysql_mutex_unlock(&buf_pool.mutex); buf_block_t *block= reinterpret_cast<buf_block_t*>(bpage); ut_ad(bpage->frame == block->frame_address()); mtr->memo_push(block, MTR_MEMO_PAGE_X_FIX);#ifdef BTR_CUR_HASH_ADAPT
drop_hash_entry= block->index;#endif
} else { page_hash_latch &hash_lock= buf_pool.page_hash.lock_get(chain); /* It does not make sense to use transactional_lock_guard here,
because buf_relocate() would likely make the memory transaction too large. */ hash_lock.lock();
mysql_mutex_lock(&buf_pool.flush_list_mutex); buf_relocate(bpage, &free_block->page); free_block->page.lock.x_lock(); buf_flush_relocate_on_flush_list(bpage, &free_block->page); mysql_mutex_unlock(&buf_pool.flush_list_mutex);
buf_unzip_LRU_add_block(free_block, FALSE);
mysql_mutex_unlock(&buf_pool.mutex); hash_lock.unlock();#if defined SUX_LOCK_GENERIC || defined UNIV_DEBUG
bpage->lock.x_unlock(); bpage->lock.free();#endif
ut_free(bpage); mtr->memo_push(free_block, MTR_MEMO_PAGE_X_FIX); bpage= &free_block->page; } } else { mysql_mutex_unlock(&buf_pool.mutex); ut_ad(bpage->frame == reinterpret_cast<buf_block_t*>(bpage)->frame_address());#ifdef BTR_CUR_HASH_ADAPT
ut_ad(!reinterpret_cast<buf_block_t*>(bpage)->index);#endif
const auto state= bpage->state(); ut_ad(state >= buf_page_t::FREED); bpage->set_reinit(state < buf_page_t::UNFIXED ? buf_page_t::FREED : state & buf_page_t::LRU_MASK); }
#ifdef BTR_CUR_HASH_ADAPT
if (drop_hash_entry) btr_search_drop_page_hash_index(reinterpret_cast<buf_block_t*>(bpage), false);#endif /* BTR_CUR_HASH_ADAPT */
if (ibuf_exist && !recv_recovery_is_on()) ibuf_merge_or_delete_for_page(nullptr, page_id, zip_size);
return reinterpret_cast<buf_block_t*>(bpage); }
/* If we get here, the page was not in buf_pool: init it there */
DBUG_PRINT("ib_buf", ("create page %u:%u", page_id.space(), page_id.page_no()));
bpage= &free_block->page;
ut_ad(bpage->state() == buf_page_t::MEMORY); bpage->lock.x_lock();
/* The block must be put to the LRU list */ buf_LRU_add_block(bpage, false); { transactional_lock_guard<page_hash_latch> g {buf_pool.page_hash.lock_get(chain)}; bpage->set_state(buf_page_t::REINIT + 1); buf_pool.page_hash.append(chain, bpage); }
if (UNIV_UNLIKELY(zip_size)) { bpage->zip.data= buf_buddy_alloc(zip_size);
/* To maintain the invariant block->in_unzip_LRU_list ==
block->page.belongs_to_unzip_LRU() we have to add this block to unzip_LRU after block->page.zip.data is set. */ ut_ad(bpage->belongs_to_unzip_LRU()); buf_unzip_LRU_add_block(reinterpret_cast<buf_block_t*>(bpage), FALSE); }
buf_pool.stat.n_pages_created++; mysql_mutex_unlock(&buf_pool.mutex);
mtr->memo_push(reinterpret_cast<buf_block_t*>(bpage), MTR_MEMO_PAGE_X_FIX);
bpage->set_accessed();
/* Delete possible entries for the page from the insert buffer:
such can exist if the page belonged to an index which was dropped */ if (page_id < page_id_t{SRV_SPACE_ID_UPPER_BOUND, 0} && !srv_is_undo_tablespace(page_id.space()) && !recv_recovery_is_on()) ibuf_merge_or_delete_for_page(nullptr, page_id, zip_size);
static_assert(FIL_PAGE_PREV + 4 == FIL_PAGE_NEXT, "adjacent"); memset_aligned<8>(bpage->frame + FIL_PAGE_PREV, 0xff, 8); mach_write_to_2(bpage->frame + FIL_PAGE_TYPE, FIL_PAGE_TYPE_ALLOCATED);
/* FIL_PAGE_FILE_FLUSH_LSN_OR_KEY_VERSION is only used on the
following pages: (1) The first page of the InnoDB system tablespace (page 0:0) (2) FIL_RTREE_SPLIT_SEQ_NUM on R-tree pages (3) key_version on encrypted pages (not page 0:0) */
memset(bpage->frame + FIL_PAGE_FILE_FLUSH_LSN_OR_KEY_VERSION, 0, 8); memset_aligned<8>(bpage->frame + FIL_PAGE_LSN, 0, 8);
#ifdef UNIV_DEBUG
if (!(++buf_dbg_counter % 5771)) buf_pool.validate();#endif /* UNIV_DEBUG */
return reinterpret_cast<buf_block_t*>(bpage);}
/** Initialize a page in the buffer pool. The page is usually not read
from a file even if it cannot be found in the buffer buf_pool. This is oneof the functions which perform to a block a state transition NOT_USED =>FILE_PAGE (the other is buf_page_get_gen).@param[in,out] space space object@param[in] offset offset of the tablespace or deferred space id if space object is null@param[in] zip_size ROW_FORMAT=COMPRESSED page size, or 0@param[in,out] mtr mini-transaction@param[in,out] free_block pre-allocated buffer block@return pointer to the block, page bufferfixed */buf_block_t*buf_page_create(fil_space_t *space, uint32_t offset, ulint zip_size, mtr_t *mtr, buf_block_t *free_block) noexcept{ space->free_page(offset, false); return buf_page_create_low({space->id, offset}, zip_size, mtr, free_block);}
/** Initialize a page in buffer pool while initializing the
deferred tablespace@param space_id space identfier@param zip_size ROW_FORMAT=COMPRESSED page size or 0@param mtr mini-transaction@param free_block pre-allocated buffer block@return pointer to the block, page bufferfixed */buf_block_t* buf_page_create_deferred(uint32_t space_id, ulint zip_size, mtr_t *mtr, buf_block_t *free_block) noexcept{ return buf_page_create_low({space_id, 0}, zip_size, mtr, free_block);}
/** Monitor the buffer page read/write activity, and increment corresponding
counter value in MONITOR_MODULE_BUF_PAGE.@param bpage buffer page whose read or write was completed@param read true=read, false=write */ATTRIBUTE_COLDvoid buf_page_monitor(const buf_page_t &bpage, bool read) noexcept{ monitor_id_t counter;
const byte* frame = bpage.zip.data ? bpage.zip.data : bpage.frame;
switch (fil_page_get_type(frame)) { ulint level; case FIL_PAGE_TYPE_INSTANT: case FIL_PAGE_INDEX: case FIL_PAGE_RTREE: level = btr_page_get_level(frame);
/* Check if it is an index page for insert buffer */ if (fil_page_get_type(frame) == FIL_PAGE_INDEX && btr_page_get_index_id(frame) == (index_id_t)(DICT_IBUF_ID_MIN + IBUF_SPACE_ID)) { if (level == 0) { counter = MONITOR_RW_COUNTER( read, MONITOR_INDEX_IBUF_LEAF_PAGE); } else { counter = MONITOR_RW_COUNTER( read, MONITOR_INDEX_IBUF_NON_LEAF_PAGE); } } else { if (level == 0) { counter = MONITOR_RW_COUNTER( read, MONITOR_INDEX_LEAF_PAGE); } else { counter = MONITOR_RW_COUNTER( read, MONITOR_INDEX_NON_LEAF_PAGE); } } break;
case FIL_PAGE_UNDO_LOG: counter = MONITOR_RW_COUNTER(read, MONITOR_UNDO_LOG_PAGE); break;
case FIL_PAGE_INODE: counter = MONITOR_RW_COUNTER(read, MONITOR_INODE_PAGE); break;
case FIL_PAGE_IBUF_FREE_LIST: counter = MONITOR_RW_COUNTER(read, MONITOR_IBUF_FREELIST_PAGE); break;
case FIL_PAGE_IBUF_BITMAP: counter = MONITOR_RW_COUNTER(read, MONITOR_IBUF_BITMAP_PAGE); break;
case FIL_PAGE_TYPE_SYS: counter = MONITOR_RW_COUNTER(read, MONITOR_SYSTEM_PAGE); break;
case FIL_PAGE_TYPE_TRX_SYS: counter = MONITOR_RW_COUNTER(read, MONITOR_TRX_SYSTEM_PAGE); break;
case FIL_PAGE_TYPE_FSP_HDR: counter = MONITOR_RW_COUNTER(read, MONITOR_FSP_HDR_PAGE); break;
case FIL_PAGE_TYPE_XDES: counter = MONITOR_RW_COUNTER(read, MONITOR_XDES_PAGE); break;
case FIL_PAGE_TYPE_BLOB: counter = MONITOR_RW_COUNTER(read, MONITOR_BLOB_PAGE); break;
case FIL_PAGE_TYPE_ZBLOB: counter = MONITOR_RW_COUNTER(read, MONITOR_ZBLOB_PAGE); break;
case FIL_PAGE_TYPE_ZBLOB2: counter = MONITOR_RW_COUNTER(read, MONITOR_ZBLOB2_PAGE); break;
default: counter = MONITOR_RW_COUNTER(read, MONITOR_OTHER_PAGE); }
MONITOR_INC_NOCHECK(counter);}
/** Check if the encrypted page is corrupted for the full crc32 format.
@param[in] space_id page belongs to space id@param[in] d page@param[in] is_compressed compressed page@return true if page is corrupted or false if it isn't */static bool buf_page_full_crc32_is_corrupted(ulint space_id, const byte* d, bool is_compressed) noexcept{ if (space_id != mach_read_from_4(d + FIL_PAGE_SPACE_ID)) return true;
static_assert(FIL_PAGE_LSN % 4 == 0, "alignment"); static_assert(FIL_PAGE_FCRC32_END_LSN % 4 == 0, "alignment");
return !is_compressed && memcmp_aligned<4>(FIL_PAGE_LSN + 4 + d, d + srv_page_size - FIL_PAGE_FCRC32_END_LSN, 4);}
/** Check if page is maybe compressed, encrypted or both when we encounter
corrupted page. Note that we can't be 100% sure if page is corruptedor decrypt/decompress just failed.@param[in,out] bpage page@param[in] node data file@return whether the operation succeeded@retval DB_SUCCESS if page has been read and is not corrupted@retval DB_PAGE_CORRUPTED if page based on checksum check is corrupted@retval DB_CORRUPTION if the page LSN is in the future@retval DB_DECRYPTION_FAILED if page post encryption checksum matches butafter decryption normal page checksum does not match. */static dberr_t buf_page_check_corrupt(buf_page_t *bpage, const fil_node_t &node){ ut_ad(node.space->referenced());
byte* dst_frame = bpage->zip.data ? bpage->zip.data : bpage->frame; dberr_t err = DB_SUCCESS; uint key_version = buf_page_get_key_version(dst_frame, node.space->flags);
/* In buf_decrypt_after_read we have either decrypted the page if
page post encryption checksum matches and used key_id is found from the encryption plugin. If checksum did not match page was not decrypted and it could be either encrypted and corrupted or corrupted or good page. If we decrypted, there page could still be corrupted if used key does not match. */ const bool seems_encrypted = !node.space->full_crc32() && key_version && node.space->crypt_data && node.space->crypt_data->type != CRYPT_SCHEME_UNENCRYPTED; ut_ad(!node.space->is_temporary() || node.space->full_crc32());
/* If traditional checksums match, we assume that page is
not anymore encrypted. */ if (node.space->full_crc32() && !buf_is_zeroes(span<const byte>(dst_frame, node.space->physical_size())) && (key_version || node.space->is_compressed() || node.space->is_temporary())) { if (buf_page_full_crc32_is_corrupted( bpage->id().space(), dst_frame, node.space->is_compressed())) { err = DB_PAGE_CORRUPTED; } } else { switch (buf_page_is_corrupted(true, dst_frame, node.space->flags)) { case NOT_CORRUPTED: break; case CORRUPTED_OTHER: err = DB_PAGE_CORRUPTED; break; case CORRUPTED_FUTURE_LSN: err = DB_CORRUPTION; break; } }
if (seems_encrypted && err == DB_PAGE_CORRUPTED && bpage->id().page_no() != 0) { err = DB_DECRYPTION_FAILED;
ib::error() << "The page " << bpage->id() << " in file '" << node.name << "' cannot be decrypted; key_version=" << key_version; }
return (err);}
/** Complete a read of a page.
@param node data file@return whether the operation succeeded@retval DB_PAGE_CORRUPTED if the checksum or the page ID is incorrect@retval DB_DECRYPTION_FAILED if the page cannot be decrypted */dberr_t buf_page_t::read_complete(const fil_node_t &node) noexcept{ const page_id_t expected_id{id()}; ut_ad(is_read_fixed()); ut_ad(!buf_dblwr.is_inside(id())); ut_ad(id().space() == node.space->id); ut_ad(zip_size() == node.space->zip_size()); ut_ad(!!zip.ssize == !!zip.data);
const byte *read_frame= zip.data ? zip.data : frame; ut_ad(read_frame);
dberr_t err; if (!buf_page_decrypt_after_read(this, node)) { err= DB_DECRYPTION_FAILED; goto database_corrupted; }
if (belongs_to_unzip_LRU()) { buf_pool.n_pend_unzip++; auto ok= buf_zip_decompress(reinterpret_cast<buf_block_t*>(this), false); buf_pool.n_pend_unzip--;
if (!ok) { err= DB_PAGE_CORRUPTED; goto database_corrupted_compressed; } }
{ const page_id_t read_id(mach_read_from_4(read_frame + FIL_PAGE_SPACE_ID), mach_read_from_4(read_frame + FIL_PAGE_OFFSET));
if (read_id == expected_id); else if (read_id == page_id_t(0, 0)) { /* This is likely an uninitialized (all-zero) page. */ err= DB_FAIL; goto release_page; } else if (!node.space->full_crc32() && page_id_t(0, read_id.page_no()) == expected_id) /* FIL_PAGE_SPACE_ID was written as garbage in the system tablespace
before MySQL 4.1.1, which introduced innodb_file_per_table. */; else if (node.space->full_crc32() && *reinterpret_cast<const uint32_t*> (&read_frame[FIL_PAGE_FCRC32_KEY_VERSION]) && node.space->crypt_data && node.space->crypt_data->type != CRYPT_SCHEME_UNENCRYPTED) { err= DB_DECRYPTION_FAILED; goto release_page; } else { sql_print_error("InnoDB: Space id and page no stored in the page," " read in from %s are " "[page id: space=" UINT32PF ", page number=" UINT32PF "], should be " "[page id: space=" UINT32PF ", page number=" UINT32PF "]", node.name, read_id.space(), read_id.page_no(), expected_id.space(), expected_id.page_no()); err= DB_FAIL; goto release_page; } }
err= buf_page_check_corrupt(this, node); if (UNIV_UNLIKELY(err != DB_SUCCESS)) {database_corrupted: if (belongs_to_unzip_LRU())database_corrupted_compressed: memset_aligned<UNIV_PAGE_SIZE_MIN>(frame, 0, srv_page_size);
if (!srv_force_recovery) goto release_page; }
if (err == DB_PAGE_CORRUPTED || err == DB_DECRYPTION_FAILED) {release_page: if (node.space->full_crc32() && recv_recovery_is_on() && recv_sys.dblwr.find_deferred_page(node, id().page_no(), const_cast<byte*>(read_frame))) { /* Recover from doublewrite buffer */ err= DB_SUCCESS; goto success_page; }
if (recv_sys.free_corrupted_page(expected_id, node)); else if (err == DB_FAIL) err= DB_PAGE_CORRUPTED; else { sql_print_error("InnoDB: Failed to read page " UINT32PF " from file '%s': %s", expected_id.page_no(), node.name, ut_strerr(err));
buf_page_print(read_frame, zip_size());
if (node.space->set_corrupted() && !is_predefined_tablespace(node.space->id)) sql_print_information("InnoDB: You can use CHECK TABLE to scan" " your table for corruption. %s", FORCE_RECOVERY_MSG); }
buf_pool.corrupted_evict(this, buf_page_t::READ_FIX); return err; }success_page:
const bool recovery= recv_recovery_is_on();
if (recovery && !recv_recover_page(node.space, this)) return DB_PAGE_CORRUPTED;
const bool ibuf_may_exist= !recv_no_ibuf_operations && (!expected_id.space() || !is_predefined_tablespace(expected_id.space())) && fil_page_get_type(read_frame) == FIL_PAGE_INDEX && page_is_leaf(read_frame);
if (UNIV_UNLIKELY(MONITOR_IS_ON(MONITOR_MODULE_BUF_PAGE))) buf_page_monitor(*this, true); DBUG_PRINT("ib_buf", ("read page %u:%u", id().space(), id().page_no()));
if (!recovery) { ut_d(auto f=) zip.fix.fetch_sub(ibuf_may_exist ? READ_FIX - IBUF_EXIST : READ_FIX - UNFIXED); ut_ad(f >= READ_FIX); ut_ad(f < WRITE_FIX); } else if (ibuf_may_exist) set_ibuf_exist();
lock.x_unlock(true);
return DB_SUCCESS;}
#ifdef BTR_CUR_HASH_ADAPT
/** Clear the adaptive hash index on all pages in the buffer pool. */ATTRIBUTE_COLD void buf_pool_t::clear_hash_index() noexcept{ std::set<dict_index_t*> garbage;
mysql_mutex_lock(&mutex); ut_ad(!btr_search_enabled);
for (char *extent= memory, *end= memory + block_descriptors_in_bytes(n_blocks); extent < end; extent+= innodb_buffer_pool_extent_size) for (buf_block_t *block= reinterpret_cast<buf_block_t*>(extent), *extent_end= block + pages_in_extent[srv_page_size_shift - UNIV_PAGE_SIZE_SHIFT_MIN]; block < extent_end && reinterpret_cast<char*>(block) < end; block++) { dict_index_t *index= block->index; assert_block_ahi_valid(block);
/* We can clear block->index and block->n_pointers when
holding all AHI latches exclusively; see the comments in buf0buf.h */
if (!index) {# if defined UNIV_AHI_DEBUG || defined UNIV_DEBUG
ut_a(!block->n_pointers);# endif /* UNIV_AHI_DEBUG || UNIV_DEBUG */
continue; }
ut_d(const auto s= block->page.state()); /* Another thread may have set the state to
REMOVE_HASH in buf_LRU_block_remove_hashed().
The state change in buf_pool_t::resize() is not observable here, because in that case we would have !block->index.
In the end, the entire adaptive hash index will be removed. */ ut_ad(s >= buf_page_t::UNFIXED || s == buf_page_t::REMOVE_HASH);# if defined UNIV_AHI_DEBUG || defined UNIV_DEBUG
block->n_pointers= 0;# endif /* UNIV_AHI_DEBUG || UNIV_DEBUG */
if (index->freed()) garbage.insert(index); block->index= nullptr; }
mysql_mutex_unlock(&mutex);
for (dict_index_t *index : garbage) btr_search_lazy_free(index);}#endif /* BTR_CUR_HASH_ADAPT */
#ifdef UNIV_DEBUG
/** Check that all blocks are in a replaceable state.
@return address of a non-free block@retval nullptr if all freed */void buf_pool_t::assert_all_freed() noexcept{ mysql_mutex_lock(&mutex);
for (char *extent= memory, *end= memory + block_descriptors_in_bytes(n_blocks); extent < end; extent+= innodb_buffer_pool_extent_size) for (buf_block_t *block= reinterpret_cast<buf_block_t*>(extent), *extent_end= block + pages_in_extent[srv_page_size_shift - UNIV_PAGE_SIZE_SHIFT_MIN]; block < extent_end && reinterpret_cast<char*>(block) < end; block++) { if (!block->page.in_file()) continue; switch (const lsn_t lsn= block->page.oldest_modification()) { case 0: case 1: break;
case 2: ut_ad(fsp_is_system_temporary(block->page.id().space())); break;
default: if (srv_read_only_mode) { /* The page cleaner is disabled in read-only mode. No pages
can be dirtied, so all of them must be clean. */ ut_ad(lsn == recv_sys.lsn || srv_force_recovery == SRV_FORCE_NO_LOG_REDO); break; }
goto fixed_or_dirty; }
if (!block->page.can_relocate()) fixed_or_dirty: ib::fatal() << "Page " << block->page.id() << " still fixed or dirty"; }
mysql_mutex_unlock(&mutex);}#endif /* UNIV_DEBUG */
/** Refresh the statistics used to print per-second averages. */void buf_refresh_io_stats() noexcept{ buf_pool.last_printout_time = time(NULL); buf_pool.old_stat = buf_pool.stat;}
/** Invalidate all pages in the buffer pool.
All pages must be in a replaceable state (not modified or latched). */void buf_pool_invalidate() noexcept{ /* It is possible that a write batch that has been posted
earlier is still not complete. For buffer pool invalidation to proceed we must ensure there is NO write activity happening. */
os_aio_wait_until_no_pending_writes(false); ut_d(buf_pool.assert_all_freed()); mysql_mutex_lock(&buf_pool.mutex);
while (UT_LIST_GET_LEN(buf_pool.LRU)) { buf_LRU_scan_and_free_block(); }
ut_ad(UT_LIST_GET_LEN(buf_pool.unzip_LRU) == 0);
buf_pool.freed_page_clock = 0; buf_pool.LRU_old = NULL; buf_pool.LRU_old_len = 0; buf_pool.stat.init();
buf_refresh_io_stats(); mysql_mutex_unlock(&buf_pool.mutex);}
#ifdef UNIV_DEBUG
/** Validate the buffer pool. */void buf_pool_t::validate() noexcept{ ulint n_lru = 0; ulint n_flushing = 0; ulint n_free = 0; ulint n_zip = 0;
mysql_mutex_lock(&mutex);
/* Check the uncompressed blocks. */
for (ulint i = 0; i < n_blocks; i++) { const buf_block_t* block = get_nth_page(i); ut_ad(block->page.frame == block->frame_address());
switch (const auto f = block->page.state()) { case buf_page_t::NOT_USED: ut_ad(!block->page.in_LRU_list); n_free++; break; case buf_page_t::MEMORY: case buf_page_t::REMOVE_HASH: /* do nothing */ break; default: if (f >= buf_page_t::READ_FIX && f < buf_page_t::WRITE_FIX) { /* A read-fixed block is not
necessarily in the page_hash yet. */ break; } ut_ad(f >= buf_page_t::FREED); const page_id_t id{block->page.id()}; ut_ad(page_hash.get( id, page_hash.cell_get(id.fold())) == &block->page); n_lru++; } }
/* Check dirty blocks. */
mysql_mutex_lock(&flush_list_mutex); for (buf_page_t* b = UT_LIST_GET_FIRST(flush_list); b; b = UT_LIST_GET_NEXT(list, b)) { ut_ad(b->in_file()); ut_ad(b->oldest_modification()); ut_ad(!fsp_is_system_temporary(b->id().space())); n_flushing++;
if (UNIV_UNLIKELY(!b->frame)) { n_lru++; n_zip++; } const page_id_t id{b->id()}; ut_ad(page_hash.get(id, page_hash.cell_get(id.fold())) == b); }
ut_ad(UT_LIST_GET_LEN(flush_list) == n_flushing);
mysql_mutex_unlock(&flush_list_mutex); ut_ad(n_lru + n_free <= n_blocks + n_zip); ut_ad(UT_LIST_GET_LEN(LRU) >= n_lru); ut_ad(UT_LIST_GET_LEN(free) <= n_free); ut_ad(size_in_bytes != size_in_bytes_requested || UT_LIST_GET_LEN(free) == n_free);
mysql_mutex_unlock(&mutex);
ut_d(buf_LRU_validate()); ut_d(buf_flush_validate());}#endif /* UNIV_DEBUG */
#if defined UNIV_DEBUG_PRINT || defined UNIV_DEBUG
/** Write information of the buf_pool to the error log. */void buf_pool_t::print() noexcept{ index_id_t* index_ids; ulint* counts; ulint i; index_id_t id; ulint n_found; dict_index_t* index;
mysql_mutex_lock(&mutex);
index_ids = static_cast<index_id_t*>( ut_malloc_nokey(n_blocks * sizeof *index_ids));
counts = static_cast<ulint*>( ut_malloc_nokey(sizeof(ulint) * n_blocks));
mysql_mutex_lock(&flush_list_mutex);
ib::info() << "[buffer pool: size=" << n_blocks << ", database pages=" << UT_LIST_GET_LEN(LRU) << ", free pages=" << UT_LIST_GET_LEN(free) << ", modified database pages=" << UT_LIST_GET_LEN(flush_list) << ", n pending decompressions=" << n_pend_unzip << ", n pending flush LRU=" << n_flush() << " list=" << os_aio_pending_writes() << ", pages made young=" << stat.n_pages_made_young << ", not young=" << stat.n_pages_not_made_young << ", pages read=" << stat.n_pages_read << ", created=" << stat.n_pages_created << ", written=" << stat.n_pages_written << "]";
mysql_mutex_unlock(&flush_list_mutex);
/* Count the number of blocks belonging to each index in the buffer */
n_found = 0;
for (size_t i = 0; i < n_blocks; i++) { buf_block_t* block = get_nth_page(i); const buf_frame_t* frame = block->page.frame; ut_ad(frame == block->frame_address());
if (fil_page_index_page_check(frame)) {
id = btr_page_get_index_id(frame);
/* Look for the id in the index_ids array */ for (ulint j = 0; j < n_found; j++) { if (index_ids[j] == id) { counts[j]++; goto found; } }
index_ids[n_found] = id; counts[n_found] = 1; n_found++;found: continue; } }
mysql_mutex_unlock(&mutex);
for (i = 0; i < n_found; i++) { index = dict_index_get_if_in_cache(index_ids[i]);
if (!index) { ib::info() << "Block count for index " << index_ids[i] << " in buffer is about " << counts[i]; } else { ib::info() << "Block count for index " << index_ids[i] << " in buffer is about " << counts[i] << ", index " << index->name << " of table " << index->table->name; } }
ut_free(index_ids); ut_free(counts);
validate();}#endif /* UNIV_DEBUG_PRINT || UNIV_DEBUG */
#ifdef UNIV_DEBUG
/** @return the number of latched pages in the buffer pool */ulint buf_get_latched_pages_number() noexcept{ ulint fixed_pages_number= 0;
mysql_mutex_assert_owner(&buf_pool.mutex);
for (buf_page_t *b= UT_LIST_GET_FIRST(buf_pool.LRU); b; b= UT_LIST_GET_NEXT(LRU, b)) if (b->state() > buf_page_t::UNFIXED) fixed_pages_number++;
return fixed_pages_number;}#endif /* UNIV_DEBUG */
void buf_pool_t::get_info(buf_pool_info_t *pool_info) noexcept{ mysql_mutex_lock(&mutex); pool_info->pool_size= curr_size(); pool_info->lru_len= UT_LIST_GET_LEN(LRU); pool_info->old_lru_len= LRU_old_len; pool_info->free_list_len= UT_LIST_GET_LEN(free);
mysql_mutex_lock(&flush_list_mutex); pool_info->flush_list_len= UT_LIST_GET_LEN(flush_list); pool_info->n_pend_unzip= UT_LIST_GET_LEN(unzip_LRU); pool_info->n_pend_reads= os_aio_pending_reads_approx(); pool_info->n_pending_flush_lru= n_flush(); pool_info->n_pending_flush_list= os_aio_pending_writes(); mysql_mutex_unlock(&flush_list_mutex);
double elapsed= 0.001 + difftime(time(nullptr), last_printout_time);
pool_info->n_pages_made_young= stat.n_pages_made_young; pool_info->page_made_young_rate= double(stat.n_pages_made_young - old_stat.n_pages_made_young) / elapsed; pool_info->n_pages_not_made_young= stat.n_pages_not_made_young; pool_info->page_not_made_young_rate= double(stat.n_pages_not_made_young - old_stat.n_pages_not_made_young) / elapsed; pool_info->n_pages_read= stat.n_pages_read; pool_info->pages_read_rate= double(stat.n_pages_read - old_stat.n_pages_read) / elapsed; pool_info->n_pages_created= stat.n_pages_created; pool_info->pages_created_rate= double(stat.n_pages_created - old_stat.n_pages_created) / elapsed; pool_info->n_pages_written= stat.n_pages_written; pool_info->pages_written_rate= double(stat.n_pages_written - old_stat.n_pages_written) / elapsed; pool_info->n_page_gets= stat.n_page_gets; pool_info->n_page_get_delta= stat.n_page_gets - old_stat.n_page_gets; if (pool_info->n_page_get_delta) { pool_info->page_read_delta= stat.n_pages_read - old_stat.n_pages_read; pool_info->young_making_delta= stat.n_pages_made_young - old_stat.n_pages_made_young; pool_info->not_young_making_delta= stat.n_pages_not_made_young - old_stat.n_pages_not_made_young; } pool_info->n_ra_pages_read_rnd= stat.n_ra_pages_read_rnd; pool_info->pages_readahead_rnd_rate= double(stat.n_ra_pages_read_rnd - old_stat.n_ra_pages_read_rnd) / elapsed; pool_info->n_ra_pages_read= stat.n_ra_pages_read; pool_info->pages_readahead_rate= double(stat.n_ra_pages_read - old_stat.n_ra_pages_read) / elapsed; pool_info->n_ra_pages_evicted= stat.n_ra_pages_evicted; pool_info->pages_evicted_rate= double(stat.n_ra_pages_evicted - old_stat.n_ra_pages_evicted) / elapsed; pool_info->unzip_lru_len= UT_LIST_GET_LEN(unzip_LRU); pool_info->io_sum= buf_LRU_stat_sum.io; pool_info->io_cur= buf_LRU_stat_cur.io; pool_info->unzip_sum= buf_LRU_stat_sum.unzip; pool_info->unzip_cur= buf_LRU_stat_cur.unzip; buf_refresh_io_stats(); mysql_mutex_unlock(&mutex);}
/*********************************************************************//**
Prints info of the buffer i/o. */staticvoidbuf_print_io_instance(/*==================*/ buf_pool_info_t*pool_info, /*!< in: buffer pool info */ FILE* file) /*!< in/out: buffer where to print */{ ut_ad(pool_info);
fprintf(file, "Buffer pool size " ULINTPF "\n" "Free buffers " ULINTPF "\n" "Database pages " ULINTPF "\n" "Old database pages " ULINTPF "\n" "Modified db pages " ULINTPF "\n" "Percent of dirty pages(LRU & free pages): %.3f\n" "Max dirty pages percent: %.3f\n" "Pending reads " ULINTPF "\n" "Pending writes: LRU " ULINTPF ", flush list " ULINTPF "\n", pool_info->pool_size, pool_info->free_list_len, pool_info->lru_len, pool_info->old_lru_len, pool_info->flush_list_len, static_cast<double>(pool_info->flush_list_len) / (static_cast<double>(pool_info->lru_len + pool_info->free_list_len) + 1.0) * 100.0, srv_max_buf_pool_modified_pct, pool_info->n_pend_reads, pool_info->n_pending_flush_lru, pool_info->n_pending_flush_list);
fprintf(file, "Pages made young " ULINTPF ", not young " ULINTPF "\n" "%.2f youngs/s, %.2f non-youngs/s\n" "Pages read " ULINTPF ", created " ULINTPF ", written " ULINTPF "\n" "%.2f reads/s, %.2f creates/s, %.2f writes/s\n", pool_info->n_pages_made_young, pool_info->n_pages_not_made_young, pool_info->page_made_young_rate, pool_info->page_not_made_young_rate, pool_info->n_pages_read, pool_info->n_pages_created, pool_info->n_pages_written, pool_info->pages_read_rate, pool_info->pages_created_rate, pool_info->pages_written_rate);
if (pool_info->n_page_get_delta) { double hit_rate = static_cast<double>( pool_info->page_read_delta) / static_cast<double>(pool_info->n_page_get_delta);
if (hit_rate > 1) { hit_rate = 1; }
fprintf(file, "Buffer pool hit rate " ULINTPF " / 1000," " young-making rate " ULINTPF " / 1000 not " ULINTPF " / 1000\n", ulint(1000 * (1 - hit_rate)), ulint(1000 * double(pool_info->young_making_delta) / double(pool_info->n_page_get_delta)), ulint(1000 * double(pool_info->not_young_making_delta) / double(pool_info->n_page_get_delta))); } else { fputs("No buffer pool page gets since the last printout\n", file); }
/* Statistics about read ahead algorithm */ fprintf(file, "Pages read ahead %.2f/s," " evicted without access %.2f/s," " Random read ahead %.2f/s\n",
pool_info->pages_readahead_rate, pool_info->pages_evicted_rate, pool_info->pages_readahead_rnd_rate);
/* Print some values to help us with visualizing what is
happening with LRU eviction. */ fprintf(file, "LRU len: " ULINTPF ", unzip_LRU len: " ULINTPF "\n" "I/O sum[" ULINTPF "]:cur[" ULINTPF "], " "unzip sum[" ULINTPF "]:cur[" ULINTPF "]\n", pool_info->lru_len, pool_info->unzip_lru_len, pool_info->io_sum, pool_info->io_cur, pool_info->unzip_sum, pool_info->unzip_cur);}
/*********************************************************************//**
Prints info of the buffer i/o. */voidbuf_print_io(/*=========*/ FILE* file) /*!< in/out: buffer where to print */{ buf_pool_info_t pool_info;
buf_pool.get_info(&pool_info); buf_print_io_instance(&pool_info, file);}
/** Verify that post encryption checksum match with the calculated checksum.
This function should be called only if tablespace contains crypt data metadata.@param page page frame@param fsp_flags contents of FSP_SPACE_FLAGS@return whether the page is encrypted and valid */bool buf_page_verify_crypt_checksum(const byte *page, uint32_t fsp_flags) noexcept{ if (!fil_space_t::full_crc32(fsp_flags)) { return fil_space_verify_crypt_checksum( page, fil_space_t::zip_size(fsp_flags)); }
return !buf_page_is_corrupted(true, page, fsp_flags);}
/** Print the given page_id_t object.
@param[in,out] out the output stream@param[in] page_id the page_id_t object to be printed@return the output stream */std::ostream& operator<<(std::ostream &out, const page_id_t page_id){ out << "[page id: space=" << page_id.space() << ", page number=" << page_id.page_no() << "]"; return out;}#endif /* !UNIV_INNOCHECKSUM */
|
