mariadb

Tree: 2b11a0e991

Author	SHA1	Message	Date
Sergei Golubchik	2b11a0e991	MDEV-37268 assert upon UPDATE or partitioned table with unique hash under READ-COMMITTED followup for `9703c90712` (MDEV-37199 UNIQUE KEY USING HASH accepting duplicate records) maintain the invariant, that handler::ha_update_row() is always invoked as handler::ha_update_row(record[0], record[1])	3 months ago
Sergei Golubchik	b96b5a6ccf	cleanup: ha_partition::m_rec0 it makes no sense to safe this->table->record[0] in this->m_rec0, just use table->record[0] (also, future-proof, if table->record[0] changes)	3 months ago
Sergei Golubchik	774039e410	MDEV-37268 ER_DUP_ENTRY upon REPLACE into table with unique hash under READ-COMMITTED followup for `9703c90712` (MDEV-37199 UNIQUE KEY USING HASH accepting duplicate records) when looking for long unique duplicates and the new row is already inserted, we cannot simply "skip one conflict" we must skip exactly the new row and find a conflict which isn't a new row - otherwise table->file->dup_ref can be set incorrectly and REPLACE won't work.	3 months ago
Sergei Golubchik	3a2e1f87a1	MDEV-37268 ER_NOT_KEYFILE or assertion failure upon REPLACE into table with unique hash under READ-COMMITTED followup for `9703c90712` (MDEV-37199 UNIQUE KEY USING HASH accepting duplicate records) don't forget to rnd_init()/rnd_end() around rnd_pos()	3 months ago
Raghunandan Bhat	9412cd0e62	MDEV-35330: Assertion `marked_for_read()` failed in VSec9::VSec9 \| Item_func_from_unixtime::get_date Problem: When a parameter in a prepared UPDATE statement uses DEFAULT value for a column (e.g., `EXECUTE IMMEDIATE 'UPDATE t SET b=?' USING DEFAULT`), and that column's default is derived from an expression or a function referencing another column (e.g., `FROM_UNIXTIME(a)`), the server fails to correctly mark the dependent column (`a`) for reading. This happened because the server failed to associate the `Item_param` (representing the `?` parameter) with its target field (`b`), leaving `Item_param::m_associated_field` uninitialized. This prevented the correct code path for evaluating the `DEFAULT` value, leading to the dependent column not being marked for read. When the column's value was later accessed, an assertion failed, causing a crash. Fix: In the prepare stage of the UPDATE statement, associate each value with the target field, causing the server to take the correct code path (`Item_param::assign_default`) for evaluating DEFAULT values and marking dependent column for reading beforehand. Only tests are included as it is a duplicate of MDEV-36870.	3 months ago
Julius Goryavsky	008145b968	galera: changes for transition to galera library 26.4.23	3 months ago
Julius Goryavsky	1681b6c330	MDEV-37257: unstable tests temporarily added to 'disabled' list	3 months ago
Sergei Golubchik	bfcd2674a3	MDEV-37199 disable --view-protocol creating and dropping views doesn't work very well with transactions	3 months ago
Thirunarayanan Balathandayuthapani	626d5bf832	MDEV-36287 mariabackup ignores tables-file Problem: ======== - Mariabackup ignores tables-file option because it fails to register the new entry in database hash cells. This issue was caused by the commit `3c312d247c` (MDEV-35190). xb_register_filter_entry() fails to stop when it encounters the empty list. Solution: ========= xb_register_filter_entry(): If there is no next member to deference then it return pointer to existing element.	3 months ago
Sergei Golubchik	9703c90712	MDEV-37199 UNIQUE KEY USING HASH accepting duplicate records Server-level UNIQUE constraints (namely, WITHOUT OVERLAPS and USING HASH) only worked with InnoDB in REPEATABLE READ isolation mode, when the constraint was checked first and then the row was inserted or updated. Gap locks prevented race conditions when a concurrent connection could've also checked the constraint and inserted/updated a row at the same time. In READ COMMITTED there are no gap locks. To avoid race conditions, we now check the constraint after the row operation. This is enabled by the HA_CHECK_UNIQUE_AFTER_WRITE table flag that InnoDB sets in the READ COMMITTED transactions. Checking the constraint after the row operation is more complex. First, the constraint will see the current (inserted/updated) row, and needs to skip it. Second, IGNORE operations become tricky, as we need to revert the insert/update and continue statement execution. write_row() (INSERT IGNORE) is reverted with delete_row(). Conveniently it deletes the current row, that is, the last inserted row. update_row(a,b) (UPDATE IGNORE) is reverted with a reversed update, update_row(b,a). Conveniently, it updates the current row too. Except in InnoDB when the PK is updated - in this case InnoDB internally performs delete+insert, but does not move the cursor, so the "current" row is the deleted one and the reverse update doesn't work. This combination now throws an "unsupported" error and will be fixed in MDEV-37233	4 months ago
Sergei Golubchik	2746c19a9c	MDEV-37203 UBSAN: applying zero offset to null pointer in strings/ctype-uca.inl \| my_uca_strnncollsp_onelevel_utf8mb4 \| handler::check_duplicate_long_entries_update	4 months ago
Sergei Golubchik	d8c2362912	cleanup: long unique checks consolidate and unify long unique checks. fix a bug where an update of a long unique blob was ignoring the prefix length	4 months ago
Rucha Deodhar	dc9bdb4216	MDEV-21530: json_extract STILL crashes in Item_func_json_extract::read_json Analysis: Wrong comparison function is used, resulting in wrong argument to the read_json() function. Fix: Used correct function.	7 months ago
Marko Mäkelä	024c7e881f	MDEV-37103 innodb_immediate_scrub_data_uncompressed=ON may break innodb_undo_log_truncate=ON The test innodb.undo_truncate occasionally demonstrates a race condition where scrubbing is writing zeroes to a freed undo page, and innodb_undo_log_truncate=ON truncating the same tablespace. The truncation is an exception to the rule that InnoDB tablespace file sizes can only grow, never shrink. The fields fil_space_t::size and fil_node_t::size are protected by fil_system.mutex, which used to be a highly contended resource. We do not want to revert back to acquiring the mutex in fil_space_t::io() because that would introduce an obvious scalability bottleneck. fil_space_t::flush_freed(): Do not try to scrub pages of the undo tablespace in order to prevent a race condition between io() and undo tablespace truncation. fil_space_t::io(): Prevent a null pointer dereference when reporting an out-of-bounds access to the non-first file of the system or temporary tablespace. Do not invoke set_corrupted() after an out-of-bounds asynchronous read. Note: fil_space_t::flush_freed() may only invoke PUNCH_RANGE on page_compressed tablespaces, never on an undo tablespace.	3 months ago
Marko Mäkelä	e3c5565dfb	MDEV-36330 fixup: Only fix innodb_snapsho_isolation=ON ha_innobase::store_lock(): Do not create a read view or start the transaction if innodb_snapshot_isolation=OFF. This should save some resources with the default settings.	4 months ago
Thirunarayanan Balathandayuthapani	3bcfc2ed0a	MDEV-22250 InnoDB: Failing assertion: opt_no_lock during mariabackup --backup backup_file_op_fail(): Ignore the FTS internal table if it is being created in late phase of backup. mariabackup --prepare should be handle intermediate table and orphaned fts internal table. check_if_fts_table(): Determine whether the space name belongs to internal FTS table	4 months ago
Marko Mäkelä	b7b2e009b3	MDEV-37215 SELECT FOR UPDATE crash in SERIALIZABLE ha_innobase::store_lock(): Set also trx->will_lock when starting a transaction at SERIALIZABLE isolation level. This fixes up commit `7fbbbc983f` (MDEV-36330).	4 months ago
Marko Mäkelä	499fa24d63	MDEV-27058 fixup: Fix a bogus assertion buf_page_get_low(): Do not expect a valid state of buf_page_t::in_zip_hash for blocks that are not file pages. This debug assertion had been misplaced in commit `aaef2e1d8c` (MDEV-27058) that removed the condition block->page.state() == BUF_BLOCK_FILE_PAGE.	4 months ago
Yuchen Pei	ea962ca495	MDEV-30436 [fixup] Add missing check for HAVE_PSI_INTERFACE A fixup of `3e9aa07cce`, with thanks to Daniel Black.	4 months ago
Yuchen Pei	998e765060	MDEV-32907 Spider: do not create gbh if encountering Item_aggregate_ref When spider encounters Item_aggregate_ref that is valid at gbh creation, it could become invalid at gbh execution due to item list substitution. Therefore we ban Item_aggregate_ref for spider gbh creation. To that end, we also make sure that str is NULL if and only if in the creation stage, not the execution stage, including removing a redundant check when str is not NULL. : Note that it is likely the same scenario as in MDEV-25116.	4 months ago
Yuchen Pei	3e9aa07cce	MDEV-30436 Spider: deduplicate some sts/crd code. We spare spider_table_remove_share_from_crd/sts_thread and spider_table_add_share_to_crd/sts_thread because the function body has too many fields with crd/sts in the name and macro affects debuggability.	5 months ago
Yuchen Pei	a3aab082ff	MDEV-27474 Spider: remove #WITH_PARTITION_STORAGE_ENGINE	5 months ago
Yuchen Pei	c78e906ed5	MDEV-30264 Remove unused method spider_db_result::fetch_row_from_tmp_table	5 months ago
Monty	4d19e55441	MDEV-36858 MariaDB MyISAM secondary indexes silently break for tables > 10B rows The problem was a bug in my_qsort from 1980 where we converted difference of pointers to int instead of longlong. This causes sort to fail if the size of the sort array > 2147483648 bytes. In case of myisam recovery, we are sorting pointers, which limited the max number of elements to 268435456. In the case of a longlong and a 6 byte pointers, this equals to a sort buffer of about 5G. Fixed by changing the cast to longlong.	4 months ago
Marko Mäkelä	7fbbbc983f	MDEV-36330: SERIALIZABLE read inconsistency At TRANSACTION ISOLATION LEVEL SERIALIZABLE, InnoDB would fail to flag a write/read conflict, which would be a violation already at the more relaxed REPEATABLE READ level when innodb_snapshot_isolation=ON. Fix: Create a read view and start the transaction at the same time. Thus, lock checks will be able to consult the correct read view to flag ER_CHECKREAD if we are about to lock a record that was committed after the start of our transaction. innobase_start_trx_and_assign_read_view(): At any other isolation level than READ UNCOMMITTED, do create a read view. This is needed for the correct operation of START TRANSACTION WITH CONSISTENT SNAPSHOT. ha_innobase::store_lock(): At SERIALIZABLE isolation level, if the transaction was not started yet, start it and open a read view. An alternative way to achieve this would be to make trans_begin() treat START TRANSACTION (or BEGIN) in the same way as START TRANSACTION WITH CONSISTENT SNAPSHOT when the isolation level is SERIALIZABLE. innodb_isolation_level(const THD*): A simpler version of innobase_map_isolation_level(). Compared to earlier, we will return READ UNCOMMITTED also if the :newraw option is set for the InnoDB system tablespace. Reviewed by: Vladislav Lesin	4 months ago
Marko Mäkelä	f73ffd1150	MDEV-37183 Scrubbing empty record breaks recovery page_delete_rec_list_end(): Do not attempt to scrub the data of an empty record. The test case would reproduce a debug assertion failure in branches where commit `358921ce32` (MDEV-26938) is present. MariaDB Server 10.6 only supports ascending indexes, and in those, the empty string would always be sorted first, never last in a page. Nevertheless, we fix the bug also in 10.6, in case it would be reproducible in a slightly different scenario. Reviewed by: Thirunarayanan Balathandayuthapani	4 months ago
Sergei Golubchik	3b140fed0d	bump the VERSION	4 months ago
Sergei Golubchik	c27d78beb5	MDEV-36870 Spurious unrelated permission error when selecting from table with default that uses nextval(sequence) Lots of different cases, SELECT, SELECT DEFAULT(), UPDATE t SET x=DEFAULT, prepares statements, opening of a table for the I_S, prelocking (so TL_WRITE), insert with subquery (so SQLCOM_SELECT), etc. Don't check NEXTVAL privileges in fix_fields() anymore, it cannot possibly handle all the cases correctly. Make a special method Item_func_nextval::check_access() for that and invoke it from * fix_fields on explicit SELECT NEXTVAL() (but not if NEXTVAL() is used in a DEFAULT clause) * when DEFAULT bareword in used in, say, UPDATE t SET x=DEFAULT (but not if DEFAULT() itself is used in a DEFAULT clause) * in CREATE TABLE * in ALTER TABLE ALGORITHM=INPLACE (that doesn't go CREATE TABLE path) * on INSERT helpers * Virtual_column_info::check_access() to walk the item tree and invoke Item::check_access() * TABLE::check_sequence_privileges() to iterate default expressions and invoke Virtual_column_info::check_access() also, single-table UPDATE in prepared statements now associates value items with fields just as multi-update already did, fixes the case of PREPARE s "UPDATE t SET x=?"; EXECUTE s USING DEFAULT.	4 months ago
Sergei Golubchik	1c7685f5fc	bugfix: nextval() in default, and UPDATE SET x=DEFAULT set thd->lex->default_used accordingly	4 months ago
Kristian Nielsen	4c8af2007d	MDEV-36934: semi sync makes the master unresponsive when a replica is stopped Ensure that a pending semi-sync transaction can only be signalled on its THD while it is waiting in commit_trx(); not if the wait in commit_trx() is skipped for some reason. There was a bug that if no semi-sync slaves were connected and rpl_semi_sync_master_wait_no_slave was off, the THD pointer could be left in the list of pending transactions after THD was deleted, and an invalid THD/condition variable could be signalled, causing hang or other corruption. Testcase based on work by Brandon Nesterenko <brandon.nesterenko@mariadb.com>. Reviewed-by: Brandon Nesterenko <brandon.nesterenko@mariadb.com> Signed-off-by: Kristian Nielsen <knielsen@knielsen-hq.org>	4 months ago
Aleksey Midenkov	c4a2688328	MDEV-24726 Assertion on compressed varstring as key field in optimizer temporary table Compressed field cannot be part of a key by its nature: there is no data to order, only the compressed data. For optimizer temporary table we create uncompressed substitute. In all other cases (MDEV-16808) we don't use key: add_keyuse() is skipped by !field->compression_method() condition.	5 months ago
Marko Mäkelä	31aa8b6939	MDEV-37170 Enable AVX10.1 CRC-32 on GCC 16 The AVX512 accelerated CRC-32 computation that had been added in commit `9ec7819c58` was disabled in commit `a293dfd92a` for GCC 16. Let us enable that logic by applying dr-m/crc32_simd@075bacb0cc7da8df18afa15a827ae4d11166fe32 which makes use of the avx10.1 target attribute that had been introduced in GCC 15.	4 months ago
Kostadin Shishmanov	a293dfd92a	Fix building with gcc 16 (evex512 removal) Recently, evex512 was removed from gcc trunk [1] which will eventually become gcc 16, and that leads to a build failure in mariadb, originally reported downstream in a Gentoo bug [2]. This is reproducible across all versions from 10.6 to current master. The change is as simple as adding an upper boundary to which gcc versions can use evex512. [1] https://gcc.gnu.org/git/?p=gcc.git;a=commitdiff;h=c052a6f4 [2] https://bugs.gentoo.org/956632 Signed-off-by: Kostadin Shishmanov <kostadinshishmanov@protonmail.com>	5 months ago
Thirunarayanan Balathandayuthapani	27660ff2e9	MDEV-37121 Change buffer freed pages are not removed during slow shutdown Problem: ======= - During slow shutdown, expectation is that change buffer index to merge all the change buffer pages and free all the change buffer pages except root. But InnoDB fails to free the freed change buffer pages from segment and returns it to system tablespace. Solution: ========= - During slow shutdown, remove all pages from free list of change buffer and frees it to system tablespace	4 months ago
Daniel Black	9059385262	MDEV-37048 revert MSAN my_vsnprintf_ex for double workaround 5 years ago `e843033d02` worked around a MSAN bug when retrieving a double from a va_list. The construct {{__msan_check_mem_is_initialized(ap,size)}} where ap is a va_list is undefined as {{__msan_check_mem_is_initialized}} expects a pointer. The implementation detail of va_list is architecture dependant and on aarch64 this isn't a pointer. The need to any msan action is no unnecessary since this has been corrected in recent clang versions. As such the additions from `e843033d02` have been reverted. Tested with clang-20.1 and the test cases from MDEV-22690 and MDEV-22691.	4 months ago
Thirunarayanan Balathandayuthapani	0dd6566ee4	MDEV-37123 dict_table_open_on_id() fails to release dict_sys.latch While updating the persistent defragmentation statistics for the table, InnoDB opens the table only if it is in cache. If dict_table_open_on_id() fails to find the table in cache then it fails to unfreeze dict_sys.latch. This lead to crash	4 months ago
Monty	2d5dfc47a9	Define error message for HA_ERR_INCOMPATIBLE_DEFINITION	4 months ago
Jan Lindström	56fbc0cdd7	MDEV-36953 : mysql-wsrep#198 test hangs Test changes only. INSERT may fail for lock wait because in other node we have LOCK TABLE or it may succeed if next statement i.e. UNLOCK TABLES is fast enough. Signed-off-by: Julius Goryavsky <julius.goryavsky@mariadb.com>	5 months ago
Jan Lindström	f495460689	MDEV-36968 : galera_3nodes.inconsistency_shutdown test occasionally hangs Test changes only. Remove unnecessary wsrep_on=[ON\|OFF] and use sync wait instead. Signed-off-by: Julius Goryavsky <julius.goryavsky@mariadb.com>	5 months ago
Jan Lindström	fd1266a980	MDEV-34761 : Assertion `client_state_.mode() == wsrep::client_state::m_local' failed in int wsrep::transaction::after_statement(wsrep::unique_lock<wsrep::mutex>&) @@enforce_storage_engine is local setting and there is no knowledge how other nodes are configured. Statement CREATE TABLE xxx ENGINE=yyy is replicated as it is and if required engine != enforced engine it could lead inconsistent used storage engine in the cluster. Fix is to return error and a warning if required engine is not same as enforced engine. Signed-off-by: Julius Goryavsky <julius.goryavsky@mariadb.com>	5 months ago
Jan Lindström	f41acb555d	MDEV-35523 : Server crashes with "WSREP: Unknown writeset version: -1" Cluster configuration was incorrect e.g. wsrep_node_address was missing. Therefore, Galera replication was not properly initialized and TOI is not supported. Fix is to check when user tries to start Galera replication with wsrep_on=ON that Galera replication is properly initialized and node is ready to receive operations. If Galera replication is not properly initialized return a error. Signed-off-by: Julius Goryavsky <julius.goryavsky@mariadb.com>	4 months ago
Monty	c3578720e6	Removed safemalloc warnings from myisamchk --version	4 months ago
Marko Mäkelä	3c67d73aad	MDEV-36482: Make libaio work WITH_MSAN=ON As noted in commit `c36834c832` the MemorySanitizer instrumented builds so far only work with the synchronous I/O interface (innodb_use_native_aio=OFF). It is not that hard to make WITH_MSAN=ON work with the Linux libaio, without even instrumenting that library itself. aio_linux::getevent_thread_routine(): Declare the buffer that is returned by my_getevents() as initialized. Declare the data returned by a successful aio_opcode::AIO_PREAD as initialized. Reviewed by: Daniel Black	4 months ago
Thirunarayanan Balathandayuthapani	e706324205	MDEV-35863 innodb.doublewrite_debug test case fails to start the server Problem: ======= - There are two failures occurs for this test case: (1) set global innodb_buf_flush_list_now=1 doesn't make sure that pages are being flushed. (2) InnoDB page cleaner thread aborts while writing the checkpoint information. Problem is that When InnoDB startup aborts, InnoDB changes the shutdown state to SRV_SHUTDOWN_EXIT_THREADS. By changing the shutdown state, InnoDB doesn't advance the log_sys.lsn (avoids fil_names_clear()). After InnoDB shutdown(innodb_shutdown()) is being initiated, shutdown state again changed to SRV_SHUTDOWN_INITIATED. This leads the page cleaner thread to fail with assertion ut_ad(srv_shutdown_state > SRV_SHUTDOWN_INITIATED) in log_write_checkpoint_info() Solution: ========= (1) In order to avoid (1) failure, InnoDB can make the variable innodb_max_dirty_pages_pct_lwm, innodb_max_dirty_pages_pct to 0. Also make sure that InnoDB doesn't have any dirty pages in buffer pool by adding wait_condition. (2) Avoid changing the srv_shutdown_state to SRV_SHUTDOWN_EXIT_THREADS when the InnoDB startup aborts	4 months ago
Sergei Golubchik	cda1826201	MDEV-36852 Table definition gets corrupt after adding unique hash key	5 months ago
Thirunarayanan Balathandayuthapani	6ec57588bd	MDEV-30363 InnoDB: Failing assertion: trx->error_state == DB_SUCCESS in que_run_threads Problem: ========= - During truncation of a fulltext table, InnoDB does create the table and does insert the default config fts values in fulltext common config table using create table transaction. - Before committing the create table transaction, InnoDB does update the dictionary by loading the stopword into fts cache and write the stopword configuration into fulltext common config table by creating a separate transaction. This leads to lock wait timeout error and rollbacks the transaction. - But truncate table holds dict_sys.lock and rollback also tries to acquire dict_sys.lock. This leads to assertion during rollback. Solution: ========= ha_innobase::truncate(): Commit the create table transaction before updating the dictionary after create table.	5 months ago
Vladislav Vaintroub	888663ce12	MDEV-36280 ALTER TABLE with DEFAULT NEXTVAL(sequence) fails due to insufficient grants Defer privilege checking until fix_fields. This way ALTER will behave consistently with CREATE, and require the same privileges to sequence column (SELECT/INSERT)	5 months ago
Monty	643319a7fb	MDEV-36465 MDEV-33813 Regression, Queries in 'Waiting for someone to free space' state will not automatically retry IO and hang forever MDEV-33813 caused a regressing in that when a disk got full when writing to a MyISAM or Aria table the MariaDB connection would, instead of doing a retry after 60 seconds, hang until the query was killed. Fixed by changing mysql_coind_wait() top mysql_cond_timedwait() Author: Thomas Stangner	5 months ago
Monty	0a91bbdc41	Get debug version to compile with gcc 7.5.0	5 months ago
Monty	ce4f83e6b9	MDEV-29157 SELECT using ror_merged scan fails with s3 tables handler::clone() call did not work with read only tables like S3. It gave a wrong error message (out of memory instead of a permission error) and aborted the query. The issue was that the clone call had a wrong parameter to ha_open(). This now fixed. I also changed the clone call to provide the correct error message if things fails. This patch fixes an 'out of memory' error when using the S3 engine for queries that could use multiple indexes together to find the matching rows, like the following: SELECT * FROM t1 WHERE key1 = 99 OR key2 = 2	5 months ago

1 2 3 4 5 ...

199553 Commits (2b11a0e9918ac6b04d821fdbe4192475f516946a) All Branches Search

199553 Commits (2b11a0e9918ac6b04d821fdbe4192475f516946a)

All Branches