mariadb

Commit Graph

Author	SHA1	Message	Date
Daniel Black	1055bc957e	MDEV-36182: Revert "MDEV-26674: Set innodb_use_native_aio=OFF ...when using io_uring on a potentially affected kernel" Remove version check on the kernel as it now corresponds to a working RHEL9 kernel and the problem was only there in pre-release kernels that shouldn't have been used in production. This reverts commit `1193a793c4`.	6 months ago
Daniel Black	51179067fc	MDEV-36182: Revert "MDEV-26674 workaround for mariadb-backup" Remove version check on the kernel as it now corresponds to a working RHEL9 kernel and the problem was only there in pre-release kernels that shouldn't have been used in production. This reverts commit `3dc0d884ec`.	6 months ago
Marko Mäkelä	1c9f64e54f	MDEV-36613 Incorrect undo logging for indexes on virtual columns Starting with mysql/mysql-server@02f8eaa9988dadb73dd68630dd82393cfa19bfb8 and commit `2e814d4702` the index ID of indexes on virtual columns was being encoded insufficiently in InnoDB undo log records. Only the least significant 32 bits were being written. This could lead to some corruption of the affected indexes on ROLLBACK, as well as to missed chances to remove some history from such indexes when purging the history of committed transactions that included DELETE or an UPDATE in the indexes. dict_hdr_create(): In debug instrumented builds, initialize the DICT_HDR_INDEX_ID close to the 32-bit barrier, instead of initializing it to DICT_HDR_FIRST_ID (10). This will allow the changed code to be exercised while running ./mtr --suite=gcol,vcol. trx_undo_log_v_idx(): Encode large index->id in a similar way as mysql/mysql-server@e00328b4d068c7485ac2ffe27207ed1f462c718d but using a different implementation. trx_undo_read_v_idx_low(): Decode large index->id in a similar way as mach_u64_read_much_compressed(). Reviewed by: Debarun Banerjee	6 months ago
Jan Lindström	5f2562291c	MDEV-36509 : Galera test failure on galera_sr.mysql-wsrep-features#165 Problem was that thread was holding lock_sys.wait_mutex when streaming replication transaction rollback was handled and in wsrep-lib requests THD::LOCK_thd_kill mutex causing wrong mutex usage (thd->reset_globals()). Fix is to remove streaming replication rollback handling from Deadlock::report() i.e. wsrep_handle_SR_rollback call. Purpose of Deadloc::report() is to find a cycle in the waits-for graph if exists, report it, mark victim transaction as deadlock victim and release locks it is waiting for. Actual streaming replication rollback that can take longer time can be handled later at trx_t::rollback where lock_sys.wait_mutex is not held. Signed-off-by: Julius Goryavsky <julius.goryavsky@mariadb.com>	6 months ago
Jan Lindström	ee947fae80	MDEV-36464 : Galera test failure on galera_3nodes.galera_gtid_2_cluster Test changes only. Add wait conditions after INSERT-clauses to make sure that they are replicated before checking gtid position or table contents. Signed-off-by: Julius Goryavsky <julius.goryavsky@mariadb.com>	6 months ago
Julius Goryavsky	0403f0147f	MDEV-33136: backport corrections from 10.11+	6 months ago
Julius Goryavsky	88dfa6bcee	Merge branch '10.5' into '10.6'	6 months ago
Julius Goryavsky	08f902e9ce	MDEV-36116: correction for error codes	6 months ago
Alexey Yurchenko	77391482bd	MDEV-34998: master can stop responding after cluster vote to evict a node Additional corrections: there is a natural race between closing connection to cluster in case of applying error and finishing the IST and sometimes IST finishes and tries to send JOIN message over a closed connection. This does not affect the correctness of the test or node behavior. Added error message suppression. Signed-off-by: Julius Goryavsky <julius.goryavsky@mariadb.com>	6 months ago
Julius Goryavsky	bbd0e4b2c9	MDEV-34998 addendum: post-fix corrections for SST scripts Delaying scripts on joiner after SST/IST has been made a common debug feature for all suitable SST/IST methods. Also some minor fixes have been made for new tests.	6 months ago
Alexey Yurchenko	ec5068fe59	MDEV-34998: master can stop responding after cluster vote to evict a node After cluster vote to evict a node that failed a transaction, current master can't commit anymore. Error voting for joiner in the JOINED state was broken because the group-wide commit cut (implicit SUCCESS vote) was not taken into account when processing error vote request from the JOINED node. This commit adds 3 MTR tests to verify the fix in the galera library works as designed. Requires Galera library commit `91f0090a05` (galera library version > 26.4.19). Signed-off-by: Julius Goryavsky <julius.goryavsky@mariadb.com>	1 year ago
Nikita Malyavin	cb7e39b75b	MDEV-36181 Field pointer may be uninitialized in fill_record Newer gcc reports: error: 'rfield' may be used uninitialized [-Werror=maybe-uninitialized] 9041 \| unwind_stored_field_offsets(fields, rfield); After investigation, it turned to be an impossible case: 1. The only way it could be broken is if if (!(field= fld->field_for_view_update())) line case would succeed from the first time. 2. Consequent checks initialize rfield. fld may return NULL in field_for_view_update() only for views. 3. Before fill_record, UPDATE first calls check_fields, where field_for_view_update() result is already checked. INSERT calls check_view_insertability that checks that all view fields are updateable. It all means that field_for_view_update() cannot be NULL in fill_record, so the if can be converted to DBUG_ASSERT. This essentially shifts the responsibility on preliminary field_for_view_update() check to the caller. In this patch: 1. convert field_for_view_update() check to DBUG_ASSERT 2. harden unwind_stored_field_offsets function so that it can be used even if field_for_view_update() is NULL 3. As a consequence, `field` is passed instead of `rfield` as a terminator. 4. Initialize `field` to NULL to bypass a false-positive warning!	6 months ago
Oleksandr Byelkin	ba34657cd2	MDEV-35238 (MDEV-34922) Wrong results from a tables with a single record and an aggregate The problem is that copy function was used in field list but never copied in this execution path. So copy should be performed before returning result. Protection against uninitialized copy usage added.	9 months ago
Marko Mäkelä	14389b6993	MDEV-36510 InnoDB fails to compile with clang++-20 ut_allocator(PSI_memory_key): Remove the explicit keyword, to allow implicit instantiation of purge_sys_t::unordered_map.	6 months ago
Nikita Malyavin	e6ea5d568c	MDEV-36507 fix dbug_print_row concurrent access `7544fd4cae` had to make use of a static array to avoid memory use-after-free or leak. Instead, let us make a function returning String, this is the only way to automatically manage the memory after the function returned. To make it all correct, move constructor is added. Normally, it is expected, that the constructor will be elided upon return of an object by value, but if something goes different, or -fno-elide-constructors is used, we can have a problem. So this was a move constructor avoids copy elision-related UB. dbug_print_row returning char* is still there for convenient use in a debugger.	7 months ago
Nikita Malyavin	4de5161a70	fix build with -fno-elide-constructors Linker is trying to find a copy constructor to injector::transaction::transaction ld.lld: error: undefined symbol: injector::transaction::transaction (injector::transaction const&) >>> referenced by rpl_injector.cc:164 >>> rpl_injector.cc.o:(injector::new_trans(THD*)) This constructor is declared, but is not implemented. Ok if copy elision is enabled, but causes error otherwise. Remove the constructor declaration, so that operator= will be used in reality.	6 months ago
Nikita Malyavin	24a4d4ab0c	MDEV-36422 Build fails with cmake 4.0.0 due to wsrep update wsrep-lib	6 months ago
Nikita Malyavin	2a5a12b227	MDEV-36506 Build fails with cmake 4.0 Update cmake_minimum_required to 2.8...3.12 in root cmake and mroonga. This will update "Policy Version" to 3.12, which will not prevent the build by even higher cmake versions. There is also a reason to stay on the compatible with windows "policy version", so 3.12 is conservatively chosen. On the other hand, it will require at least version 2.8.	6 months ago
Yuchen Pei	b6392c292e	MDEV-36454 MDEV-35452 Fix spider view protocol test failures caused by tampering of order by items With view protocol, a SELECT statement is transformed into two statements: 1. CREATE OR REPLACE VIEW mysqltest_tmp_v AS SELECT ... 2. SELECT * FROM mysqltest_tmp_v The first statement reconstructed the query, which is executed in the second statement. The reconstruction often replaces aliases in ORDER BY by the original item. For example, in the test spider/bugfix.mdev_29008 the query SELECT MIN(t2.a) AS f1, t1.b AS f2 FROM tbl_a AS t1 JOIN tbl_a AS t2 GROUP BY f2 ORDER BY f1, f2; is transformed to "select min(`t2`.`a`) AS `f1`,`t1`.`b` AS `f2` from (`auto_test_local`.`tbl_a` `t1` join `auto_test_local`.`tbl_a` `t2`) group by `t1`.`b` order by min(`t2`.`a`),`t1`.`b`" In such cases, spider constructs different queries to execute at the data node. So we disable view protocol for such queries.	6 months ago
Yuchen Pei	8866fba00a	MDEV-36454 MDEV-35452 Fix spider view protocol test failures caused by different queries constructed without GBH With view protocol, often during optimization, the GBH is not created because join->tables_list is the view mysqltest_tmp_v which has MEMORY as engine which does not have GBH implemented. In such cases, if without view protocol the test takes a path that does create a spider GBH, the resulting queries sent to the data node often differ. Therefore we disable view protocol for these statements.	6 months ago
Yuchen Pei	dabd51c391	MDEV-36335 MDEV-35452 Disable view protocol when calling spider_copy_tables in tests Spider needs to lock the spider table when executing the udf, but the server layer would have already locked tables in view protocol because it transforms the query: select spider_copy_table('t', 0, 1) to two queries create or replace view mysqltest_tmp_v as select spider_copy_table('t', 0, 1); select * from mysqltest_tmp_v; So spider justifiably errors out in this case by checking on thd->derived_tables and thd->locks in spider_copy_tables_body()	6 months ago
Yuchen Pei	0f7c9146cf	MDEV-36452 MDEV-35452 Fix udf output temp field name in spider tests with --view-protocol	6 months ago
Yuchen Pei	25f1e6f565	MDEV-36307 MDEV-35452 Do not create spider group by handler when dealing with derived tables or view and at least one select item is constant If one of the selected field is a MIN or MAX and it has been optimized into a constant, it is not added to the temp table used by a group by handler (GBH). The GBH therefore cannot store results to this missing field. On the other hand, when SELECTing from a view or a derived table, TMP_TABLE_ALL_COLUMNS is set. If the query has no group by or order by, an Item_temptable_field is created for this MIN/MAX field and added to the JOIN. Since the GBH could not store results to the corresponding field in the temp table, the value of this Item_temptable_field remains NULL. And the NULL value is passed to the record, then the temp row, and finally output as the (wrong) result. To fix this, we opt to not creating a spider GBH when a view or derived table is involved. This fixes spider/bugfix.mdev_26345 for --view-protocol Also fixed a comment: TABLE_LIST::belong_to_derived is NULL if the table belongs to a derived table that has non-MERGE type.	6 months ago
Yuchen Pei	59962ae2b3	MDEV-36442 MDEV-35452 Fix --view-protocol for spider tests with SELECT arguments of SELECT statements Running mtr --view-protocol transforms SELECT statements to a CREATE OR REPLACE VIEW of the statement, followed by SELECT from the view. When thus when spider tests check the query log for select statements, it often output a different one with --view-protocol compared to without. By adding disable/enable_view_protocol pairs to these statements. Most of these statements are surrounded by existing disable/enable_ps[2]_protocol pairs. Acked-by: Yuchen Pei <ycp@mariadb.com>	6 months ago
Sergey Vojtovich	b02ad4a6f8	MDEV-36427 - FTBFS with libxml2 2.14.0 Connect engine fails to build with libxml2 2.14.0. Connect engine uses "#ifndef BASE_BUFFER_SIZE" to determine if libxml2 is available. If libxml2 is unavailable it did redefine xmlElementType enum of libxml/tree.h. The reasons for this redefinition is vague, most probably some of these constants were used when connect was compiled with MSXML, while libxml2 was disabled. However BASE_BUFFER_SIZE constant was removed from libxml2 recently, as a result connect fails to build due to xmlElementType constants redefinition. Use LIBXML2_SUPPORT instead of BASE_BUFFER_SIZE for libxml2 availability check.	6 months ago
Daniel Black	690b2cf776	test: archive-big test too big for msan archive.archive-big w1 [ fail ] timeout after 900 seconds Test ended at 2025-03-19 22:27:30 Test case timeout after 900 seconds == /build/mysql-test/var/1/log/archive-big.log == CREATE TABLE t1(a BLOB) ENGINE=ARCHIVE; INSERT INTO t1 SELECT * FROM t1; ... == /build/mysql-test/var/1/tmp/analyze-timeout-mysqld.1.err == mysqltest: Could not open connection 'default' after 500 attempts: 2002 Can't connect to local MySQL server through socket '/build/mysql-test/var/tmp/1/mysqld.1.sock' (111)	7 months ago
Daniel Black	93ea4f29a4	MDEV-36347 UBSAN: plugins.auth_v0100 - runtime error: call to function do_auth_0x0100 through pointer to incorrect function type. Redoing a new plugin interface for an obsolete protocol was too much so we just remove the UBSAN testing on the function. It wasn't possible to just disable funtion-type-mismatch	6 months ago
Daniel Black	3995de0318	MDEV-36341 UBSAN: FederatedX fill_server runtime error: applying non-zero offset to null pointer UBSAN is nuanced around null pointer additions, but there is no offset on schema, its the start of the server->key.	6 months ago
Daniel Black	13dd073742	MDEV-31846: enable cursor protocol for test federatedx_create_handlers Per comment it code - it can be removed 10.5+. Not in the comment, is the SELECT INTO ... parts still required cursor protocol exclusion.	6 months ago
Daniel Black	b316a7135b	mroonga: undefined behaviour fix The value of dv[0].data being null showed up in the mtr tests: mroonga/storage.alter_table_fulltext_add_no_primary_key as: /source/storage/mroonga/vendor/groonga/lib/ii.c:2052:37: runtime error: applying non-zero offset 28 to null pointer Correct this by entrying the if condition on null pointer value. The free is valid, and the data of size is allocated.	6 months ago
Daniel Black	2ba49bd804	mroonga: correct offsetof calculation Using a null pointer as a basic for calculation was undefined behaviour.	6 months ago
Marko Mäkelä	db4763a0d1	Fix a slow test When we expect a lock wait timeout, let us override the default innodb_lock_wait_timeout=50 with the minimum timeout of 1 second.	6 months ago
Thirunarayanan Balathandayuthapani	b11772d9a5	MDEV-33167 ASAN errors in dict_sys_t::load_table / get_foreign_key_info after failing to load a table Problem: ======= - While loading the foreign key constraints for the parent table, if child table wasn't open then InnoDB uses the parent table heap to store the child table name in fk_tables list. If the consecutive foreign key relation for the parent table fails with error, InnoDB evicts the parent table from memory. But InnoDB accesses the evicted table memory again in dict_sys.load_table() Solution: ======== dict_load_table_one(): In case of error, remove the child table names which was added during dict_load_foreigns()	6 months ago
Thirunarayanan Balathandayuthapani	0d7ef4f478	MDEV-36236 Instant alter aborts when InnoDB fails to rollback instant operation Problem: ======== - InnoDB does consecutive instant alter operation, first instant DDL fails, it fails to reset the old instant information in table during rollback. This lead to consecutive instant alter to have wrong assumption about the exisitng instant column information. Fix: ==== dict_table_t::instant_column(): Duplicate the instant information field of the table. By doing this, InnoDB alter retains the old instant information and reset it during rollback operation	6 months ago
Marko Mäkelä	e7442e5eb9	MDEV-36226 fixup: format mismatch	6 months ago
Marko Mäkelä	4c0e2f1aca	MDEV-35813: even more robust test case The test in commit `1756b0f37d` is occasionally failing if there are unexpectedly many page cleaner batches that are updating the log checkpoint by small amounts. This occurs in particular when running the server under Valgrind. Let us insert the same number of records with a larger number of statements in a hope that the test would then be more likely to pass.	6 months ago
Denis Protivensky	c01bff4a10	MDEV-36360: Don't grab table-level X locks for applied inserts It prevents a crash in wsrep_report_error() which happened when appliers would run with FK and UK checks disabled and erroneously execute plain inserts as bulk inserts. Moreover, in release builds such a behavior could lead to deadlocks between two applier threads if a thread waiting for a table-level lock was ordered before the lock holder. In that case the lock holder would proceed to commit order and wait forever for the now-blocked other applier thread to commit before. Signed-off-by: Julius Goryavsky <julius.goryavsky@mariadb.com>	6 months ago
Julius Goryavsky	b983a911e9	galera mtr tests: synchronization between branches and editions	6 months ago
Julius Goryavsky	5003dac220	MDEV-36116: Post-merge correction for 10.6+	6 months ago
Julius Goryavsky	03c31ab099	Merge branch '10.5' into '10.6'	6 months ago
Jan Lindström	25737dbab7	MDEV-33850 : For Galera, create sequence with low cache got signal 6 error: [ERROR] WSREP: FSM: no such a transition REPLICATING -> COMMITTED Problem was that transacton was BF-aborted after certification succeeded and transaction tried to rollback and during rollback binlog stmt cache containing sequence value reservations was written into binlog. Transaction must replay because certification succeeded but transaction must not be written into binlog yet, it will be done during commit after the replay. Fix is to skip binlog write if transaction must replay and in replay we need to reset binlog stmt cache. Signed-off-by: Julius Goryavsky <julius.goryavsky@mariadb.com>	7 months ago
Denis Protivensky	dd54ce9e10	MDEV-36116: Remove debug assert in TOI when executing thread is killed Signed-off-by: Julius Goryavsky <julius.goryavsky@mariadb.com>	8 months ago
Daniele Sciascia	d698b784c8	MDEV-35658 Assertion `commit_trx' failed in test galera_as_master The test issues a simple INSERT statement, while sql_log_bin = 0. This option disables writes to binlog. However, since MDEV-7205, the option does not affect Galera, so changes are still replicated. So sql_log_bin=off, "partially" disabled the binlog and the INSERT will involve both binlog and innodb, thus requiring internal 2 phase commit (2PC). In 2PC INSERT is first prepared, which will make it transition to PREPARED state in innodb, and later committed which causes the new assertion from MDEV-24035 to fail. Running the same test with sql_log_bin enabled also results in 2PC, but the execution has one more step for ordered commit, between prepare and commit. Ordered commit causes the transaction state to transition back to TRX_STATE_NOT_STARTED. Thus avoiding the assertion. This patch makes sure that when sql_log_bin=off, the ordered commit step is not skipped, thus going through the expected state transitions in the storage engine. Signed-off-by: Julius Goryavsky <julius.goryavsky@mariadb.com>	10 months ago
Daniele Sciascia	88eec3a9a1	MDEV-36393 Test failure on galera_sr.GCF-572 Fix a regression on test galera_sr.GCF-572 introduced by commit `c9a6adba`. This commit partially reverted a trivial test fix for the galera_sr.GCF-572 test (commit 11ef59fb), which was targeted at 10.6. This is backporting the fix from commit 11ef59fb fix to 10.5, so that the test stays the same all versions >= 10.5. Signed-off-by: Julius Goryavsky <julius.goryavsky@mariadb.com>	6 months ago
Julius Goryavsky	fa55b36c1e	galera tests: corrections for garbd-related tests	6 months ago
Julius Goryavsky	41565615c5	galera: synchronization changes to stop random test failures	7 months ago
Vladislav Vaintroub	10b2187a94	Fix random mariabackup crashes with latest pcre2. Seen on Windows using newest pcre2-10.45. Tests that use regular expression e.g., mariabackup.partial, fail very often. To fix, serialize regexec calls, which are documented as non-thread-safe in https://www.pcre.org/current/doc/html/pcre2posix.html.	6 months ago
Aleksey Midenkov	c1492f3d07	MDEV-36115 InnoDB: assertion: node->pcur->rel_pos == BTR_PCUR_ON in row_update_for_mysql `932ec586` (MDEV-23644) in TABLE::delete_row() added ha_delete_row() for the case of HA_ERR_FOREIGN_DUPLICATE_KEY. The problem is ha_update_row() called beforewards may change m_last_part which is required for ha_delete_row() to delete from correct partition. The fix reverts m_last_part in case ha_partition::update_row() fails.	6 months ago
Aleksey Midenkov	1db7ccc124	MDEV-31122 Server crash in get_lock_data / mysql_lock_abort_for_thread While ALTER thread tries to notify SELECT thread about lock conflict it accesses its TABLE object (THD::notify_shared_lock()) and lock data (mysql_lock_abort_for_thread()). As part of accessing lock data it calls ha_partition::store_lock() which iterates over all partitions and does their store_lock(). The problem is SELECT opened 2 read partitions, but ha_partition::store_lock() tries to access all partitions as indicated in m_tot_parts which is 4. So the last 2 partitions m_file[2] and m_file[3] are uninitialized and store_lock() accesses uninitialized data. The code in ha_partition::store_lock() does this wrong handling to use all partitions specifically for the case of mysql_lock_abort_for_thread(), this is conducted with comment: /* This can be called from get_lock_data() in mysql_lock_abort_for_thread(), even when thd != table->in_use. In that case don't use partition pruning, but use all partitions instead to avoid using another threads structures. */ if (thd != table->in_use) { for (i= 0; i < m_tot_parts; i++) to= m_file[i]->store_lock(thd, to, lock_type); } The explanation is "to avoid using another threads structures" does not really explain why this change was needed. The change was originally introduced by: commit 9b7cccaf319 Author: Mattias Jonsson <mattias.jonsson@oracle.com> Date: Wed May 30 00:14:39 2012 +0200 WL#4443: final code change for dlenevs review. - Don't use pruning in lock_count(). - Don't use pruning in store_lock() if not owning thd. - Renamed is_fields_used_in_trigger to is_fields_updated_in_trigger() and check if they may be updated. - moved out mark_fields_used(TRG_EVENT_UPDATE) from mark_columns_needed_for_update(). And reverted the changed call order. And call mark_fields_used(TRG_EVENT_UPDATE) instead. which also fails to explain the rationale of the change. The original idea of WL#4443 is to reduce locks and this change does not happen to serve this goal. So reverting this change restores original behaviour of using only partitions marked for use and fixes invalid access to uninitialized data.	6 months ago
Yuchen Pei	866c06ac2d	MDEV-36441 MDEV-35452 Fix extra spider_same_server_link warnings in view-protocol With --view-protocol, mtr transforms a SELECT query to two queries: 1. CREATE OR REPLACE VIEW mysqltest_tmp_v AS ... 2. SELECT * FROM mysqltest_tmp_v where ... is the original query. Further mtr may run the first query in a separate connection. On the other hand if the data node is the same as the spider node, spider_same_server_link is required for connection to the data node. Therefore, for mtr --view-protocol tests often spider_same_server_link needs to be set on both session and global levels. In this patch we add the missing "SET GLOBAL spider_same_server_link=1" queries to tests that fail with wrong results due to this issue. It does not fix --view-protocol for all the affected tests, because there are other issues fixed in subsequent patches.	6 months ago

... 3 4 5 6 7 ...

199616 Commits (10.6) All Branches Search

199616 Commits (10.6)

All Branches