Marko Mäkelä added a comment - 2024-10-16 06:08 It looks like this covered the changes from MySQL 8.0.38 and MySQL 8.0.39, not 8.0.37 and 8.0.38 as it was claimed earlier. Below is the list of InnoDB changes between MySQL 8.0.36 ( MDEV-33276 ) and 8.0.37. Not applicable to MariaDB Server Bug#35240825 performance_schema.data_locks query increased RAM usage to OOM This view does not exist in MariaDB Server. Bug#35993300 Server assertion crash after upgrade on 8.0.30 This is related to ALTER INSTANCE DISABLE INNODB REDO_LOG . In MariaDB Server, the closest equivalent was something that was deprecated and ignored in MDEV-19747 . Bug#36122452 Update copyright headers to 2024 Oracle copyright years in the MariaDB Server source code files are updated when importing code changes from specific files. Bug#35789759 omit multi-value index columns when extracting rows from fetch cache MariaDB Server does not support other inverted indexes than FULLTEXT INDEX yet. Bug#36180360 Backport Bug#35625510 to 8.0 This is something related to the MySQL implementation of MDEV-11369 or MDEV-15562 . They would store a version number in each record. The MariaDB implementation is simpler, only storing the number of instantly added fields in each affected clustered index record, and storing metadata in a special record at the start of the index while not changing the format of any dictionary tables. Bug#35740866 parallel read is slow when innodb_use_native_aio = off This is disabling asynchronous read-ahead during parallel index scans. MariaDB does not support parallel index scans yet. Bug#36142806 innodb_parallel_read_threads>1 makes simple select count more expensive MariaDB does not support parallel index scans yet. Bug#36154818 Redolog files on Windows are not using Overlapped mode According to wlad , this could be beneficial when the same file is being concurrently accessed from multiple threads. In MariaDB Server, that is not the case; WriteFile and NtFlushBuffersFileEx would be invoked from the same thread, guaranteed by the group commit mechanism ( MDEV-21534 ). Bug#35183686: ALTER_STORED_COLUMN_ORDER with algorithm=INSTANT makes wrong INSERT REDO log Starting with MDEV-12353 , InnoDB will not write any index metadata as part of logging changes to data pages. This might theoretically affect MariaDB Server 10.4, which has reached EOL. I did not investigate that. Bug#36118112 srv_io_thread_function are assigned in incorrect order This is fixing something that was introduced by an earlier refactoring, which was not applied to MariaDB Server. Bug#36159317 INNODB: DEADLOCK BETWEEN FIL_SHARD::OPEN_FILE, FIL_OP_REPLAY_RENAME ON WINDOWS 2 This is following up some earlier change. In MariaDB Server, fil_system.mutex was never split to shards; contention on it was addressed by different means. Bug#36108397 Upgrade to latest protobuf library This seems to bundle an unrelated change in InnoDB, to address a warning about designated initializers, which we had removed from that file in MDEV-26443 . GPL License Exception Update This is changing the wording of some copyright headers in many source code files. We can update such comments when actually incorporating changes to the MariaDB Server code base. Bug#35952353 SELECT COUNT(*) degraded performance on 8.0 compared to 5.7 This is due to parallel index scan, which MariaDB Server does not support yet. Bug#35277407 InnoDB：trx hangs due to wrong trx->in_innodb value TrxInInnoDB had been removed in MDEV-13935 . This bug could possibly duplicate MDEV-11197 . Bug#36124625 HCS-11410 : Restore Dbsystem Failing with error "at least one redo file is missi This is related to MySQL 8.0 writing multiple log files. Starting with MariaDB Server 10.5, there is only one ib_logfile0 . Bug#35289234 Can't disable encryption once redo log encryption is enabled This is something specific to the MySQL implementation of innodb_redo_log_encrypt . The logic of innodb_encrypt_log in MariaDB Server is somewhat different. Bug#35932118 innodb.redo_log_archive_01 fails with log0write.cc:2115:ib::fatal MariaDB Server does not support log archiving. Bug#36319083 Merge sort buffer can be too small Bug#36342792 IO write to merge file aligned past end of buffer Bug#35836581 Server crashes when adding a fulltext index Bug#36342792 IO write to merge file aligned past end of buffer These are fixing regressions due to parallel index creation. MariaDB does not implement MDEV-16281 yet. Bug#35835864 Crash during background rollback if both prepare and active transaction exist MariaDB Server is resurrecting table locks for XA PREPARE transactions. Apparently an earlier change in MySQL 8.0.33 had broken the logic. Bug#36343647 Assertion failure:ibuf0ibuf.cc:3833:ib::fatal This is fixing a regression that had been introduced in MySQL 8.0.11 . ( Bug#36343647 Assertion failure:ibuf0ibuf.cc:3833:ib::fatal This had been fixed in MDEV-15916 back in 2018. Bug#36425219 log_writer_write_buffer must double-check log.write_lsn after reacquiring mutex This is fixing something related to the multi-threaded, multi-file log writer in MySQL 8.0. MariaDB has group commit on a single ib_logfile0 that seems to work well, especially ever since MDEV-27774 and MDEV-33515 were addressed. Applicable to MariaDB Server Bug#35127747 The n_rows value becomes 0 when repeated altering with algorithm=copy within 10 seconds Filed as MDEV-35163 . Bug#34929814 Inconsistent FTS state in concurrent scenarios ( follow-up ) Filed as MDEV-34057 .