Details
-
Bug
-
Status: Closed (View Workflow)
-
Blocker
-
Resolution: Done
-
N/A
-
Not for Release Notes
Description
InnoDB changes between MySQL 8.4.10 and MySQL 8.4.11
Not applicable to MariaDB Server
Bug#25579578 INNODB: FAILING ASSERTION: !FIELD->PREFIX_LEN ¦¦ FIELD->FIXED_LEN == FIELD->PREF
This fixes a compilation error in a MySQL 8.4.9 bug fix that was deemed not applicable to MariaDB in MDEV-39457. Starting with MariaDB Server 10.2, this type of compilation errors would be impossible, because we define the following:
typedef size_t ulint; |
Bug#39129182 Mysqld Crash due to "Deadlock Detected!" Error
This is changing something in the function btr_cur_search_to_nth_level(), which was heavily refactored in MDEV-30400 and MDEV-29835 to fix similar problems. The fix is not directly applicable as such. I posted comments to MDEV-38814 and MDEV-40617 about this.
Bug#37105825 Skip value when add an auto_increment column
This is fixing a bug in a parallel CREATE INDEX code path. MDEV-16281 has not been implemented in MariaDB Server yet.
Bug#39244016 missed release of FIL_SHARD leads to deadlock under high load
In MariaDB Server, the fil_system.mutex was never split in the first place; instead, the root cause of the scalability bottleneck was addressed in MDEV-23855.
BUG#39157211 InnoDB: SET PERSIST innodb_adaptive_hash_index=ON may persist OFF during buffer pool resize
There is no SET PERSIST in MariaDB Server yet. The buffer pool resizing was redesigned in MDEV-29445, but we do still disable the adaptive hash index while shrinking the buffer pool.
BUG#39165747 Assertion field.field_no < dict_index_get_n_fields(index) for MLOG_REC_UPDATE_IN_PLACE with ROW_FORMAT=REDUNDANT
Starting with MDEV-12353 in MariaDB Server 10.5, no index metadata is being written to log records. This change is relaxing a debug check that is related to the MySQL implementation of instant ALTER TABLE. Even in MariaDB Server 10.3 or 10.4 using the old log record format, the implementation of MDEV-11369 and MDEV-15562 were simpler and not affected by this.
Bug#39245844 CLONE client might write to arbitrary path
There is currently no plan to implement a CLONE function in MariaDB Server; MDEV-36733 was cancelled. We plan to implement BACKUP SERVER (MDEV-14992) intead.
Bug#39091376 Failed DDL may not cleanup all builders
MDEV-16281 has not been implemented in MariaDB Server yet.
Bug#39245805 Possible null ptr derference in clone0apply.cc:1400
This fixes something related to a new log format and recovery logic that was introduced for the MySQL CLONE function. The MariaDB BACKUP SERVER (MDEV-14992) builds upon MDEV-37949, which introduces a innodb_log_archive=ON format variation to the existing redo log.
Bug#38169053 MySQL 8.0.39/40 crashed (post-push fix)
MySQL 8.0 introduced a Global Data Dictionary that stores information of all InnoDB data files in a dictionary table. In MariaDB Server, SYS_TABLESPACES and SYS_DATAFILES were removed in MDEV-22343. In MariaDB, the discovery of data files is based on FILE_ records in the redo log as well as the contents of the SYS_TABLES table.
Bug#39249507 Regression : "Row size too large (> 8126)" error with table using InnoDb DYNAMIC row format
This partly reverts the fix of Bug#39129182, which was rejected from MariaDB Server in MDEV-39457, because the change seems to be related to a MySQL specific implementation of instant ALTER TABLE, which appeared after MDEV-11369 and MDEV-15562 had been released.
Applicable to MariaDB Server
Bug#39433811 Can't create some of tables with SPATIAL KEY because of ER_TOO_BIG_ROWSIZE
Filed as MDEV-40617. This is not a security bug fix, but an enhancement. This is by far not the only bug in our implementation of SPATIAL INDEX; bugs like MDEV-15284 are more critical.
Attachments
Issue Links
- blocks
-
MDEV-40327 Community Server Q3 2026 release merge
-
- Open
-
- is blocked by
-
MDEV-40614 Merge applicable changes from InnoDB 8.4.10
-
- Closed
-
- relates to
-
MDEV-40617 Bogus ER_TOO_BIG_ROWSIZE reported for SPATIAL INDEX
-
- Open
-