MariaDB 11.5.2

What's new in this version:

Storage Engines:
InnoDB:
- ALTER TABLE...ALGORITHM=COPY now builds indexes more efficiently
- Unsetting the innodb_alter_copy_bulk system variable restores the old behavior.
- Fix unexpected storage read IO for the redo log, reintroduce innodb_log_write_ahead_size
- Correctly terminate transaction early with ER_LOCK_TABLE_FULL when lock memory is growing
- Alter operation on redundant table no longer aborts the server
- Fix MariaDB crash with SIGILL because the OS does not support AVX512
- Fix InnoDB: Failing assertion: stat_n_leaf_pages > 0 in ha_innobase::estimate_rows_upper_bound
- wait_for_read in buf_page_get_low no longer hurts performance
- Fix InnoDB: Assertion failure in file ./storage/innobase/page/page0zip.cc line 4211

Aria:
- Fix Aria index corruption

Spider:
- UBSAN: runtime error: applying non-zero offset in spider_free_mem and SIGSEGV in spider_free_mem on SELECT
- SIGSEGV in ha_resolve_storage_engine_name, UBSAN: runtime error: index 4294967295 out of bounds for type 'st_plugin_int *[64]'
- SIGSEGV in spider_db_conn::fin_loop_check, and ASAN: heap-use-after-free in spider_db_mbase::fin_loop_check on SHOW TABLE STATUS
- SIGSEGV in ha_spider::lock_tables on BEGIN after table lock
- SIGSEGV in spider_conn_first_link_idx and others on DELETE, INSERT and SELECT
- Spider: Crashes, asserts, hangs, memory corruptions and ASAN heap-use-after-free's
- Spider: @@insert_id 128 to TINYINT: Assertion `!is_set() || (m_status == DA_OK_BULK && is_bulk_op())' failed.
- ASAN errors in spider_db_free_result after partition DDL
- Server crashes when calling spider UDF after aria_encrypt_tables is enabled

Partitioning:
- SIGSEGV in parse_engine_part_options on INSERT, SELECT or ALTER
- Assertion `auto_increment_value' failed in ha_partition::info on INSERT into MEMORY table

Character Sets:
- On startup: UBSAN: applying zero offset to null pointer in my_copy_fix_mb from strings/ctype-mb.c and other locations
- On startup: UBSAN: runtime error: applying zero offset to null pointer in skip_trailing_space and my_hash_sort_utf8mb3_general1400_nopad_as_ci
- SHOW CREATE DATABASE statement crashes the server when db name contains some unicode characters, ASAN stack-buffer-overflow
- Wrong result set with utf8mb4_danish_ci and BNLH join

Optimizer:
- On startup: UBSAN: runtime error: applying non-zero offset in JOIN::make_aggr_tables_info in sql/sql_select.cc
- Crash after killing query while it is processed by test_quick_select
- Extend condition normalization to include 'NOT a'
- Constant subquery causing a crash in pushdown optimization
- Crash when pushing condition with CHARSET()/COERCIBILITY() into derived table
- 2nd execution name resolution problem with pushdown into unions
- Assertion `(key_part->key_part_flag & 4) == 0' failed key_hashnr
- Crash caused by query containing constant having clause
- Using NAME_CONST() (or executing query from the stored procedure and referring to a local variable) changes the plan and may make execution slower
- ORDER BY DESC causes ROWID Filter optimization performance degradation

Replication:
- Auto-generated DELETE from HEAP table no longer breaks replication
- Fix replication failure when XA transactions are used where the replica has replicate_do_db set and the client has touched a different database when running DML such as inserts.
- Fix replication error when CHANGE MASTER TO is used in conjunction with a password longer than 41 ascii characters
- The --init-rpl-role option is utilized to avoid a possible error state in semisync recovery

Backup:
- Hide password passed on commandline from xtrabackup_info

Galera:
- Galera updated to 26.4.19
- galera_gtid_2_cluster: Assertion `thd->wsrep_next_trx_id() != (0x7fffffffffffffffLL * 2ULL + 1)'
- table gtid_slave_pos entries never been deleted with wsrep_gtid_mode = 0
- Deadlock found when trying to get lock during applying
- Change error code for Galera unkillable threads
- 10.11.8 cluster becomes inconsistent when using composite primary key and partitioning
- wsrep_sst_mariabackup use /tmp dir during SST rather then user defined tmpdir

General:
- As per the MariaDB Deprecation Policy, this will be the last release of MariaDB 11.5 for Debian 10 "Buster", RHEL/CentOS 7, Ubuntu 23.10 "Mantic", and Fedora 38
- Repositories for Ubuntu 24.04 "Noble" have been added
- IMPORT TABLESPACE no longer unnecessarily traverses tablespaces list
- Fix unknown variable defaults-group-suffix= with mariadb-secure-installation
- mariadb-install-db no longer hangs on macOS
- Fix transaction termination with ER_LOCK_TABLE_FULL when lock memory is growing
- Disable new connections in case of fatal signal
- Control over memory allocated for SP/PS
- Triggers now work correctly with bulk insert
- Fix assertion `table->field[0]->ptr >= table->record[0] && table->field[0]->ptr <= table->record[0] + table->s->reclength' failed in void handler::assert_icp_limitations(uchar*)
- sandbox mode - now compatible with --binary-mode