Home - Waterfall Grid T-Grid Console Builders Recent Builds Buildslaves Changesources - JSON API - About

Console View


Categories: connectors experimental galera main
Legend:   Passed Failed Warnings Failed Again Running Exception Offline No data

connectors experimental galera main
fdfd7487deab...
fdfd7487deabe948dd765426dfcef11e24eb644b
Vladislav Vaintroub
MDEV-38741 Windows: watch with a dedicated one-thread pool

The waits were on the system thread pool, so the callbacks ran on
arbitrary threads, several at a time, and not registered in mysys.
Use a pool of our own with a single thread instead. It runs all the
callbacks one by one, like the main loop handles the epoll events, and
registers itself in mysys with a thread_local guard on the first one.

Shutdown no longer counts running callbacks and spins: they do nothing
once the server is stopping, and closing the pool lets its thread exit
when the connections have removed their waits. my_thread_global_end()
waits for that thread.

remove() no longer clears the event selection of the socket, which is
closed right after.

Polling the socket with the AFD driver and one completion port was
tried and rejected: while such a poll is pending, select(), which the
vio layer waits with, loses the data notification of the same socket.

Assisted-By: Claude:claude-sonnet-5-5
f93ec1983fc6...
f93ec1983fc65c7cc0adedf015b48f1512ec7339
Sergei Golubchik
reject out of range rec_per_key in replay context

Index_statistics::set_avg_frequency() casts rec_per_key * 100000 to
ulonglong, which is undefined for negative or too large values.

Reject them when parsing the context.

Assisted-by: Claude:claude-5.5-opus
f420bd4249bb...
f420bd4249bbe4cfb6b5b10e94793160b4ec76b3
Sergei Golubchik
MDEV-41541 uninitialized memory read in optimizer replay

init all frequencies - not all of them might be present in the replay
ebac0469efba...
ebac0469efbadff39a0578c8cb0daf5dace6e3c4
Oleksandr Byelkin
Merge branch '13.1' into bb-13.1-release
d858d6033ae5...
d858d6033ae5cda70fc53e73fde81461c5dbc714
Oleksandr Byelkin
Merge branch 'bb-12.3-release' into bb-13.1-release
d553b2dd4cfe...
d553b2dd4cfe8c56c37c0d4623698b57331fd3da
Sergei Golubchik
disable DuckDB in the server in favor of the externally built

Fixes #6045
d057e2199596...
d057e2199596dfed3a3462b7ec036c2c449aa2ab
Sergei Golubchik
cleanup: remove obsolete CMAKE_VERSION checks

main CMakeLists.txt sets

CMAKE_MINIMUM_REQUIRED(VERSION 3.12.0)
cee366562810...
cee3665628103a7841a9861ab2a442a6e1dfd912
Vladislav Vaintroub
MDEV-38741 disconnect watcher: end the connection if a statement is aborted

KILL_DISCONNECT is statement-level. When a disconnect does abort a
statement, the connection carried on and ran the commands the client
had pipelined behind it, so a client that closed its sending side got
one statement aborted and the others executed.

Now the connection ends too, after the error of the aborted statement
is sent and the reason is logged. If the kill aborts nothing, for
example the statement has finished and the client already sent
COM_QUIT, nothing happens, so an orderly quit or a close without it is
handled as before, and logs nothing.

The condition in end_connection() for the "Aborted connection" line
needs no change, as the killed connection does not get it.

main.disconnect_lock_wait checks that orderly disconnects log no
"Aborted connection" line, and that an aborted connection is logged
once. It waits for the lock wait of the connection that it closes, not
for any lock wait.

Assisted-by: Claude:claude-5.5-sonnet
ce5f63733dd3...
ce5f63733dd38f244c25f81aef822e8cc2facff4
Sergei Golubchik
CREATE VIEW with a long name writes .frm outside of the database directory

fn_format() silently drops the directory when the path is too long.
Use MY_SAFE_PATH in sql_create_definition_file() and fail with
ER_IDENT_CAUSES_TOO_LONG_PATH.

Assisted-by: Claude:claude-5.5-opus
b75b72e706e2...
b75b72e706e2173de52aff29dc3cb1eff0fd52c2
Sergei Golubchik
unix_socket: reject connections from the server process itself

CONNECT, FederatedX and Spider connect to the local server via the unix
socket with user-supplied credentials. unix_socket saw mysqld's own uid
as the peer, so anyone allowed to create such a table could log in as
the OS user mysqld runs under (mysql@localhost, ALL PRIVILEGES by default).

Let's fail authentication when the peer pid is the server's own pid.
On old FreeBSD (<13) that cannot get the peer pid allow everything
as before.

Assisted-by: Claude:claude-5.5-opus
b25f3e321c99...
b25f3e321c99195bb79f7c961306b486891ae19a
Sergei Golubchik
cleanup: don't malloc every single String for every single byte
b0e59c7633f0...
b0e59c7633f0394ca5afd9d3e862fed2b8d95d8b
Alexey Botchkov
MDEV-41430 auth_pam_tool leaves zombie processes.

As the auth_pam_tool claims root privileges, the mariadbd can't kill
it if it hangs.
So instead of sending the SIGKILL to the auth_pam_tool we set the
alarm() there so it kills itself after the timeout.
The alarm counter is hardcoded to 100 seconds so the authentication
can be halted even if the connect_timeout allows it to be longer.
The auth_pam.c now calls the waitpid() without the WNOHANG, so it waits
for the auth_pam_tool to return. If the PAM module aggressively resets
the alarm or handles the SIGALARM the pam_auth() can hang forever.
a1cf8c048ed3...
a1cf8c048ed3b97fd4d1f8e50a2a71f53849c541
Sergei Golubchik
stack overflow opening a table with a too long path

CREATE TABLE allows .frm path of up to FN_REFLEN bytes, but buffers
for normalized_path + reg_ext were FN_REFLEN, leaving no room for \0.
A nonexistent table name could produce a path even longer than that.

* make buffers FN_REFLEN + 1
* in open_table_def(), treat a path that cannot fit as a missing frm

Assisted-by: Claude:claude-5.5-opus
9dff871612ab...
9dff871612ab355281000b5216b069347cccc73d
Sergei Golubchik
MDEV-41830 ASAN stack-buffer-overflow in option file parser
9c105ead5fe1...
9c105ead5fe1b9a7ce055d638f74ecafe0002c9f
Oleksandr Byelkin
Merge branch 'bb-13.1-release' into bb-13.2-release
9320f20b88ca...
9320f20b88ca300590eebb609061075e164242fc
Sergei Golubchik
MDEV-40862 mariadb-backup --backup crashes on database and table names that expand to nearly 255 bytes in the file system

* make_path_n() never grew the buffer and never updated its length
* local_open(): use MY_SAFE_PATH, a too long path must not fall back
  to a relative path in the current (source data) directory

Assisted-by: Claude:claude-5.5-opus
8c01931bbe17...
8c01931bbe1797daff18dd8ac127779924e6caf4
Oleksandr Byelkin
Merge branch '12.3' into bb-12.3-release
8ab8b3991085...
8ab8b399108532a908c71e6d23ff1c03e3738779
Sergei Golubchik
CREATE VIEW with a long name writes .frm outside of the database directory

fn_format() silently drops the directory when the path is too long.
Use MY_SAFE_PATH in sql_create_definition_file() and fail with
ER_IDENT_CAUSES_TOO_LONG_PATH.

Assisted-by: Claude:claude-5.5-opus
86523d901286...
86523d9012866f02e5d8d483f0a8128474cea985
Sergei Golubchik
MDEV-40374 I_S.metadata_lock_info shows schema/table that user has no privileges on

Apply the SHOW PROCESSLIST visibility rule: without PROCESS a user
sees only locks held by threads of the same account. That is enough
to find who blocks one's own statement in the case where the user
can KILL the blocker.

Assisted-by: Claude:claude-opus-5-5
813c3e8a4651...
813c3e8a4651a00a24a24ea943bf5f2a4996f1df
Sergei Golubchik
replay infuses statistics into INFORMATION_SCHEMA tables

fill_schema_table_by_open() opens and closes tables in the middle of
the statement. set_statistics_for_table() in get_schema_stat_record()
infused replayed statistics into them, which then went back to the
table cache before the replay restored them, and I_S.STATISTICS
showed the replayed cardinality.

Disable the replay while I_S opens tables.

Assisted-by: Claude:claude-5.5-opus
7128f50a142b...
7128f50a142b19a506801c5e1ab54ef729578b9f
Sergei Golubchik
MDEV-38859 federated use-after-free with temp tables and savepoints

ha_init() added sizeof(SAVEPOINT) to savepoint_alloc_size after
startup engines were initialized, so an engine installed at runtime
got a savepoint_offset that already included sizeof(SAVEPOINT),
and (sv+1)+savepoint_offset counted it twice, pointing past
the end of the allocated SAVEPOINT.

Start savepoint_alloc_size at sizeof(SAVEPOINT) and make
savepoint_offset relative to the SAVEPOINT itself.

Assisted-by: Claude:claude-opus-5-5
6b587bf94319...
6b587bf943196ad2d91afab0d02f7b124eb54d94
Sergei Golubchik
MDEV-39752 use-of-uninitialized-value in ftb_parse_query_internal

maria_ft_get_word() does not set the word for bracket and quote tokens,
yet ftb_parse_query_internal() passes w.pos and w.len to
mysql_add_word() by value for every token. clang-16+ MSAN reports
passing uninitialized values as arguments.

Initialize w, as MDEV-33478 did for the MyISAM copy of this function.

Assisted-by: Claude:claude-opus-5-5
672dbe816b3e...
672dbe816b3e5686ce9f4d1da321845fba317814
Sergei Golubchik
MDEV-39615 CONNECT: Privilege validation gap

A CONNECT sub-table of PROXY, TBL, XCL, OCCUR, PIVOT was opened
in-process without any privilege check, so a user with CREATE could
read and write any existing CONNECT table, including file-based ones
that otherwise require FILE.

Check privileges on the sub-table as if it were used directly instead
of the PROXY table, like in a SQL SECURITY INVOKER view: SELECT, INSERT,
SELECT+UPDATE or SELECT+DELETE depending on the mode, plus the
sub-table's own FILE privilege checks. Info commands fail silently,
as for other sub-table errors.

Non-CONNECT sub-tables are still accessed via a client connection with
the table's credentials.

Assisted-by: Claude:claude-5.5-opus
5925a64f8a0c...
5925a64f8a0c771862bcb2e2b5e04b72242f9e69
Sergei Golubchik
MDEV-40862 mariadb-backup --backup crashes on database and table names that expand to nearly 255 bytes in the file system

* make_path_n() never grew the buffer and never updated its length
* local_open(): use MY_SAFE_PATH, a too long path must not fall back
  to a relative path in the current (source data) directory

Assisted-by: Claude:claude-5.5-opus
546350c61837...
546350c61837b237e93c1393b1aae6535c085ef9
Vladislav Vaintroub
MDEV-38741 Do not report the client closing after COM_QUIT

The client closes the socket right after COM_QUIT, which is not a
failure, and there is no statement left to abort.

Unregister the watcher when COM_QUIT is parsed, so that on Linux most
of the closes are not reported at all, and ignore a disconnect in a
quitting connection like in an idle one.

Assisted-By: Claude:claude-sonnet-5-5
4bb6501ba27a...
4bb6501ba27a3382d41ce0989d37e5fdc19d6881
Sergei Golubchik
MDEV-41362 ASAN heap-buffer-overflow in validate/plugin_foreach_with_mask after password_reuse_check

allocate one byte more for `buff[key_len]=0` to work
4641beb5f8d0...
4641beb5f8d00d645d94362b6a031ecca6fb7269
bsrikanth-mariadb
MDEV-40553 fixup: GCC -Og -Wmaybe-uninitialized
3d07a9d6eee9...
3d07a9d6eee959d1b252d1636d173a1d6bb66f26
Oleksandr Byelkin
Merge branch 'bb-11.8-release' into bb-12.3-release
342379092564...
342379092564c51fbd371d47af4c843ded0bb46c
Vladislav Vaintroub
MDEV-38741 disconnect_watcher system variable

A global variable to switch the disconnect watcher on and off, ON by
default. It is dynamic and affects the connections made after the change.
The variable exists only where there is a watcher: on Linux, Windows and
the BSD systems, not elsewhere or in the embedded server.

The tests that depend on the watcher run with it ON and OFF, see
include/disconnect_watcher.combinations. With the watcher OFF a lock wait
is not aborted, so the test releases the lock instead, see the rdiff files.
A combination that does not fit the server value is skipped.
innodb.xa_recovery, rpl_password_boundaries and rpl_ssl1 do not test the
watcher and behave the same in both modes.

Assisted-By: Claude:claude-sonnet-5-5
30ece046ab68...
30ece046ab68788b280121f956869a2b13c37f66
Sergei Golubchik
stack overflow opening a table with a too long path

CREATE TABLE allows .frm path of up to FN_REFLEN bytes, but buffers
for normalized_path + reg_ext were FN_REFLEN, leaving no room for \0.
A nonexistent table name could produce a path even longer than that.

* make buffers FN_REFLEN + 1
* in open_table_def(), treat a path that cannot fit as a missing frm

Assisted-by: Claude:claude-5.5-opus
210c57988e5d...
210c57988e5dab9fcf98891208c029c7669505cb
Sergei Golubchik
MDEV-41944 CONNECT CSV discovery header overflows colname[]/len[]/typ[]/prc[]

CSVColumns() stored a column name pointer for every separator in the
header line without checking MAXCOL, so a header with more than 200
fields wrote past the end of the colname[] stack array, and the
resulting column count indexed len[], typ[], prc[] out of bounds.

Reject such headers with TOO_MANY_FIELDS, the same way data lines
are rejected.

Assisted-by: Claude:claude-opus-5-5
20f27639cfa0...
20f27639cfa05f1c2175701f97b23cea491b6154
Sergei Golubchik
replay loses infused statistics in stored functions

Every sub-statement (stored function, trigger) replaced
thd->opt_ctx_replay, leaking the top-level one, so the top-level
tables were never restored and kept key_info->read_stats pointing
into the freed statement mem_root. Saved and infused statistics were
allocated on whatever arena was active, e.g. the stored function
execution arena, freed before the restore.

* create the replay object only for a top-level statement, and let
  sub-statements use it; restore only at the end of the top-level one
* allocate everything the replay object owns in its own MEM_ROOT
* empty the saved list after restoring

Assisted-by: Claude:claude-5.5-opus
19906247ebe7...
19906247ebe7bbf1b3c1f6656bd24997fcc3492b
Sergei Golubchik
compilation failure

storage/innobase/row/row0import.cc:678:17: error: conversion from 'uint32_t' {aka 'unsigned int'} to 'uint16_t' {aka 'short unsigned int'} may change value [-Werror=conversion]
  678 |                ? 0 : FSP_FLAGS_GET_ZIP_SSIZE(m_space_flags);
13a98bf11bde...
13a98bf11bde9101fd1c633d0b2dbfe283e3a156
Sergei Golubchik
replay leaves stale index statistics after DELETE

A table can be infused twice per statement (DELETE without WHERE on an
engine without delete_all_rows() calls set_statistics_for_table() twice).
The second save holds the first infused Index_statistics as "original",
and restore_modified_table_stats() restored it last, leaving
key_info->read_stats pointing into the freed statement mem_root.

Restore in the reverse order of saving.

Assisted-by: Claude:claude-5.5-opus
111a6a98d8ba...
111a6a98d8ba6035a351fa52833a809b488fdcaa
Sergei Petrunia
MDEV-41862: Followup: print single-table DELETE without USING.
10b04bc5cf3d...
10b04bc5cf3d63df33d3e617a8f2b645b8df2a02
Oleksandr Byelkin
Merge branch 'main' into bb-13.2-release
078df7d81c9c...
078df7d81c9c645dd30ed8bcf0ee7ef9d0d3fa29
Sergei Golubchik
MDEV-41830 ASAN stack-buffer-overflow in option file parser
04992eecd488...
04992eecd48851080070080d8e47546b6ad3d168
Sergei Golubchik
MDEV-40690 auth_pam_tool loses setuid-root bit in RPM packages

without %dir, rpm list of files includes the content of auth_pam_tool_dir
recursively, and auth_pam_tool_dir has its own line too. The end result
depends on the order in which two lines are applied which in turn
depends on the order the sort will put them in. rpm 6 was rewritten
in C++ and moved from libc qsort to std::sort.

11.4+ fix, because `%dir %attr` only works in cmake >= 3.10
034b4c26f895...
034b4c26f8957287242026734fb5fab8fbe3c214
Sergei Golubchik
MDEV-41362 ASAN heap-buffer-overflow in validate/plugin_foreach_with_mask after password_reuse_check

allocate one byte more for `buff[ley_len]=0` to work
BuildBot (0.8.9) working for the MariaDB project.
Page built: Sat 31 Oct 2026 22:26:05 (UTC)