Commits
8.0
Name already in use
Commits on Oct 12, 2023
-
Bug#35897778: Upgrade to CURL 8.4.0 [remove old]
Remove all old source files. Change-Id: Ifb43e581715178079b5a5a6e0ebcf201d2aae8f2
-
Bug#35897778: Upgrade to CURL version 8.4.0 [patches]
curl cmake patches for 8.0 Update CURL_VERSION_DIR Set options in extra/curl/CMakeLists.txt - PICKY_COMPILER OFF - BUILD_SHARED_LIBS OFF - BUILD_STATIC_LIBS OFF - SHARE_LIB_OBJECT OFF Disable in curl cmake files: - CMAKE_DEBUG_POSTFIX - find_package(OpenSSL) - Do not add "/MP" for clang on Windows - install(...) Enable in curl cmake files: - Keep the FILE protocol for HTTP_ONLY Add comments about 'libcurl' and 'libcurl_static' Remove obsolete INCLUDE_DIRECTORIES and LINK_DIRECTORIES from extra/curl/CMakeLists. Change-Id: I108d8c961b13e040305b8bf2a167d89b41147481
-
Bug#35897778: Upgrade to CURL 8.4.0 [add sources]
Unpack curl-8.4.0.tar.gz rm buildconf.bat configure ltmain.sh config.guess config.sub Makefile rm -rf docs m4 packages plan9 projects src tests winbuild git add curl-8.4.0 Change-Id: I839e94338e4bed0ea132522ca41e01a51d1e9447
-
Bug#35855488 Build on Ubuntu 23.10 [noclose]
Backport -Wno-alloc-size-larger-than suppressions from trunk. from the patch for: Bug#34099162 Add support for mold linker Change-Id: I3996c472b423daaf3002c477ccb757d2e55846f9 -
Bug#34278103 Compile MySQL with GCC 13 [noclose]
For a Debug build, with -O2, we get: In file included from /usr/include/string.h:548, from extra/googletest/googletest-release-1.12.0/googletest/include/gtest/xxxxxx/gtest-port.h:262, from extra/googletest/googletest-release-1.12.0/googletest/include/gtest/gtest-message.h:55, from extra/googletest/googletest-release-1.12.0/googletest/include/gtest/gtest-assertion-result.h:46, from extra/googletest/googletest-release-1.12.0/googletest/include/gtest/gtest.h:59, from unittest/gunit/item_json_func-t.cc:23: In function void* memset(void*, int, size_t), inlined from void TRASH(void*, size_t) at include/memory_debugging.h:72:9, inlined from static void Item::operator delete(void*, size_t) at sql/item.h:962:10, inlined from void item_json_func_unittest::BM_JsonSearch(size_t) at unittest/gunit/item_json_func-t.cc:721:78: /usr/include/x86_64-linux-gnu/bits/string_fortified.h:59:33: error: <anonymous> may be used uninitialized [-Werror=maybe-uninitialized] 59 | return __builtin___memset_chk (__dest, __ch, __len, | ~~~~~~~~~~~~~~~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~ 60 | __glibc_objsize0 (__dest)); | ~~~~~~~~~~~~~~~~~~~~~~~~~~ unittest/gunit/item_json_func-t.cc: In function void item_json_func_unittest::BM_JsonSearch(size_t): unittest/gunit/item_json_func-t.cc:721:78: note: <anonymous> was declared here 721 | auto search = new Item_func_json_search(table.in_use, new Item_field(&field), | ^ gcc somehow thinks we may call this: static void operator delete(void *ptr [[maybe_unused]], size_t size [[maybe_unused]]) { TRASH(ptr, size); } The fix is to evaluate 'new Item_field(&field)' separately. Change-Id: I5d73561f900043092bcd982fffc8462707627f1a (cherry picked from commit e536c8a446b97ac9c90ce957ba1c34547b798993) -
Bug#35855488 Build on Ubuntu 23.10 [noclose]
sql-common/oci/signing_key.cc:87:42: error: ignoring attributes on template argument int (*)(FILE*) [-Werror=ignored-attributes] std::unique_ptr<FILE, decltype(&fclose)> fp(fopen(file_name.c_str(), "rb"), This is the declaration for Ubuntu 23.10 extern int fclose (FILE *__stream) __nonnull ((1)); Use a scope guard closure, rather than a unique_ptr with custom deleter. Change-Id: Iecb41b238dc9b72d87f33f54aea272391f3b5977 (cherry picked from commit 5d2e8c1aadb23c9d9acba692d2dd7afb24a0897a)
-
Bug#35847193 Cluster code fails to compile on Ubuntu23.10
Fixing compiler warning by adding a check to ensure accessing array element within the range. Reviewed by: Mauritz Sundell <mauritz.sundell@oracle.com> (backported from commit f21d8a05d5a8180cd0d2e8b49833532aaedf8f6a) Change-Id: I5c56b5ef1fa73908a3bd5c070ab7106fee4cc9ad
-
Bug#35847193 Cluster code fails to compile on Ubuntu23.10
Fix alloc-size-larger-than warning in Qmgr by explicitly only allocate non-zero sized array. Compiler can not deduce that that is actually always the case. The confusing warning fixed is: storage/ndb/src/kernel/blocks/qmgr/QmgrInit.cpp: In member function 'initData': storage/ndb/src/kernel/blocks/qmgr/QmgrInit.cpp:204:60: warning: argument 1 value '18446744073709551615' exceeds maximum object size 9223372036854775807 [-Walloc-size-larger-than=] 204 | receivedProcessInfo = new ProcessInfo[numOfApiAndMgmNodes]; | ^ /usr/lib/gcc/x86_64-pc-linux-gnu/13/include/g++-v13/new:128:26: note: in a call to allocation function 'operator new []' declared here 128 | _GLIBCXX_NODISCARD void* operator new[](std::size_t) _GLIBCXX_THROW (std::bad_alloc) | ^ Change-Id: Ie7aea5e8681db1c420d2a0e80d1cda8319f74a2b
Commits on Sep 25, 2023
-
Bug#35801349 Build fails on Windows due to CMake errors in storage/nd…
…b/wix Disable WiX builds for NDB. Change-Id: Ied0482eb52148ae8cc65a99a32b518a664d44d20
Commits on Sep 22, 2023
-
WL#13220: Deprecate --character-set-client-handshake server option
Added a deprecation warning. Added test. WL#8609: Deprecate old/new system variables Added a deprecation warning on command line and SET GLOBAL Added test WL#11091: Deprecate mysql_ssl_set() C API function Marked mysql_ssl_set() as deprecated using the C++14 macro Removed use of it from the mysql codebase. Post-push fix: Fix for issue reported by linktest Change-Id: I7d329c47c0a0a9ef4710215531dec72b59fd8812
-
WL#8609: Deprecate old/new system variables
Post push fix. Rerecord ndb test results due to deprecation warning for new. Change-Id: If8805d3bd5713e007f5d22145d459e217ab68d4f
Commits on Sep 20, 2023
-
WL#13220: Deprecate --character-set-client-handshake server option
Added a deprecation warning. Added test. WL#8609: Deprecate old/new system variables Added a deprecation warning on command line and SET GLOBAL Added test WL#11091: Deprecate mysql_ssl_set() C API function Marked mysql_ssl_set() as deprecated using the C++14 macro Removed use of it from the mysql codebase. Change-Id: I3596a504c3587e26389d6250d9eec36cf93136ce
-
WL#15800: Deprecate usage of weak ciphers in server
- Added deprecation message for weak ciphers WL#15006: WebauthN authentication plugin - Added authentication_fido deprecation Change-Id: I1545b6695c98051be284a61c37b187d6c4c0fe36
-
Bug#35814434 memory-usage increases with connection-sharing and bench…
…marking When benchmarking mysqlrouter with connection-sharing enabled, the RES of the mysqlrouter process increases over time even though it stays more-or-less constant when running the connection-sharing disabled. ASAN reports no memory leaks. Using "valgring --tool=massif", it is seen that the "timer_ops" are allocated a lot and not destroyed fast enough. Those timer_ops are from Timer.async_wait() and are already cancel()ed, but their result is not yet delivered to the async-callers. Cancelled timers are expected to be handled with the highest priority, but here they are handled only if there is no other work. Change ====== - treat "expired timer" and "cancelled timer" as "execute now". Change-Id: I76868b52850f81cbcc3f189e6128e4ff98063299
-
Bug#35398028 : crash when persisted GR variable exist and
GR plugin isn't loaded (during upgrade) Description: Server crashes on start up if the group replication plugin is disabled or uninstalled after upgrade from version 8.0.28 in which group replication variables were persisted. Server does not load certain persisted group replication variables when the plugin is installed after server start up. Fix: Handled the case of nullptr return value. Added functionality to load parse early variables during plugin installation. Change-Id: I04a96c8435f77e187648b7e5a9cc88cecaacd4f0
-
Bug#35812604 heap-use-after-free after read() fail
ASAN reports a use-after-free when handling queries larger than 16 Mbyte. Background ========== The use-after-free happens when using the read_buffer_view() after - Channel::read_to_plain() grew the read-buffer - the socket-read() failed - the read-buffer was resized to its original size. Change ====== - sync the read_buffer_view() even if the read() failed as it may have moved the read-buffer to a new memory area. Change-Id: I2511052394e83d217e4349557ec00664472b48da
-
Bug#35404584 Multi-column indexes with writest will cause HA_ERR_FOUN…
…D_DUPP_KEY Issue description ----------------- Replication uses write-sets to detect dependencies and/or conflicts among transactions. The write-set is extracted from primary, not null unique and not null foreign keys when the binary log format is ROW and the option`transaction_write_set_extraction` is enabled, both options are enabled by default. For string columns, indexes can be created that use only the leading part of column values, using col_name(length) syntax to specify an index prefix length. Analysis -------- It was discovered that multi-column unique key write-sets were not properly generated, more precisely it was considering that unique keys could only refer the leading part of column values, while the whole values of referenced key columns are added to writeset. Proposed solution ----------------- In add_pke(), the values of the columns related to the multi-column key should not be added to the writeset without truncation. Only the specified width should be added. Change-Id: I6d22b2e703096feb315bb85c50e6f2768ff25380
-
Bug#35613791 - Upgrade from Wix 3.11 to 4.0.1 for Server/Cluster MSI …
…builds WiX Toolset v3 is no longer actively maintained. Other new and old bugs corrected: Bug#35729602 - MEB msi does not contains the MySQL image like others msi. Bug#35734657 - Server MSI installing in "C:\Program Files (x86)" instead of "C:\Program Files" Change-Id: I08922db98fd7234ffab57e5d05f1c3cb6e297ae8
-
Bug#35685515: Upgrade libfido2 to v1.13.0
Post-push fix: Disabled -D_FORTIFY_SOURCE=2 usage Change-Id: I594eaada8ef81f90242fb87f0e4703157eb8a36d
-
Bug#35769610 statement with open comment crashes connection sharing
When connection sharing is enabled, a statement like: DO 1 /* crashes mysqlrouter. ASAN reports a heap-buffer-overrun. Change ------ - handle ABORT_SYM in the lexer as end-symbol. - add a fuzzer for the SqlLexer Change-Id: Ice21cac61f9e04208a8d0b6492a3af02ff078a12
-
Bug#35769702 queries larger than 16M aren't analyzed if sharing is en…
…abled When connection sharing is enabled, sending queries longer than 16Mbyte lead to: * only the first 16Mbyte are analyzed if sharing is allowed * data after 16Mbyte isn't properly discarded. Change ------ - assembled payloads for payloads larger than 16Mbyte to allow lexing/parsing on them. Change-Id: Ibbfb7cc3d46fe2133040ad46370414badbb53f80
-
Bug#35493871 authentication with empty password and PHP fails
Running php -r 'mysqli_real_connect(mysqli_init(), "127.0.0.1", "empty", "");' through router fails with: PHP Warning: mysqli_real_connect(): Premature end of data (mysqlnd_wireprotocol.c:703) in Command line code on line 1 PHP Warning: mysqli_real_connect(): AUTH_RESPONSE packet 1 bytes shorter than expected in Command line code on line 1 Background ---------- PHP sends empty password as empty packet, while libmysqlclient sends '\0'. Change ------ - accept empty packet as "empty password" - added tests which call php with various authentication variants. Change-Id: I34ae07409b74b8ec7d5dbdf8939f3f7f6c9c3d0a
-
WL#15915 Deprecate INFORMATION_SCHEMA.PROCESSLIST
Use of INFORMATION_SCHEMA.PROCESSLIST is deprecated, use PERFORMANCE_SCHEMA.PROCESSLIST instead. Change-Id: I2564471ede42d9da1ccd61269b073e57dbee6fcf
Commits on Sep 4, 2023
-
Approved-by: Erlend Dahl <erlend.dahl@oracle.com>
-
Bug#35755396: audit_plugin: The chrono::steady_clock::now() usage causes
performance drop Fixed a bug in the scheduler logic: The mysql conditional variable timed wait is expecting an *absolute* time and the wrapper was giving it a relative time. This was causing the scheduler background loop to not wait at all and busyloop check the queue. Fixed by: - making sure the mysql conditional variable timed wait gets an absolute time and also fixed the unit test mock code to process the time passed to it as an absolute time. - Replaced the usage of steady_clock::now with system_clock::from_time_t(time(nullptr)). This cuts down on the timer resolution to 1sec, but is faster. - added a missing file in one of the guid unit tests on windows causing it to not be compiling on windows. - added one extra second to the wait for the top task to avoid some limited (to 1 sec) busy-looping on linux when executing the task. Addressed review comments. Change-Id: Id6cdd974384c4132470af7036e8920e50cbcecdc
-
Null merge branch 'mysql-5.7' into mysql-8.0
Vidhya Sree Rajuladevi committedSep 4, 2023 -
Bug #23036096 - INNODB LAST DETECTED DEADLOCK NOT SHOWING FULL QUERY
POST-PUSH FIX: - Includes fixing of failed test case Change-Id: I8dd955b6028bf6a032f760e8039155c9cc483f99
Vidhya Sree Rajuladevi committedSep 4, 2023
Commits on Sep 1, 2023
-
Bug#34306497 - Low limit heuristic used unnecessarily with descending
scans POST-PUSH FIX: Disabled warnings for a query in `engines/iuds.insert_calendar.test` because it behaves differently for Myisam engine and Innodb engine. Change-Id: Iafa765ca4851b0a286dd25ee4aad35998985dbe6
Nimita Joshi committedSep 1, 2023 -
Bug#35713721 redo log is never flushed
(backport from trunk to 8.0) Flushing logs was skipped on Windows, since the value of Log_file_handle::s_skip_fsyncs was not in line with what is written in innobase/os/os0file.cc . On Windows, log files are always opened without the FILE_FLAG_NO_BUFFERING Windows attribute, thus they always need flushing, regardless what the value of SRV_WIN_IO_UNBUFFERED is. Change-Id: Ia8a608ccaec58c84453cffbeaa80842184d27c47
-
Bug#35685515: Upgrade libfido2 to v1.13.0
Post-push fix: Added fido plugin check header to excludenoskip.list Change-Id: Ib7e071ab0d1c77811714e19b71ee04811c101624
-
BUG #35352168 : Debug build asserts in innodb.ddl_crash_alter_partiti…
…on test PROBLEM: - ALTER TABLE EXCHANGE PARTITIONS doesn't mark its transaction in undo log as DDL/DD operation. - Undo log entry is allocated earlier than we add items to DDL log or access DD. - Call for dropping SDI is done before exchanging partitions. FIX: - Delay the call for dropping SDI post exchanging partition. Change-Id: I2077cf330d1c9a6ac93b32b0f22243d685da066b
Mohammad Tafzeel Shams committedSep 1, 2023
Commits on Aug 31, 2023
-
Bug#35763112 Compile MySQL Cluster with clang 15
Compiler warnings occur when compiling MySQL Cluster 8.0.35 with clang 15 Description: /export/home2/tmp/magnus/mysql/8.0/storage/ndb/src/kernel/blocks/dbtup/ DbtupExecQuery.cpp:4774:14: error: argument 'sizes' of type 'Uint32[2]' (aka 'unsigned int[2]') with mismatched bound [-Werror,-Warray-parameter] Uint32 sizes[2], ^ /export/home2/tmp/magnus/mysql/8.0/storage/ndb/src/kernel/blocks/dbtup/ Dbtup.hpp:3968:28: note: previously declared as 'Uint32[4]' (aka 'unsigned int[4]') here Uint32 sizes[4], ^ 1 error generated. /export/home2/tmp/magnus/mysql/8.0/storage/ndb/src/kernel/blocks/dbtup/ DbtupTrigger.cpp:2414:52: error: argument 'sec' of type 'LinearSectionPtr[3]' with mismatched bound [-Werror,-Warray-parameter] LinearSectionPtr sec[3]) ^ /export/home2/tmp/magnus/mysql/8.0/storage/ndb/src/kernel/blocks/ dbtup/../dblqh/../dbtup/Dbtup.hpp:3369:52: note: previously declared as 'LinearSectionPtr[]' here LinearSectionPtr ptr[]); Fix: Backport relevant fixes from trunk for developer convenience. Change-Id: Ib095263b72a13865bbd2fff443af4a4fee16ff1f -
Bug#33542939 - mysqlsh util.loadDump failed to load DDL error Duplica…
…te entry '0' for key -- Backporting the fix for Bug#34846823 Change-Id: I0d14922709966c91d8c03620f0fae46a4f189957
Aditya A committedAug 31, 2023
Commits on Aug 30, 2023
-
Bug#35685515: Upgrade libfido2 to v1.13.0
Description: - Removed libfido2-1.8.0 source code - Removed libcbor-0.7.0 source code Change-Id: I542eac1fc4d98bea0c30ec9740cccac4b95dec51 Change-Id: Ia170809bf7ebe8ec1a954af2f4ac7dd69ea8e911