2388 Commits

Author SHA1 Message Date
f288f86a71 Merge branch '3.3' into 3.4 2025-08-04 18:36:43 +02:00
c80f221f1c CONC-783 fixup - make sure that proxy header is not sent twice
in case of TLS connection.
2025-08-04 16:53:54 +02:00
0ad0cc0f42 Merge branch '3.3' into 3.4 2025-08-04 15:17:37 +02:00
c71865eca4 CONC-783 Fix potential loss of "Proxy header not accepted from host" error
Send the proxy header and handshake response in a single write.

If the client uses two separate send() calls, the server's error message
"Proxy header not accepted from host" may be lost. This occurs because
the server sends a TCP RST (reset) instead of a FIN if it closes the socket
while the client is still sending data.

As a result, the client may receive ECONNRESET or EPIPE, without seeing
the actual error from the server.
2025-08-04 15:17:11 +02:00
b790c6c149 bump the VERSION 2025-07-22 23:07:00 +02:00
ba3a172dfd Merge branch '3.3' into 3.4 2025-07-17 15:04:50 +02:00
77bdf5a572 Fixed double close (introduced in PR275) 2025-07-17 10:36:37 +02:00
e240810b38 fix: fixed the possible nullPointerAritheticOutOfMemory by using
SET_CLIENT_ERROR and return 1;
because jumping to the error label is not viable in this case as pos is
not initialized and it could raise more errors, this way we still handle
the error while not clashing with the original error handling
implementation
2025-07-17 09:15:04 +02:00
1edef5b207 fix: added a failsafe return 1; in case the malloc at 559 fails so that
we don't work with NULL
2025-07-17 09:14:53 +02:00
f2117d4225 fix: fixed the posibility of undefined behaviour of the strchr previously
at 252 if the strdup at 248 fails and thus returns NULL by
setting a client error signalling that the client ran out of
memory for easier communication of what exactly failed during the
loading of the plugins and why. Returning from the function just after
2025-07-17 09:14:53 +02:00
f72654e324 fix: fixed a memory leak due to the socket not being closed after
encountering an error by closing the socket before return
2025-07-17 09:14:53 +02:00
9e4042c149 fix: fixed a possible leak when one of the files (fp1 or fp2) fails to
open by adding a simple condition that checks whether one of them opens
while the other fails and closing it
and added the extra prentheses for the assignment inside the if statement
for easier debugging and to avoid static analysis flagging it as a wrong
statement inside the if clause
2025-07-17 09:14:53 +02:00
c10a3479fb fix: added the closing of the test_file before various fail statements
to avoid memory leaks
2025-07-17 09:14:53 +02:00
056f09bb06 fix: added frees for any allocated memory if the allocation of other
memory fails and the FAIL_UNLESS calls needed the stop the test if this
happens
2025-07-17 09:14:53 +02:00
7009f60416 fix: added the closing of the file descriptor if the check_mysql_rc
macro will return FAIL to avoid memory leaks
2025-07-17 09:14:53 +02:00
9fb319a7e7 fix: added the necessary frees for the bind var before any of the
check_stmt_rc() and check_mysql_rc() macros potentialy return from the
fuction
2025-07-17 09:14:53 +02:00
92a8b3b529 fix: fixed the posible read of NULL when strdup fails thus resulting in
rpl->filename being NULL by setting an error signalling that we ran out
of memory to ease the debugging of the mariadb_rpl_options function and make
it clearer if it ever fails because of this. Also added va_end in this case to
avoid the va_list (ap) leaking when we return from this function and
then returning 1 from the fucntion.
2025-07-17 09:14:53 +02:00
5ea5d7ae01 fix: fixed the order of the mariadb_free_rpl_event() and rpl_set_error()
calls to avoid using the rpl_even var after freeing it
2025-07-17 09:14:53 +02:00
06e041cfa2 Added the FAIL_UNLESS call for the bind variable that is being allocated
to handle the possibility of the calloc failing and ensure we are not
working with NULL in the rest of the tests
2025-07-17 09:14:53 +02:00
0fc940259a Added the call to FAIL_UNLESS to handle the possibility of the
allocation of buffer failing to ensure we don't work with NULL further
in the test, also added a free to bind in this case to avoid a memory
leak
2025-07-17 09:14:53 +02:00
ba0da1df94 Merge pull request #289 from qobood/3.3
Fix miscellaneous typos
2025-07-16 16:56:05 +02:00
9d77fe8321 bump the VERSION 2025-07-16 12:49:15 +02:00
40be423f08 Fix miscellaneous typos 2025-07-13 18:42:30 +03:00
cacd251e20 Merge pull request #275 from PavolSloboda/static-analysis-result
Static analysis result bug fixes
2025-07-10 09:56:57 +02:00
dad72b6342 Merge branch '3.3' into 3.4 2025-07-10 09:45:46 +02:00
c58dce932a Merge pull request #279 from grooverdan/static_analsis_fix
Static analysis fix
2025-07-10 09:04:02 +02:00
f3d862759b Merge pull request #281 from grooverdan/caching_sha2_pw_err
caching_sha2_pw: free filebuffer and cleanup on Public key import failure
2025-07-10 05:54:49 +02:00
96b73e2052 Merge pull request #282 from grooverdan/win_get_password
get_tty_password(Windows) - CreateFile error
2025-07-09 11:52:15 +02:00
0eb34a2619 Merge pull request #267 from ericherman/eherman-io-size-20241231
Create #define for network buffer alignment
2025-07-09 11:48:32 +02:00
67972e131d ci-fix: Test 10.x mtr, not 11.4 and above 2025-06-25 09:52:18 +02:00
a7a3db7149 ci fixes:
- Don't set C/C unittest environment variables globally
- Removed 10.6 and 10.11 mtr tests
2025-06-25 09:51:44 +02:00
f410977447 Add diagnostic output for ci test 2025-06-25 09:51:29 +02:00
92b09b6727 [TODO-5373] add macos to non mandatory for now 2025-06-25 09:51:13 +02:00
5b32575fc1 [TODO-5373] remove node.js installation 2025-06-25 09:51:04 +02:00
0a3f217396 [TODO-5373] add github action 2025-06-25 09:50:30 +02:00
d3facb7ca9 ci fixes:
- Don't set C/C unittest environment variables globally
- Removed 10.6 and 10.11 mtr tests
2025-06-25 06:27:46 +02:00
189c2318b5 Add diagnostic output for ci test 2025-06-24 10:53:01 +02:00
29c43ff92c Merge pull request #285 from rusher/3.4
[TODO-5373] add github action
2025-06-24 06:18:30 +02:00
3139702bfb [TODO-5373] add macos to non mandatory for now 2025-06-20 15:14:30 +02:00
6774fb9da5 [TODO-5373] remove node.js installation 2025-06-18 16:09:13 +02:00
1bf3571146 [TODO-5373] add github action 2025-06-16 19:43:16 +02:00
b3adb24128 Merge branch '3.3' into 3.4 2025-06-07 16:50:38 +02:00
aebe28b8eb CONC-778: TLSv1.3 support for Windows Server 2022
Windows Server 2022 uses version number 10.0.20348, which is less than 10.0.22000.
Therefore, instead of checking for build numbers >= 22000 (which would exclude Server 2022),
we must check for build numbers >= 20348 to correctly include both Windows Server 2022
and Windows 11.

This is safe because TLSv1.3 was first supported in:
- Windows 11 (starting with build 22000)
- Windows Server 2022 (starting with build 20348)

Earlier versions did not support TLSv1.3:
- Windows 10 (up to build 19044)
- Windows Server 2019 (latest build 17763)
2025-06-07 15:32:44 +02:00
d226315ef8 get_tty_password(Windows) - CreateFile error
The Windows error condition for failing to open a file is
INVALID_HANDLE_VALUE is returned.
2025-05-29 16:40:50 +10:00
6fde019bfc caching_sha2_pw: free filebuffer and cleanup on Public key import failure 2025-05-29 15:42:34 +10:00
f431c6eaef CONC-776: enable rpl_set_error 2025-05-27 16:48:48 +10:00
bd233f8ac0 break; after return never reached 2025-05-27 16:48:39 +10:00
d5d330f82a mariadb_reconnect: reconnection may fail to allocate extensions
If this the case, tmp_mysql.extensions will be a null pointer
and the rest of the reconnection cannot continue.
2025-05-27 16:48:39 +10:00
b22b85bcdf mysql_init: early alloc failure not freed
Its possible for the net extensions allocation to
succeed but the other extensions to fail to allocate.

In this later case free the net alternatives explicitly
if we where going to free the entire allocation.
2025-05-27 16:48:39 +10:00
ba69f71f7d Added the call to FAIL_UNLESS to handle the possibility of the
allocation of buffer failing to ensure we don't work with NULL further
in the test, also added a free to bind in this case to avoid a memory
leak
2025-05-26 12:27:39 +02:00