mariadb/vio
Vladislav Vaintroub 01d994b39a Post-fix 567c097359
Do *not* check if socket is closed by another thread. This is
race-condition prone, unnecessary, and harmful. VIO state was introduced
to debug the errors, not to change the behavior.

Rather than checking if socket is closed, add a DBUG_ASSERT that it is
*not* closed, because this is an actual logic error, and can potentially
lead to all sorts of funny behavior like writing error packets to Innodb
files.

Unlike closesocket(), shutdown(2) is not actually race-condition prone,
and it breaks poll() and read(), and it worked for longer than a decade,
and it does not need any state check in the code.
2024-03-18 23:43:56 +01:00
..
docs MDEV-17926 FederatedX TODO is obsolete 2018-12-12 00:31:04 +01:00
CMakeLists.txt Merge branch '10.3' into 10.4 2019-05-19 20:55:37 +02:00
test-ssl.c Update FSF Address 2019-05-11 21:29:06 +03:00
test-sslclient.c Update FSF Address 2019-05-11 21:29:06 +03:00
test-sslserver.c Update FSF Address 2019-05-11 21:29:06 +03:00
vio.c MDEV-33582 Add more warnings to be able to better diagnose network issues 2024-03-05 20:19:49 +02:00
vio_priv.h Merge 10.4 into 10.5 2020-08-01 14:42:51 +03:00
viopipe.c MDEV-23348 vio_shutdown does not prevent later ReadFile on named pipe 2020-08-03 13:23:38 +02:00
viosocket.c Post-fix 567c097359 2024-03-18 23:43:56 +01:00
viossl.c Merge 10.5 into 10.6 2021-03-27 18:03:03 +02:00
viosslfactories.c Merge branch '10.5' into 10.6 2022-11-02 14:33:20 +01:00
viotest-ssl.c Update FSF Address 2019-05-11 21:29:06 +03:00
viotest-sslconnect.cc Update FSF Address 2019-05-11 21:29:06 +03:00
viotest.cc Update FSF Address 2019-05-11 21:29:06 +03:00