This change primarily splits SAPI deactivation to module and destroy
parts. The reason is that currently some SAPIs might bail out
on deactivation. One of those SAPI is PHP-FPM that can bail out on
request end if for example the connection is closed by the client
(web sever). The problem is that in such case the resources are not
freed and some values reset. The most visible impact can have not
resetting the PG(headers_sent) which can cause erorrs in the next
request. One such issue is described in #77780 bug which this fixes
and is also cover by a test in this commit. It seems reasonable
to separate deactivation and destroying of the resource which means
that the bail out will not impact it.
With request timeouts configured, php-fpm occasionally prints the
following warning:
WARNING: failed to acquire scoreboard
This is happens when php-fpm checks the child scoreboards for timeouts,
but fails to acquire a lock immediately. As this can (and does) occur
during normal operation, this commit downgrades this to a notice.
Closes#9019.
We revert the commits which caused this regression from the PHP-8.0 and
PHP-8.1 branches for now. We keep it in "master" because of PR #8833
which may offer a proper fix without BC break.
Extensions may (and do) write to stderr in mshutdown and similar. In
the best case, with the stderr stream closed, it's just swallowed.
However, some libraries will do things like try to detect color, and
these will outright fail and cause an error path to be taken.
Extensions may (and do) write to stderr in mshutdown and similar. In
the best case, with the stderr stream closed, it's just swallowed.
However, some libraries will do things like try to detect color, and
these will outright fail and cause an error path to be taken.
This issue might happen if there is change of the fcgi stream when
the buffer is full. Then the empty record is created which signals
end of stream which is incorrect.
The actual fix without a test was contributed by GitHub user @loveharmful
in GH-3198.
This change introduces subsequent kill of the process when idle process quit
(SIGQUIT) does not succeed. It can happen in some situations and means that FPM
is not able to scale down in dynamic pm. Using SIGKILL fixes the issue.
The fix introduces early locking of scoreboard when it is updated
which prevents the race condition causing an incorrect number of
active processes being set.
