Is there a chunk of dead code below this change or am I misreading? https://svn.apache.org/viewvc/httpd/httpd/trunk/server/mpm/event/event.c?revision=1916241&view=markup&pathrev=1916241#l2758
The first thread creation failure results in clean_child_exit(), so what's all this about starting the listener as long as we have some workers or AH03271 stuff about not taking over slots quickly? On Mon, Mar 11, 2024 at 4:36 PM <cove...@apache.org> wrote: > > Author: covener > Date: Mon Mar 11 20:35:27 2024 > New Revision: 1916241 > > URL: http://svn.apache.org/viewvc?rev=1916241&view=rev > Log: > event: avoid possible hang in clean_child_exit > > If the pthread_create failure isn't on the first worker thread, > another one is likely to hold the queue mutex already. The cleanup > of pchild will try to cleanup the queue and block on destroying > the condition. > > ST_UNGRACEFUL as we have no listener thread yet. > > > Modified: > httpd/httpd/trunk/server/mpm/event/event.c > > Modified: httpd/httpd/trunk/server/mpm/event/event.c > URL: > http://svn.apache.org/viewvc/httpd/httpd/trunk/server/mpm/event/event.c?rev=1916241&r1=1916240&r2=1916241&view=diff > ============================================================================== > --- httpd/httpd/trunk/server/mpm/event/event.c (original) > +++ httpd/httpd/trunk/server/mpm/event/event.c Mon Mar 11 20:35:27 2024 > @@ -2749,6 +2749,7 @@ static void *APR_THREAD_FUNC start_threa > APLOGNO(03104) > "ap_thread_create: unable to create worker > thread"); > /* let the parent decide how bad this really is */ > + signal_threads(ST_UNGRACEFUL); > clean_child_exit(APEXIT_CHILDSICK); > } > threads_created++; > > -- Eric Covener cove...@gmail.com
