httpd-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Dean Gaudet <>
Subject Re: cvs commit: apache-apr/pthreads/src/main http_main.c
Date Wed, 17 Feb 1999 18:24:59 GMT

On Wed, 17 Feb 1999, Ryan Bloom wrote:

> There are no longer ever any calls to pthreads functions from any signal
> handlers.  The only reason we ever had them, was for SIGTERM, which called
> clean_child_exit, but clean_child_exit never should have had any pthreads
> calls in it.  The underlying problem perhaps were stated incorrectly.  It
> manifests in really any signal I guess.  It is because of the accept
> serialization that was put in yesterday.  This problem has been resolved.
> I think I just commited the patch, but I can't remember.  If I don't get a
> message about it soon, I'll look into it.

I'm curious why you're handling signals in all threads.  Why not block
signals in all threads except one?  signals are mostly useless in
pthreaded programs, you can't control which thread will receive the signal
in a portable manner.

For example, that means SIGPIPE needs to be blocked completely, so that
you get a -1 and errno on read/write to a half-closed socket.  And you
can't use SIGALRM for timeouts because any thread could receive the


for example.


View raw message