qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: "Daniel P. Berrange" <berrange@redhat.com>
To: Gerd Hoffmann <kraxel@redhat.com>
Cc: qemu-devel@nongnu.org
Subject: Re: [Qemu-devel] [PATCH] catch signals
Date: Wed, 6 Aug 2008 10:48:53 +0100	[thread overview]
Message-ID: <20080806094853.GF9055@redhat.com> (raw)
In-Reply-To: <48996CE7.5010205@redhat.com>

On Wed, Aug 06, 2008 at 11:20:39AM +0200, Gerd Hoffmann wrote:
> Daniel P. Berrange wrote:
> > On Tue, Aug 05, 2008 at 06:09:39PM +0200, Gerd Hoffmann wrote:
> >>  
> >> +#ifndef _WIN32
> >> +static void termsig_handler(int signal)
> >> +{
> >> +    switch (signal) {
> >> +    case SIGSEGV:
> >> +    case SIGBUS:
> >> +        /* returning from signal handler most likely isn't going to work */
> >> +        fprintf(stderr, "qemu: got signal %d (%s), taking emergency exit\n",
> >> +                signal, strsignal(signal));
> >> +        exit(1);
> > 
> > Neither of these functions are on the POSIX  async-signal-safe list,
> > so their use from signal handlers is not a good idea.
> 
> We are in dead water already and also don't plan to ever return from the
>  signal handler.  Is it really a problem then?

Yes, because if you further corrupt state in the signal handler it makes
debugging what went wrong even harder than it already is. In SEGV/BUS
case you really want to be able to get a decent core dump to analyse, so
the utmost care should be taken to avoid further messing up state. 

The STDIO libs are not re-entrant safe so if the original code was in a
STDIO function, and a SEGV comes in on another thread, the signal handler
will likely deadlock in a mutex. 'exit' will call into STDIO to flush
buffers, so will suffer the same problem.

> I want to be able to do cleanups (well, at least attempt) even in case
> of a segfault.  If exit() + atexit handlers isn't going to fill the bill
> we'll have to create some signal-save emergency cleanup handlers.

You could use 'sigaltstack' to setup an alternate pre-allocated stack
and carefully code the handler so it only uses pre-allocated memory,
or mem from the alt-stack & signal-safe functions. If you need something
really complex, you could 'fork' a cleanup program and communicate any
info to it using a pipe.

Regards,
Daniel
-- 
|: Red Hat, Engineering, London   -o-   http://people.redhat.com/berrange/ :|
|: http://libvirt.org  -o-  http://virt-manager.org  -o-  http://ovirt.org :|
|: http://autobuild.org       -o-         http://search.cpan.org/~danberr/ :|
|: GnuPG: 7D3B9505  -o-  F3C9 553F A1DA 4AC2 5648 23C1 B3DF F742 7D3B 9505 :|

  reply	other threads:[~2008-08-06  9:49 UTC|newest]

Thread overview: 11+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2008-08-05 16:09 [Qemu-devel] [PATCH] catch signals Gerd Hoffmann
2008-08-05 16:29 ` Samuel Thibault
2008-08-06  9:11   ` Gerd Hoffmann
2008-08-05 16:35 ` Daniel P. Berrange
2008-08-05 16:53   ` Samuel Thibault
2008-08-05 17:00     ` Daniel P. Berrange
2008-08-05 18:39       ` Jamie Lokier
2008-08-06  9:20   ` Gerd Hoffmann
2008-08-06  9:48     ` Daniel P. Berrange [this message]
2008-08-11 16:50 ` Ian Jackson
2008-08-11 19:43   ` Gerd Hoffmann

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20080806094853.GF9055@redhat.com \
    --to=berrange@redhat.com \
    --cc=kraxel@redhat.com \
    --cc=qemu-devel@nongnu.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).