Use external tool for creating backtraces on crashes if available. (v3)xserver-cygwin-1.12.1-1

Based on a patch by Matthias Hopf mhopf at suse.de

On 08/10/2010 21:43, Alan Coopersmith wrote:
> Matthias Hopf wrote:
>> This calls /usr/bin/xorg-backtrace to create reasonable commented backtraces
>> with gdb. On errors it falls back to the generic method.
>
> There's already similar code in the backtrace.c file in the
> # ifdef HAVE_PSTACK case, where we fork the external pstack
> process on Solaris to get stack traces that have more information
> than the libc builtins allow, like getting full symbol tables from
> libelf.
>
> Could we make the fork/exec wrapper common between the two?   I'd
> even be okay with making the entire code in backtrace.c common and
> just letting the script handle calling gdb or pstack.   (Solaris
> systems are more likely to have pstack installed than gdb.)

Attached is an updated version of this patch which uses a common fork/exec
wrapper and looks for the backtrace script in $bindir.

I had to drop the use of fork1() and closefrom() as they aren't portable to my
platform of interest.  Unfortunately,  gdb can't backtrace through a signal
handler on cygwin, so this patch isn't as much use to me as I hoped it might
be, so I'll just leave it here :-)

From 8b19f11d6e5ab82e510cdcc94f2db322843d6c73 Mon Sep 17 00:00:00 2001
From: Matthias Hopf <mhopf-l3A5Bk7waGM-XMD5yJDbdMReXY1tMh2IBg@public.gmane.org>
Date: Fri, 8 Oct 2010 18:25:12 +0200
Subject: [PATCH] Use external tool for creating backtraces on crashes if available.

Make the code for fork/exec-ing a helper program to perform a backtrace
more generic, so it can be used to call /usr/bin/xorg-backtrace, a
script which invokes gdb to create reasonable backtraces.
On errors it falls back to the generic method.

Signed-off-by: Jon TURNEY <jon.turney@dronecode.org.uk>

0 files changed, 0 insertions, 0 deletions