[apport] metacity crashed with SIGSEGV in event_callback()

Attached X11 configuration.

StacktraceTop:event_callback (event=0xbfdb6ec8, data=0x8102188) at display.c:1840
filter_func (xevent=0xbfdb6ec8, event=0x80f5100, data=0x810dcf8) at ui.c:88
gdk_event_apply_filters (xevent=0xbfdb6ec8, event=0x80f5100, filters=0x90) at gdkevents-x11.c:343
gdk_event_translate (display=0x80f1098, event=0x147e5, xevent=0xbfdb6ec8, return_exposes=0)
_gdk_events_queue (display=0x80f1098) at gdkevents-x11.c:2252

Thank you for your bug. Do you have gtk-qt-engine from kubuntu installed? Does it happen without it?

On Thu, 2007-03-15 at 13:10 +0000, Sebastien Bacher wrote:

> Do you have gtk-qt-engine from kubuntu
> installed? Does it happen without it?

1. Yes:
mbt@pepper:~$ dpkg -l gtk-qt-engine
Desired=Unknown/Install/Remove/Purge/Hold
|
Status=Not/Installed/Config-files/Unpacked/Failed-config/Half-installed
|/ Err?=(none)/Hold/Reinst-required/X=both-problems (Status,Err:
uppercase=bad)
||/ Name Version Description
+++-=================-=================-==================================================
ii gtk-qt-engine 0.71~svn20070224- theme engine using Qt for GTK+
2.x
mbt@pepper:~$

2. That would be an affirmative. I removed gtk-qt-engine and the GTK2
package for it as well, and it still occured. When this happens, the X
server also crashes with Signal 11, and GDM goes to 100% CPU usage until
it is restarted (“sudo /etc/init.d/gdm stop && sleep 2 &&
sudo /etc/init.d/gdm start” is what I did).

Could you get a new crash file after restarting GNOME then?

Nope—apport doesn’t seem to be kicking off when I restart to let me know. Perhaps the Metacity crash went away and it is just X11 dying a horrible Sig11 death? I don’t know, and I suspect that if it were possible to try to trace everything starting with gdm and everything that was forked/exec’d from it, I probably would get nowhere, really quickly.

I am attaching /var/log/Xorg.0.log.old. Perhaps that may be of some help?

Thanks again, Sebastien!

Could you try to remove the crash file from /var/crash and note if apport creates a new one then?

A-ha! That did it. You know, all these new things are wonderful… once you start figuring out how they work and where they dump things. :-) New crash file attached from /var/crash.

I've opened a bug upstream: http://bugzilla.gnome.org/show_bug.cgi?id=422242

In case it's useful, another file with traces. Running with intel dual-head as well.

This bug was fixed in the package metacity - 1:2.23.55-0ubuntu1

---------------
metacity (1:2.23.55-0ubuntu1) intrepid; urgency=low

  * New upstream release
    - Add some extra null checks (LP: #92502)
    - Check for double-freeing at the time of workspace freeing (LP: #64818)
  * debian/control:
    - Add Vcs-Bzr header
    - Bump Standards-Version to 3.8.0

 -- Pedro Fragoso <email address hidden> Tue, 15 Jul 2008 01:17:00 +0100