metacity crashed with SIGABRT in raise()

Reproduced. Setting to triaged/high and targeting to Oneiric

I tried this on the current daily in kvm. What I see is that the first time I start a session, it crashes almost immediately, and dmesg and apport report a segfault of lightdm (which tears down the user session as well). But when I restart lightdm, and log in again (ubuntu 2d), it works. Is that what you are seeing, or do you have different symptoms?

I also tried "metacity --replace" locally, and it doesn't crash. Jean-Baptiste, how did you reproduce this?

Thanks!

@Martin: you are correct -- it happens on _first_ login; but I do not see LightDM segfaulting (I do see other LightDM sigsegv's, but they happened minutes to hours after logging in). In my case (both LightDM and GDM) the login completes successfully.

I did not try 'metacity --replace': I am running compiz...

I just marked bug 792999 a duplicate of this bug, after comparing https://bugs.launchpad.net/ubuntu/+source/metacity/+bug/792999/comments/5 there to https://bugs.launchpad.net/ubuntu/+source/metacity/+bug/797078/+attachment/2168109/+files/metacity.gdb here.

"Apport retracing service" had previously marked bug 792999 as a duplicate of private bug 793981. So private bug 793981 is probably the same as this bug, too. Maybe someone here with access to private bug 793981 could check (and if appropriate, mark it as such).

I am using Gnome-shell and GDM. This also happening to me but and i am not using LightDM. I get errors when logging in to Gnome-classic but it doesn't crash. Also as long as i do not touch the activities button gnome-shell works.

for me, only append when I launch gns3 and when I try to re-launch metacity, I obtain this:

mmoussouni@VAL1-ubuntu:~$ metacity
Avertissement du gestionnaire de fenêtres : La fenêtre 0x3e00005 (GNS3) positionne un indice MWM indiquant qu'elle n'est pas redimensionnable, mais positionne une taille minimale de 16383 x 16383 et une taille maximale de 2147483647 x 2147483647 ; ceci n'a pas beaucoup de sens.
Avertissement du gestionnaire de fenêtres : La fenêtre 0x3e00005 (GNS3) positionne un indice MWM indiquant qu'elle n'est pas redimensionnable, mais positionne une taille minimale de 16383 x 16383 et une taille maximale de 2147483647 x 2147483647 ; ceci n'a pas beaucoup de sens.
Anomalie dans le gestionnaire de fenêtres : Unexpected X error: BadAlloc (insufficient resources for operation) serial 3165 error_code 11 request_code 53 minor_code 0)
Abandon (core dumped)

I experienced this when launching XChat tonight though I'm not sure if that was just coincidence. I'm running Oneiric with all the updates available from this morning.

This bug happens to be similar to Bug 823386 in Compiz (both crashed using the same signal in the same function, though the Compiz issue happens to be caused by trying to attempt the workaround to Bug 685552...)

Also confirmed crash when running GNOME Shell (It seems as though Metacity is trying to start with Mutter running...)

For me, metacity crashed when starting this PyQt script (which draws a window with a toolbar and a textedit).

Simplified test case.

The same happened to me twice.

I had a crash in metacity but when the crash report was generated it
found a crash in akonadi. What's with this??

On 09/11/2011 10:14 AM, Luiz Santos wrote:
> The same happened to me twice.
>

The same happened to me every time I start using my pc after a shutdown.

Had a compiz crash in 3D unity so tried in 2D. When I launch Calibre or FBReader, Metacity crashes just as Compiz does in 3D

@Jason: for crashes with Qt4 apps, see also bug 805303.

I ran this with --sync and also ran it through xtrace, which shows we are trying to delete a property on an already destroyed window, and this was introduced with the fix for bug 696864.

Uploading a fix in a few moments...

This bug was fixed in the package metacity - 1:2.34.1-1ubuntu4

---------------
metacity (1:2.34.1-1ubuntu4) oneiric; urgency=low

  * Fix LP: #797078 - metacity crashed with SIGABRT due to unhandled X error.
    Update previous patch to trap errors when accessing properties whilst
    handling an UnmapNotify event, just in case the window has already
    been destroyed
    - update debian/patches/16-capture-before-unmap.patch
 -- Chris Coulson <email address hidden> Fri, 16 Sep 2011 21:21:58 +0100

I previously reported bug 844849, which was marked as a duplicate of this one.

Now there are at least 4 identical bugs: bug 846621, private bug 848808, bug 849781, bug 855603. Maybe it's a separate issue or this one was not fixed properly?

Just updated, but it still affects my computer. Maybe the fix is not in standard repositories (I use spanish one)