GNOME Bugzilla – Bug 314442
Metacity crashes when opening Gimp dialog
Last modified: 2005-09-01 22:29:42 UTC
Version details: 2.11.3 Distribution/Version: Ubuntu Breezy current Metacity crashes and restarts itself when opening Script-Fu -> Shadow -> Drop Shadow in Gimp 2.2.8
Thanks for the bug report. Without a stack trace from the crash it's very hard to determine what caused it. Can you provide us with one? Please see http://live.gnome.org/GettingTraces for more information on how to do so.
May be related to bug 169684 (32 bit visuals cause Metacity to crash). I can't duplicate with gimp 2.0.5 and I can't get gimp to compile from CVS (either HEAD or with the GIMP_2_2_8 tag). Since I can't reproduce, we'd need a stack trace or debugging log as Teppo pointed out, though getting one for Metacity is a little bit different. See bug 169684 comment 1 for how to get the latter (which is probably easier); for just a stack trace you would need to run gdb in a separate virtual console and attach it to the running Metacity process before it crashes.
Created attachment 51335 [details] Debug log
Will it suffice just a debug log, or do I need to make a stack trace as well?
Well the debugging log merely showed that a "Script Fu" window was being centered onscreen before the log ended without showing any X errors, failed assertions, or backtraces. So, yes, we'll need a stack trace obtained with gdb as well.
The relevant ubuntu but is http://bugzilla.ubuntu.com/show_bug.cgi?id=14145: The stack trace is:
+ Trace 62650
Rebuilding metacity on ubuntu to get a debug stacktrace seems to fix the crash, and in fact just a plain re-build fixes the crash... There is also a report with this stacktrace in bug 311819 comment 3
*** Bug 314534 has been marked as a duplicate of this bug. ***
meta_window_place() doesn't call meta_accel_label_set_accelerator(); I'd like to know the real function it's crashing in as meta_window_place() is pretty huge... Also, since all the people getting this are on Debian/Ubuntu and I can't duplicate (bug 314534 even claimed it was 100% repeatable but I couldn't get it), I'd like to know if anyone can get this with the Debian/Ubuntu patches removed.
Brent Smith found and fixed this nasty crasher bug (well, provided the patch that fixes it; it hasn't been committed yet though it works great). It was an uninitialized variable, which explains why others could get the crash when it didn't happen for me doing the same thing. (When I re-ran Metacity under valgrind to test things out, I could indeed get the crash) *** This bug has been marked as a duplicate of 315000 ***