gnome-settings-daemon crashed with signal 5 in _XReply()

StacktraceTop:
 XGetWindowAttributes (dpy=0x11bea10, w=<optimized out>, attr=<optimized out>) at ../../src/GetWAttrs.c:149
 xkl_engine_select_input_merging (engine=0x132ada0, win=37748737, mask=6291456) at xklavier.c:459
 xkl_engine_process_create_window_evt (engine=0x132ada0, cev=0x7fff5fed5240) at xklavier_evt.c:465
 xkl_engine_filter_events (engine=0x132ada0, xev=0x7fff5fed5240) at xklavier_evt.c:52
 gsd_keyboard_xkb_evt_filter (xev=<optimized out>, event=<optimized out>) at gsd-keyboard-xkb.c:735

Status changed to 'Confirmed' because the bug affects multiple users.

I think I've just run into this bug; we'll wait for the retracer to confirm it. But this looks like a recent regression somewhere in the X stack that's causing a lot of duplicates, so marking as 'high' accordingly.

Chris, you debugged similar issues before, do you remember what was useful in such cases? Do we need a sync version of the stacktrace or the .xsession-errors after the issue?

(note that --sync got replaced by GDK_SYNCHRONIZE=1 use nowadays, if a sync stacktrace is required)

it seems that the issue started some weeks ago but g-s-d didn't change a lot, could be something else in the stack...

The crash is because XGetWindowAttributes is throwing a BadMatch error, which is unexpected. It will throw a BadMatch error (from dixLookupWindow) if the specified window isn't really a window. This might happen if a window is destroyed and replaced with a drawable of the same ID whilst gnome-settings-daemon is handling the original CreateNotify event.

This looks like it's basically the same issue as https://bugs.freedesktop.org/show_bug.cgi?id=23562, except with XGetWindowAttributes instead of XGetWindowProperty.

Perhaps XGetWindowAttributes needs a similar fix?

(Subscribing the other Chris for a second opinion)

Urgh. It looks like the patch you originally submitted upstream is more correct than the patch that got committed - all manner of window calls are broken, not just XGetWindowAttributes.

Let's see if the xorg-devel mailing list agrees.

Discussion on http://lists.x.org/archives/xorg-devel/2012-January/028361.html

I have not had gnome session daemon crash for a long time. In the past it was due to obsolete packages and so was not reportable, but easily fixed. This time I ran Update Manager to get the daily updates. This completed without any problems but as it included a kernel update, I rebooted and as soon as the desktop appeared Gnome setting daemon crashed. And this time it was reportable.

Bug 907052 may be a duplicate of this one.

Bug 907052 looks like it's a different issue

closing the g-s-d part it seems an xorg issue and the fix is pending commit

This bug was fixed in the package xorg-server - 2:1.11.3-0ubuntu8

---------------
xorg-server (2:1.11.3-0ubuntu8) precise; urgency=low

  * debian/patches/224_return_BadWindow_not_BadMatch.diff:
    - Fix the error values of a whole slew of window-related calls.
      Fixes some gnome-settings-daemon aborts in XLib (LP: #903973)
 -- Christopher James Halse Rogers <email address hidden> Mon, 23 Jan 2012 16:09:29 +1100