gnome-settings-daemon crashed with SIGSEGV in g_variant_unref()

StacktraceTop:
 g_variant_unref (value=0x0) at /build/buildd/glib2.0-2.29.8/./glib/gvariant-core.c:617
 g_settings_get_boolean (settings=0x0, key=0x810e222 "enable") at /build/buildd/glib2.0-2.29.8/./gio/gsettings.c:1955
 maybe_show_status_icon (manager=0x97275e8) at gsd-a11y-keyboard-manager.c:445
 g_cclosure_marshal_VOID__STRING (closure=0x9794b20, return_value=0x0, n_param_values=2, param_values=0x97e2540, invocation_hint=0xb5afc5c0, marshal_data=0x0) at /build/buildd/glib2.0-2.29.8/./gobject/gmarshal.c:498
 g_closure_invoke (closure=0x9794b20, return_value=0x0, n_param_values=2, param_values=0x97e2540, invocation_hint=0xb5afc5c0) at /build/buildd/glib2.0-2.29.8/./gobject/gclosure.c:771

On first reboot right after on-line upgrade on Jul 11 from Natty to Oneiric.

I haven't been able to replicate, but I think I have a fix, so can anyone attach his/her ~/.xsession-errors just after g-s-d crashes, please?

This bug was fixed in the package gnome-settings-daemon - 3.1.3-0ubuntu5

---------------
gnome-settings-daemon (3.1.3-0ubuntu5) oneiric; urgency=low

  * debian/patches/01_git_shorter_syndaemon_delay.patch:
    - Remove this patch, it fixes the annoyance for some users and breaks it
      for others
  * debian/patches/17_react_when_streams_removed.patch:
    - Remove patch not being applied and already upstream
  * debian/patches/01_git_a11y_keyboard_cleanup.patch:
    - Do proper cleanup when the plugin is stopped (LP: #804946)
  * debian/patches/06_use_application_indicator.patch:
    - Rebased
 -- Rodrigo Moya <email address hidden> Tue, 12 Jul 2011 13:04:51 +0200

I still see the issue with 3.1.3-0ubuntu7.

Yes, so bug #814635 is not a duplicate of this bug.

Rodrigo, it seems some users still get that issue?

Ok, so can anyone still experiencing this paste their ~/.xsession-errors file?

Here you go. It's rather large so I have attached it in a file xsession-errors.txt

Hey Pedro, could you upstream that one?

This should hopefully be fixed by the latest libglib upload (2.29.90), so can people please install that and check if it still crashes?

I haven't seen it in a while,
but it was really bad to my CPU,
potentially overheating, and making
using the machine a royal pain (ITA)
and not killable as off some gdm or other
process, so if it returns,
I'll scream bloody murder again ;-/
but until then all is right, relatively
right in the world... ;-/

On Wed, Sep 14, 2011 at 7:29 AM, Rodrigo Moya
<email address hidden> wrote:
> This should hopefully be fixed by the latest libglib upload (2.29.90),
> so can people please install that and check if it still crashes?
>
> ** Changed in: gnome-settings-daemon (Ubuntu)
>       Status: Triaged => Incomplete
>
> --
> You received this bug notification because you are subscribed to a
> duplicate bug report (819972).
> https://bugs.launchpad.net/bugs/804946
>
> Title:
>  gnome-settings-daemon crashed with SIGSEGV in g_variant_unref()
>
> Status in Gnome Settings Daemon:
>  New
> Status in “gnome-settings-daemon” package in Ubuntu:
>  Incomplete
>
> Bug description:
>  crashed while logging in.
>
>  ProblemType: Crash
>  DistroRelease: Ubuntu 11.10
>  Package: gnome-settings-daemon 3.1.2-0ubuntu1
>  ProcVersionSignature: Ubuntu 3.0-2.3-generic 3.0.0-rc4
>  Uname: Linux 3.0-2-generic i686
>  Architecture: i386
>  Date: Sat Jul  2 19:25:09 2011
>  ExecutablePath: /usr/lib/gnome-settings-daemon/gnome-settings-daemon
>  InstallationMedia: Ubuntu 10.10 "Maverick Meerkat" - Release i386 (20101007)
>  ProcCmdline: /usr/lib/gnome-settings-daemon/gnome-settings-daemon
>  ProcEnviron:
>   LANG=es_ES.UTF-8
>   SHELL=/bin/false
>   PATH=(custom, no user)
>  SegvAnalysis:
>   Segfault happened at: 0xe43c73 <g_variant_unref+35>: lock xadd %eax,0x14(%esi)
>   PC (0x00e43c73) ok
>   source "%eax" ok
>   destination "0x14(%esi)" (0x00000014) not located in a known VMA region (needed writable region)!
>   Stack memory exhausted (SP below stack segment)
>  SegvReason: writing NULL VMA
>  Signal: 11
>  SourcePackage: gnome-settings-daemon
>  StacktraceTop:
>   g_variant_unref (value=0x0) at /build/buildd/glib2.0-2.29.8/./glib/gvariant-core.c:617
>   g_settings_get_boolean (settings=0x0, key=0x810e222 "enable") at /build/buildd/glib2.0-2.29.8/./gio/gsettings.c:1955
>   ?? () from /usr/lib/gnome-settings-daemon-3.0/liba11y-keyboard.so
>   g_cclosure_marshal_VOID__STRING (closure=0x9794b20, return_value=0x0, n_param_values=2, param_values=0x97e2540, invocation_hint=0xb5afc5c0, marshal_data=0x0) at /build/buildd/glib2.0-2.29.8/./gobject/gmarshal.c:498
>   g_closure_invoke (closure=0x9794b20, return_value=0x0, n_param_values=2, param_values=0x97e2540, invocation_hint=0xb5afc5c0) at /build/buildd/glib2.0-2.29.8/./gobject/gclosure.c:771
>  Title: gnome-settings-daemon crashed with SIGSEGV in g_variant_unref()
>  UpgradeStatus: Upgraded to oneiric on 2011-06-15 (17 days ago)
>  UserGroups:
>
> To manage notifications about this bug go to:
> https://bugs.launchpad.net/gnome-settings-daemon/+bug/804946/+subscriptions
>

--
Jerry W

Anyone else with the latest glib?

This should be fixed in libglib2.0-0 2.29.92, so closing. Please reopen if it happens again after having upgraded

Link the right upstream bug for this crash.

Hy there!

Bug 865115 was appeared the same time as this bug, can it be related?

This bug was fixed in the package glib2.0 - 2.30.0-0ubuntu4

---------------
glib2.0 (2.30.0-0ubuntu4) oneiric; urgency=low

  [ Rodney Dawes ]
  * debian/patches/10_gvariant_check_null.patch:
    - Check that value is NULL before continuing (LP: #804946)

  [ Michael Terry ]
  * debian/libglib2.0-0.postrm.in:
    - Only clear the gsettings schema cache on purge (this also helps us
      deal with the above bug LP: #804946)
 -- Michael Terry <email address hidden> Thu, 06 Oct 2011 10:05:02 -0400