gnome-shell crashed with SIGTRAP from error "Can't create a GtkStyleContext without a display connection" in gtk_style_context_init

[Hans-Petter Falao]

GNOME Shell isn't meant to use GTK at all, but it's repeatedly crashing in GTK since the introduction of GNOME 40.

https://errors.ubuntu.com/problem/45d4ce4e8f26828ab6993261e8a20208ecf261b0
https://errors.ubuntu.com/problem/e60359af65f32c18a4a6ccc93fd3469b0e675844
https://errors.ubuntu.com/problem/73a93ba578f68516e680373ae324db01336e8bad
https://errors.ubuntu.com/problem/d69ccaf1379f588b04ecac2a6cc93b9be31100b4

This seems to be a result of Xwayland crashing and gnome-shell then trying to recover (and failing). Although gnome-shell shouldn't be using GTK, and Wayland sessions shouldn't be dependent on having an X11 server (Xwayland) available. Fixing either of those should resolve this.

WORKAROUND:

sudo apt remove libcanberra-gtk3-module

TEST CASE:

1. Log into a Wayland session.
2. Open a Terminal.
3. $ xrandr # Just to ensure Xwayland starts
4. $ killall Xwayland

Expected: gnome-shell still responds.
Observed: gnome-shell freezes or exits.

---

ProblemType: Crash
DistroRelease: Ubuntu 22.04
Package: gnome-shell 40.5-1ubuntu2
ProcVersionSignature: Ubuntu 5.13.0-19.19-generic 5.13.14
Uname: Linux 5.13.0-19-generic x86_64
ApportVersion: 2.20.11-0ubuntu72
Architecture: amd64
CasperMD5CheckResult: pass
CrashCounter: 1
CurrentDesktop: ubuntu:GNOME
Date: Fri Oct 29 15:11:43 2021
DisplayManager: gdm3
ExecutablePath: /usr/bin/gnome-shell
InstallationDate: Installed on 2021-10-27 (2 days ago)
InstallationMedia: Ubuntu 22.04 LTS "Jammy Jellyfish" - Alpha amd64 (20211023)
ProcCmdline: /usr/bin/gnome-shell
ProcEnviron:
 LANG=nb_NO.UTF-8
 LANGUAGE=nb_NO:nb:no_NO:no:nn_NO:nn:en
 PATH=(custom, no user)
 SHELL=/bin/bash
 XDG_RUNTIME_DIR=<set>
RelatedPackageVersions: mutter-common 40.5-1ubuntu2
Signal: 5
SourcePackage: gnome-shell
StacktraceTop:
 g_log_writer_default () from /lib/x86_64-linux-gnu/libglib-2.0.so.0
 g_log_structured_array () from /lib/x86_64-linux-gnu/libglib-2.0.so.0
 g_log_structured_standard () from /lib/x86_64-linux-gnu/libglib-2.0.so.0
 ?? () from /lib/x86_64-linux-gnu/libgtk-3.so.0
 g_type_create_instance () from /lib/x86_64-linux-gnu/libgobject-2.0.so.0
Title: gnome-shell crashed with signal 5 in g_log_writer_default()
UpgradeStatus: No upgrade log present (probably fresh install)
UserGroups: adm cdrom dip lpadmin lxd plugdev sambashare sudo
separator:

[Apport retracing service (apport)]

StacktraceTop:
 g_log_writer_default (log_level=<optimized out>, fields=0x7ffeb0f37560, n_fields=6, user_data=0x0) at ../../../glib/gmessages.c:556
 g_log_structured_array (n_fields=6, fields=0x7ffeb0f37560, log_level=G_LOG_LEVEL_ERROR) at ../../../glib/gmessages.c:1972
 g_log_structured_array (log_level=G_LOG_LEVEL_ERROR, fields=0x7ffeb0f37560, n_fields=6) at ../../../glib/gmessages.c:1945
 g_log_structured_standard (log_domain=log_domain@entry=0x7f2752a4c063 "Gtk", log_level=log_level@entry=G_LOG_LEVEL_ERROR, file=file@entry=0x7f2752ab20e0 "../../../../gtk/gtkstylecontext.c", line=line@entry=0x7f2752a6be7c "348", func=func@entry=0x7f2752ab2c30 <__func__.70> "gtk_style_context_init", message_format=message_format@entry=0x7f2752ab2400 "Can't create a GtkStyleContext without a display connection") at ../../../glib/gmessages.c:2029
 gtk_style_context_init (context=0x559c3bcdd5d0) at ../../../../gtk/gtkstylecontext.c:348

[Daniel van Vugt (vanvugt)]

Unfortunately:

 * we don't have all the debug symbols in that crash

 * I can't find any reports of similar crashes from other people

 * Many of your installed packages are from [origin: unknown]

If you are using any PPAs then please remove them from the system. Please also update your system by:

  sudo apt update
  sudo apt full-upgrade

and then reboot.

Next time a crash happens, please open a new bug. Hopefully we can match this up to some future crash, or it may never happen again after the packages are updated.

[Daniel van Vugt (vanvugt)]

Please also ensure you don't have the 'proposed' channel enabled.

[Daniel van Vugt (vanvugt)]

Sorry I was focussing on the attached files when comment #2 actually looks quite detailed.

[Daniel van Vugt (vanvugt)]

I wonder if this bug might be caused by missing the fix:

  https://gitlab.gnome.org/GNOME/gdm/-/merge_requests/156

[Daniel van Vugt (vanvugt)]

Also sounds like https://gitlab.gnome.org/GNOME/mutter/-/issues/1475 ... that was closed without a fix because it was found to be caused by a faulty extension.

[Daniel van Vugt (vanvugt)]

Incomplete because I would like to see this crash occur on a clean system without so many unknown packages installed. Although if it turns out to be bug 1948617 then it certainly warrants attention.

[Apport retracing service (apport)]

StacktraceTop:
 g_log_writer_default (log_level=<optimized out>, fields=0x7ffeb0f37560, n_fields=6, user_data=0x0) at ../../../glib/gmessages.c:556
 g_log_structured_array (n_fields=6, fields=0x7ffeb0f37560, log_level=G_LOG_LEVEL_ERROR) at ../../../glib/gmessages.c:1972
 g_log_structured_array (log_level=G_LOG_LEVEL_ERROR, fields=0x7ffeb0f37560, n_fields=6) at ../../../glib/gmessages.c:1945
 g_log_structured_standard (log_domain=log_domain@entry=0x7f2752a4c063 "Gtk", log_level=log_level@entry=G_LOG_LEVEL_ERROR, file=file@entry=0x7f2752ab20e0 "../../../../gtk/gtkstylecontext.c", line=line@entry=0x7f2752a6be7c "348", func=func@entry=0x7f2752ab2c30 <__func__.70> "gtk_style_context_init", message_format=message_format@entry=0x7f2752ab2400 "Can't create a GtkStyleContext without a display connection") at ../../../glib/gmessages.c:2029
 gtk_style_context_init (context=0x559c3bcdd5d0) at ../../../../gtk/gtkstylecontext.c:348

[Daniel van Vugt (vanvugt)]

Interesting the above attachments show the problem originating from:

#29 0x00007f2746b35bfe in free_sound_event (d=0x7f273808bf20) at /build/libcanberra-NELJAC/libcanberra-0.30/src/canberra-gtk-module.c:184

which is kind of similar to https://gitlab.gnome.org/GNOME/mutter/-/issues/1475 where GTK is the problem.

What gnome-shell extensions are you using?

[Daniel van Vugt (vanvugt)]

The instruction addresses (modulo 16 at least) seem to match up too well to be a coincidence so I believe this is:

https://errors.ubuntu.com/problem/45d4ce4e8f26828ab6993261e8a20208ecf261b0
https://errors.ubuntu.com/problem/e60359af65f32c18a4a6ccc93fd3469b0e675844
https://errors.ubuntu.com/problem/73a93ba578f68516e680373ae324db01336e8bad

which cumulatively make this the top gnome-shell crasher in impish.

[Daniel van Vugt (vanvugt)]

Like the similar upstream bug https://gitlab.gnome.org/GNOME/mutter/-/issues/1475, this appears to be caused by the intersection of:

 (a) GTK being used in the gnome-shell process (usually by extensions); and

 (b) The absence of an Xserver for GTK to connect to (Xwayland is only started on-demand now).

It looks like one possible offender using GTK is gnome-shell-extension-appindicator. I'm not sure about the other extensions, and it's hard to tell with desktop-icons-ng which in theory uses GTK in a safe way (in a separate client process).

[Daniel van Vugt (vanvugt)]

If we could then it would be nice to mitigate this by simply ensuring GTK in the gnome-shell process can find the Wayland server automatically instead of X11. Because if we don't find a blanket solution like that then we'll keep getting similar gnome-shell crash reports.

[Daniel van Vugt (vanvugt)]

Interestingly, the source of this particular crash is canberra-gtk-module.c from libcanberra 0.30. That's the latest version but it's also from 2012 so Wayland support might not be great. It appears you can safely just uninstall it:

  sudo apt remove libcanberra-gtk3-module

so I wonder if anyone finds that avoids the crash?

[Daniel van Vugt (vanvugt)]

Seems my development (impish) machine encountered the crash only once, yesterday, and logged:

Nov 02 13:40:39 kab gnome-shell[1614]: Connection to xwayland lost
Nov 02 13:40:39 kab pulseaudio[1472]: X11 I/O error handler called
Nov 02 13:40:39 kab pulseaudio[1472]: X11 I/O error exit handler called, preparing to tear down X11 modules
Nov 02 13:40:39 kab gnome-shell[1614]: Xwayland just died, attempting to recover
Nov 02 13:40:39 kab systemd[1463]: org.gnome.SettingsDaemon.XSettings.service: Failed with result 'exit-code'.
Nov 02 13:40:39 kab gnome-shell[1614]: invalid unclassed pointer in cast to 'GObject'
Nov 02 13:40:39 kab gnome-shell[1614]: g_object_get_data: assertion 'G_IS_OBJECT (object)' failed
Nov 02 13:40:39 kab gnome-shell[1614]: gdk_screen_get_display: assertion 'GDK_IS_SCREEN (screen)' failed
Nov 02 13:40:39 kab gnome-shell[1614]: gdk_display_get_name: assertion 'GDK_IS_DISPLAY (display)' failed
Nov 02 13:40:39 kab gnome-shell[1614]: gdk_screen_get_number: assertion 'GDK_IS_SCREEN (screen)' failed
Nov 02 13:40:39 kab gnome-shell[1614]: gtk_settings_get_for_screen: assertion 'GDK_IS_SCREEN (screen)' failed
Nov 02 13:40:39 kab gnome-shell[1614]: invalid unclassed pointer in cast to 'GObject'
Nov 02 13:40:39 kab gnome-shell[1614]: g_object_set_data_full: assertion 'G_IS_OBJECT (object)' failed
Nov 02 13:40:39 kab gnome-shell[1614]: Can't create a GtkStyleContext without a display connection

But errors.ubuntu.com failed to associate it with a known error ID. Maybe we also need to make sure PulseAudio isn't doing any X11.

[Launchpad Janitor (janitor)]

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

[Daniel van Vugt (vanvugt)]

Seems like killing your Xwayland process is an easy way to reproduce this crash.

[Daniel van Vugt (vanvugt)]

Removed appindicator task because I've confirmed it happens without any extensions loaded (duplicate bug 1966862).

[Daniel van Vugt (vanvugt)]

Confirmed! The workaround/fix is:

  sudo apt remove libcanberra-gtk3-module

Also I can't find an active bug tracker for libcanberra and its last release was in 2012.

[Heather Ellsworth (hellsworth)]

If the solution is to remove libcanberra-gtk3-module and since it hasn't been touched since 2012, do we need to provide it at all?

[Marco Trevisan (Treviño) (3v1n0)]

This is the root issue: https://gitlab.gnome.org/GNOME/glib/-/issues/389

Stacktrace:

(gdb) bt
#0 g_log_writer_default (log_level=<optimized out>, fields=0x7ffcf2f236c0, n_fields=6, user_data=0x0) at ../../../glib/gmessages.c:557
#1 0x00007f355eb3cc93 in g_log_structured_array (n_fields=6, fields=0x7ffcf2f236c0, log_level=G_LOG_LEVEL_ERROR) at ../../../glib/gmessages.c:1973
#2 g_log_structured_array (log_level=G_LOG_LEVEL_ERROR, fields=0x7ffcf2f236c0, n_fields=6) at ../../../glib/gmessages.c:1946
#3 0x00007f355eb3ce93 in g_log_structured_standard (log_domain=log_domain@entry=0x7f355e6be063 "Gtk", log_level=log_level@entry=G_LOG_LEVEL_ERROR,
    file=file@entry=0x7f355e724138 "../../../../gtk/gtkstylecontext.c", line=line@entry=0x7f355e6ddea4 "348",
    func=func@entry=0x7f355e724c70 <__func__.70> "gtk_style_context_init",
    message_format=message_format@entry=0x7f355e724458 "Can't create a GtkStyleContext without a display connection") at ../../../glib/gmessages.c:2030
#4 0x00007f355e5b45cf in gtk_style_context_init (context=0x555ee32475d0) at ../../../../gtk/gtkstylecontext.c:348
#5 0x00007f355ec57efa in g_type_create_instance (type=<optimized out>) at ../../../gobject/gtype.c:1929
#6 0x00007f355ec3ef4d in g_object_new_internal (class=class@entry=0x555ee2ec1c60, params=params@entry=0x0, n_params=n_params@entry=0)
    at ../../../gobject/gobject.c:2011
#7 0x00007f355ec401ad in g_object_new_with_properties (object_type=93866027507792, n_properties=0, names=names@entry=0x0, values=values@entry=0x0)
    at ../../../gobject/gobject.c:2181
#8 0x00007f355ec40cb1 in g_object_new (object_type=<optimized out>, first_property_name=<optimized out>) at ../../../gobject/gobject.c:1821
#9 0x00007f355e393f80 in _gtk_style_new_for_path (screen=0x0, path=path@entry=0x555ee2ccd4b0) at ../../../../gtk/deprecated/gtkstyle.c:854
#10 0x00007f355e3941f3 in gtk_style_new () at ../../../../gtk/deprecated/gtkstyle.c:888
#11 0x00007f355e398b71 in gtk_widget_get_default_style () at ../../../../gtk/deprecated/gtkstyle.c:4061
#12 gtk_widget_get_default_style () at ../../../../gtk/deprecated/gtkstyle.c:4050
#13 0x00007f355e65e8ee in gtk_widget_real_destroy (object=0x555ee3d54320) at ../../../../gtk/gtkwidget.c:12356
#14 0x00007f355dec62a7 in ?? () from /lib/x86_64-linux-gnu/libmutter-10.so.0
#15 0x00007f355ec2ed2f in g_closure_invoke (closure=0x555ee10ccac0, return_value=0x0, n_param_values=1, param_values=0x7ffcf2f24180,
    invocation_hint=0x7ffcf2f24100) at ../../../gobject/gclosure.c:830
#16 0x00007f355ec4aae0 in signal_emit_unlocked_R (node=node@entry=0x555ee0fcd450, detail=detail@entry=0, instance=instance@entry=0x555ee3d54320,
    emission_return=emission_return@entry=0x0, instance_and_params=instance_and_params@entry=0x7ffcf2f24180) at ../../../gobject/gsignal.c:3861
#17 0x00007f355ec4c554 in g_signal_emit_valist (instance=<optimized out>, signal_id=<optimized out>, detail=<optimized out>,
    var_args=var_args@entry=0x7ffcf2f24330) at ../../../gobject/gsignal.c:3496
#18 0x00007f355ec4c7a3 in g_signal_emit (instance=instance@entry=0x555ee3d54320, signal_id=<optimized out>, detail=detail@entry=0)
    at ../../../gobject/gsignal.c:3553
#19 0x00007f...

[Launchpad Janitor (janitor)]

This bug was fixed in the package libcanberra - 0.30-10ubuntu1

---------------
libcanberra (0.30-10ubuntu1) jammy; urgency=medium

  * Merge with debian, remaining changes:
    - Disable login sound by default and hide sound from startup applications
    - Install the gtk plugin in the unity-settings-daemon dir
  * debian/gbp.conf: Update with ubuntu settings
  * debian/control: Update reference to ubuntu branches
  * debian/changelog: Add missing entries lost during merges
  * debian/changelog: Fix Heather's email address
  * debian/patches: Refresh

libcanberra (0.30-10) unstable; urgency=medium

  * debian/patches/gtk_dont_assume_x11.patch: Avoid more X11-only codepaths

libcanberra (0.30-9) unstable; urgency=medium

  * debian/patches: Refresh
  * debian/patches: Handle display closing gracefully (LP: #1949200)

 -- Marco Trevisan (Treviño) <email address hidden> Sat, 09 Apr 2022 01:34:34 +0200

[Alberts Muktupāvels (muktupavels)]

3v1n0, gtk-module-Handle-display-closing-gracefully.patch introduced regression!

Why are you setting d->object to NULL in on_object_disposed? By doing that you are introducing memory leak, no? Creating SoundEventData structure strong reference is added, it must be removed otherwise object will never get finalized!?

I am attaching small test app that can be used to reproduce problem - it simply prints warning messages when GObject / GtkWindow is disposed and/or finalized.

Basically looks like now canberra-gtk-module adds two extra refs that are not removed. Also with attached test app free_sound_event now is never called with d->object != NULL.

[Marco Trevisan (Treviño) (3v1n0)]

> Why are you setting d->object to NULL in on_object_disposed? By doing that you are introducing memory leak, no?

Well the point was that I didn't want this to happen when last reference was removed, but this can't really happen in a sane setup unless something played some dirty game with the object.

So that's fine, I've handled it as part of fix for #1983794

[Timo Aaltonen (tjaalton)]

Hello Hans-Petter, or anyone else affected,

Accepted libcanberra into kinetic-proposed. The package will build now and be available at https://launchpad.net/ubuntu/+source/libcanberra/0.30-10ubuntu1.22.10.1 in a few hours, and then in the -proposed repository.

Please help us by testing this new package. See https://wiki.ubuntu.com/Testing/EnableProposed for documentation on how to enable and use -proposed. Your feedback will aid us getting this update out to other Ubuntu users.

If this package fixes the bug for you, please add a comment to this bug, mentioning the version of the package you tested, what testing has been performed on the package and change the tag from verification-needed-kinetic to verification-done-kinetic. If it does not fix the bug for you, please add a comment stating that, and change the tag to verification-failed-kinetic. In either case, without details of your testing we will not be able to proceed.

Further information regarding the verification process can be found at https://wiki.ubuntu.com/QATeam/PerformingSRUVerification . Thank you in advance for helping!

N.B. The updated package will be released to -updates after the bug(s) fixed by this package have been verified and the package has been in -proposed for a minimum of 7 days.

[Utkarsh Gupta (utkarsh)]

Ubuntu 22.10 (Kinetic Kudu) has reached end of life, so this bug will not be fixed for that specific release.