After an evaluation, GNOME has moved from Bugzilla to GitLab. Learn more about GitLab.
No new issues can be reported in GNOME Bugzilla anymore.
To report an issue in a GNOME project, go to GNOME GitLab.
Do not go to GNOME Gitlab for: Bluefish, Doxygen, GnuCash, GStreamer, java-gnome, LDTP, NetworkManager, Tomboy.
Bug 775337 - Shouldn't try to run the fail whale if gnome-shell crashed on Wayland
Shouldn't try to run the fail whale if gnome-shell crashed on Wayland
Status: RESOLVED OBSOLETE
Product: gnome-session
Classification: Core
Component: general
3.22.x
Other Linux
: Normal critical
: ---
Assigned To: Session Maintainers
Session Maintainers
Wayland
: 772368 (view as bug list)
Depends on:
Blocks:
 
 
Reported: 2016-11-29 14:59 UTC by Dominique Leuenberger
Modified: 2021-06-14 18:21 UTC
See Also:
GNOME target: ---
GNOME version: ---



Description Dominique Leuenberger 2016-11-29 14:59:14 UTC
While running GNOME inside a qemu/kvm VM (installed using virt-manager), I could observe GNOME not coming up and leaving a black screen.

The underlying issue could be found (well, sort of.. seems libinput is causing some trouble there)

But, at the same time, I observed gnome-session-failed crashing in gtk, and thus not being able to show the 'fail whale' picture (not, that this would have made things work, but it would have been more obvious than a black screen)

a backtrace of gnome-session-failed looked like this:

(gdb) bt
  • #0 _gtk_style_provider_private_get_settings
    at gtkstyleproviderprivate.c line 123
  • #1 gtk_css_value_initial_compute
    at gtkcssinitialvalue.c line 52
  • #2 gtk_css_static_style_compute_value
    at gtkcssstaticstyle.c line 237
  • #3 _gtk_css_lookup_resolve
    at gtkcsslookup.c line 122
  • #4 gtk_css_static_style_new_compute
    at gtkcssstaticstyle.c line 195
  • #5 gtk_css_static_style_get_default
    at gtkcssstaticstyle.c line 164
  • #6 gtk_css_node_init
    at gtkcssnode.c line 663
  • #7 g_type_create_instance
    at gtype.c line 1860
  • #8 g_object_new_internal
    at gobject.c line 1783
  • #9 g_object_newv
    at gobject.c line 1930
  • #10 g_object_new
    at gobject.c line 1623
  • #11 gtk_css_widget_node_new
    at gtkcsswidgetnode.c line 301
  • #12 gtk_widget_init
    at gtkwidget.c line 4409
  • #13 g_type_create_instance
    at gtype.c line 1860
  • #14 g_object_new_internal
    at gobject.c line 1783
  • #15 g_object_newv
    at gobject.c line 1930
  • #16 g_object_new
    at gobject.c line 1623
  • #17 main
    at gsm-fail-whale-dialog.c line 382

Comment 1 Dominique Leuenberger 2016-11-29 15:28:54 UTC
(adjusting version - this was actually tested / seen with gtk 3.22.4)
Comment 2 Bastien Nocera 2016-12-01 13:18:04 UTC
*** Bug 775463 has been marked as a duplicate of this bug. ***
Comment 3 Bastien Nocera 2016-12-01 13:19:43 UTC
(In reply to Dominique Leuenberger from comment #0)
> While running GNOME inside a qemu/kvm VM (installed using virt-manager), I
> could observe GNOME not coming up and leaving a black screen.
> 
> The underlying issue could be found (well, sort of.. seems libinput is
> causing some trouble there)
> 
> But, at the same time, I observed gnome-session-failed crashing in gtk, and
> thus not being able to show the 'fail whale' picture (not, that this would
> have made things work, but it would have been more obvious than a black
> screen)

I'm guessing that, in your logs, you'll see gnome-shell crashing before the gnome-session helper here, and that you're using Wayland, correct?
Comment 4 Dominique Leuenberger 2016-12-01 13:23:53 UTC
(In reply to Bastien Nocera from comment #3)
> I'm guessing that, in your logs, you'll see gnome-shell crashing before the
> gnome-session helper here, and that you're using Wayland, correct?

gnome-shell did report to have 'failed' before - but there were no usabe crash logs.

The best I could find was some X issues with XInputExtension like:

Nov 21 08:58:21 germ202 org.gnome.Shell.desktop[1828]: X Error of failed request:  BadValue (integer parameter out of range for operation)
Nov 21 08:58:21 germ202 org.gnome.Shell.desktop[1828]: Major opcode of failed request:  131 (XInputExtension)
Nov 21 08:58:21 germ202 org.gnome.Shell.desktop[1828]: Minor opcode of failed request:  57 ()
Nov 21 08:58:21 germ202 org.gnome.Shell.desktop[1828]: Value in failed request:  0x108
Nov 21 08:58:21 germ202 org.gnome.Shell.desktop[1828]: Serial number of failed request:  274
Nov 21 08:58:21 germ202 org.gnome.Shell.desktop[1828]: Current serial number in output stream:  275

This then lead me to remove xf86-libinput to get the system 'running'
Comment 5 Bastien Nocera 2016-12-01 13:36:44 UTC
Likely 5c46094d67f2e5cdafe6edcfd8c0d5894a2dafc6 in mutter (for bug 771744).

If you could verify, I can repurpose the bug for the gnome-session-failed crash.
Comment 6 Dominique Leuenberger 2016-12-01 14:06:25 UTC
(In reply to Bastien Nocera from comment #5)
> Likely 5c46094d67f2e5cdafe6edcfd8c0d5894a2dafc6 in mutter (for bug 771744).
> 
> If you could verify, I can repurpose the bug for the gnome-session-failed
> crash.

Happy to confirm that this fix indeed fixes the underlying issue of mutter / gnome-shell disappearing and thus eliminating the need for gnome-session-failed trying to show (libinput driver installed again)
Comment 7 Matthias Clasen 2016-12-01 15:10:36 UTC
(In reply to Bastien Nocera from comment #5)
> Likely 5c46094d67f2e5cdafe6edcfd8c0d5894a2dafc6 in mutter (for bug 771744).
> 
> If you could verify, I can repurpose the bug for the gnome-session-failed
> crash.

Moving to gnome-session, then
Comment 8 Olivier Fourdan 2017-01-02 14:53:18 UTC
On Wayland, gnome-shell/mutter is the Wayland compositor so if/when it crashes it's unlikely anything at all will be able to show up on the screen (unlike X11 where the X server is still running even after gnome-shell/mutter has vanished)
Comment 9 Bastien Nocera 2017-02-10 18:18:14 UTC
*** Bug 772368 has been marked as a duplicate of this bug. ***
Comment 10 Dominique Leuenberger 2017-04-27 07:27:32 UTC
*** Bug 776096 has been marked as a duplicate of this bug. ***
Comment 11 André Klapper 2021-06-14 18:21:24 UTC
GNOME is going to shut down bugzilla.gnome.org in favor of gitlab.gnome.org.
As part of that, we are mass-closing older open tickets in bugzilla.gnome.org
which have not seen updates for a longer time (resources are unfortunately
quite limited so not every ticket can get handled).

If you can still reproduce the situation described in this ticket in a recent
and supported software version of gnome-session, then please follow
  https://wiki.gnome.org/GettingInTouch/BugReportingGuidelines
and create a new ticket at
  https://gitlab.gnome.org/GNOME/gnome-session/-/issues/

Thank you for your understanding and your help.