[amdgpu] gnome-shell gets SIGKILL'd when lock screen or under heavy load in Wayland

[Martin Randau]

[ Impact ]

gnome-shell gets unceremoniously SIGKILLed on some Ryzen systems, sometimes when the screen locks, sometimes when launching particular apps.

[ Workaround ]

Add this to /etc/environment:

  MUTTER_DEBUG_KMS_THREAD_TYPE=user

and then reboot.

[ Test Plan ]

Not all Ryzen systems (including one I just purchased) are able to reproduce the bug. We have no choice but to leave final verification to the community. Anyone affected should try locking their screen and verify they are not instantly returned to the login screen.

[ Where problems could occur ]

Anywhere in frame scheduling and particularly for mouse cursor movement since that's what the real-time thread exists to optimize.

[ Original Description ]

I have this issue on Ubuntu 23.10. Lock screen works only with an external monitor connected. Otherwise the session is ended and the user is logged out and brought to the gdm screen.
All works in xorg.

ProblemType: Crash
DistroRelease: Ubuntu 23.10
Package: gnome-shell 45~beta.1-0ubuntu2
ProcVersionSignature: Ubuntu 6.3.0-7.7-generic 6.3.5
Uname: Linux 6.3.0-7-generic x86_64
ApportVersion: 2.27.0-0ubuntu2
Architecture: amd64
CasperMD5CheckResult: unknown
CurrentDesktop: ubuntu:GNOME
Date: Wed Sep 6 22:32:22 2023
DisplayManager: gdm3
ExecutablePath: /usr/bin/gnome-shell
InstallationDate: Installed on 2023-09-03 (3 days ago)
InstallationMedia: Ubuntu 23.10 "Mantic Minotaur" - Daily amd64 (20230901.1)
ProcCmdline: /usr/bin/gnome-shell
RelatedPackageVersions: mutter-common 45~beta.1-1ubuntu2
UpgradeStatus: No upgrade log present (probably fresh install)
UserGroups: adm cdrom dip lpadmin plugdev sudo users
separator:

[Apport retracing service (apport)]

Stacktrace:
 #0 0x00007fa67dc9999b in ?? ()
 No symbol table info available.
 Backtrace stopped: Cannot access memory at address 0x7ffda41cde60
StacktraceSource: #0 0x00007fa67dc9999b in ?? ()
StacktraceTop: ?? ()

[Daniel van Vugt (vanvugt)]

Thank you for taking the time to report this bug and helping to make Ubuntu better.

However, processing it in order to get sufficient information for the developers failed (it does not generate an useful symbolic stack trace). This might be caused by some outdated packages which were installed on your system at the time of the report. Please upgrade your system to the latest package versions. If you still encounter the crash, please file a new report.

Thank you for your understanding, and sorry for the inconvenience!

[Daniel van Vugt (vanvugt)]

Before reporting the next crash it's probably a good idea to try removing these unsupported extensions:

  '<email address hidden>',
  '<email address hidden>'

You can do it quickly with:

  cd ~/.local/share/gnome-shell
  rm -rf extensions

and then reboot.

[Martin Randau (cmmrandau)]

Yes, sorry, my bad, but that doesn't fix it. I attach dmesg -t --level=alert,crit,err,warn and journalctl after a fresh boot where I have just logged in, tried to lock screen with win-L and been brought back to gdm with the session ended.

In journalctl notice this which I have googled without success to this error:

Sep 07 06:33:30 ThinkPad-P14s-Gen-3 /usr/libexec/gdm-wayland-session[1399]: dbus-daemon[1399]: [session uid=123 pid=1399] Activated service 'org.freedesktop.systemd1' failed: Process org.freedesktop.systemd1 exited with status 1

Note that everything works when I have an external monitor connected. Also - it sometimes works to lock screen ONCE on a fresh session. It did just now, so I thought it was removing extensions that did it, but it failed when I tried to lock screen a second time.

[Martin Randau (cmmrandau)]

Here is journalctl

[Daniel van Vugt (vanvugt)]

Those logs appear to be incomplete and also not relevant to this bug. Next time it happens please run:

  journalctl -b0 > journal.txt

or if you had to reboot then run:

  journalctl -b-1 > prevjournal.txt

and attach the resulting text file here.

[Martin Randau (cmmrandau)]

Hello Daniel and thanks for the help.
I attach two logs, both after a fresh reboot. The first (_lock) is when locking with win-L, the second after 1 minute screen blank (_blank. I have set it not to automatically lock at screen blank. Both cause wayland to crash and I return to the gdm with the session ended. There is of course some overlap in the logs.

I'm not an expert as you, but very curious as to what's causing it. There is at least some indication of a wayland crash...

Sep 07 09:09:41 ThinkPad-P14s-Gen-3 systemd[3855]: Started snap.snapd-desktop-integration.snapd-desktop-integration.service - Service for snap application snapd-desktop-integration.snapd-desktop-integration.
Sep 07 09:09:41 ThinkPad-P14s-Gen-3 gnome-shell[3352]: Connection to xwayland lost
Sep 07 09:09:41 ThinkPad-P14s-Gen-3 gnome-shell[3352]: Xwayland terminated, exiting since it was mandatory
Sep 07 09:09:41 ThinkPad-P14s-Gen-3 gnome-shell[3352]: (../src/core/meta-context.c:533):meta_context_terminate: runtime check failed: (g_main_loop_is_running (priv->main_loop))
Sep 07 09:09:41 ThinkPad-P14s-Gen-3 systemd[3855]: Started tracker-miner-fs-3.service - Tracker file system data miner.
Sep 07 09:09:41 ThinkPad-P14s-Gen-3 gnome-shell[3352]: JS ERROR: Gio.IOErrorEnum: Xwayland exited unexpectedly
                                                       @resource:///org/gnome/shell/ui/init.js:21:20
Sep 07 09:09:41 ThinkPad-P14s-Gen-3 gnome-shell[3352]: Execution of main.js threw exception: Module resource:///org/gnome/shell/ui/init.js threw an exception
Sep 07 09:09:41 ThinkPad-P14s-Gen-3 gdm-launch-environment][3244]: pam_unix(gdm-launch-environment:session): session closed for user gdm

[Daniel van Vugt (vanvugt)]

I don't see any SIGSEGV in those logs (or "signal 11" or "segfault"). But it does look like gnome-shell got manually killed (SIGKILL) for some reason. The kill seems to occur soon after bug 2034665.

[Martin Randau (cmmrandau)]

It also displays an error in the extension manager. I have noticed that this error is not always present. The extension seems to work, i.e., I can tile windows by dragging them towards corners.

[Daniel van Vugt (vanvugt)]

Indeed that might be an the error leading to gnome-shell's demise but it's hard to tell because 'JS ERROR' is not uncommon and not in itself fatal.

[Martin Randau (cmmrandau)]

I think it's worthwhile to note that both screen blank and lock screen works when I have an external monitor connected (through a lenovo usb dock), regardless of whether the laptop display is set as primary or not. I don't know if you wants logs from that, or if I can provide any other relevant information?

[Daniel van Vugt (vanvugt)]

Try disabling the tiling assistant extension and see if the bug goes away.

[Martin Randau (cmmrandau)]

Unfortunately not. Here is a log where you can see the crash even though tiling-manager was disabled with "gnome-extensions disable <email address hidden>". The first crash was with and the second without tiling-manger, no reboot in-between though.

[Martin Randau (cmmrandau)]

Hello again!
I've found this bug, which was similar in causing crashes at suspend/blank/lock screen, and also with "Error reading events from display: Broken pipe": https://askubuntu.com/questions/1465560/lunar-lobster-wayland-crashes and the bug: https://bugs.launchpad.net/ubuntu/+source/mutter/+bug/2012230

[Martin Randau (cmmrandau)]

I just realized I can't even login via xorg anymore, not sure what has changed. It crashes and tells me to log out and I'm back on gdm. I think the xorg login is around 19:03:07 in this log (right after boot).

[Martin Randau (cmmrandau)]

Tried a fresh install of 23.10 and the problem persists, only thing I've done is to set amd_pstate=active at kernel boot parameter, which should not matter here. In fact, xorg crashes at the first login attempt, and sends me back to gdm, where wayland is unavailable. After reboot, wayland is again working, but with this same error of suspend/blank/lock described above.

[Daniel van Vugt (vanvugt)]

Tiling Assistant is still enabled in your logs. It might be one of those cases where an extension still causes problems when disabled.

I do see gnome-shell being SIGKILL'd by something external still. Not sure why but it's understandable that feels like a crash.

[Daniel van Vugt (vanvugt)]

Please try commenting out or removing line 274 of /<email address hidden>/src/extension/altTab.js

        AltTab.AppSwitcherPopup = this._originalAltTab;

and then log in again.

[Martin Randau (cmmrandau)]

Hello Daniel
No luck doing that either. This is with the line commented on a fresh install and without manually disabling the extension.

[Daniel van Vugt (vanvugt)]

I don't think this is a gnome-shell bug. gnome-shell is just the part of the problem you can see. What's happening before gnome-shell is killed is a bunch of daemons going down:

Sep 07 14:46:56 ThinkPad-P14s-Gen-3 gnome-calendar[5010]: Error reading events from display: Broken pipe
Sep 07 14:46:56 ThinkPad-P14s-Gen-3 xdg-desktop-por[4930]: Error reading events from display: Broken pipe
Sep 07 14:46:56 ThinkPad-P14s-Gen-3 gjs[5196]: Error reading events from display: Broken pipe
Sep 07 14:46:56 ThinkPad-P14s-Gen-3 gjs[5213]: Error reading events from display: Broken pipe
Sep 07 14:46:56 ThinkPad-P14s-Gen-3 gnome-terminal-[5019]: Error reading events from display: Broken pipe
Sep 07 14:46:56 ThinkPad-P14s-Gen-3 gnome-control-c[5018]: Error reading events from display: Broken pipe
Sep 07 14:46:56 ThinkPad-P14s-Gen-3 snapd-desktop-i[4860]: Error reading events from display: Broken pipe
Sep 07 14:46:56 ThinkPad-P14s-Gen-3 xdg-desktop-por[4754]: Error reading events from display: Broken pipe
Sep 07 14:46:56 ThinkPad-P14s-Gen-3 evolution-alarm[4440]: Error reading events from display: Broken pipe
Sep 07 14:46:56 ThinkPad-P14s-Gen-3 gsd-media-keys[4364]: Error reading events from display: Broken pipe
Sep 07 14:46:56 ThinkPad-P14s-Gen-3 gsd-color[4351]: Error reading events from display: Broken pipe
Sep 07 14:46:56 ThinkPad-P14s-Gen-3 gsd-power[4366]: Error reading events from display: Broken pipe
Sep 07 14:46:56 ThinkPad-P14s-Gen-3 gsd-wacom[4387]: Error reading events from display: Broken pipe
Sep 07 14:46:56 ThinkPad-P14s-Gen-3 gsd-keyboard[4362]: Error reading events from display: Broken pipe
Sep 07 14:46:56 ThinkPad-P14s-Gen-3 systemd[2065]: org.gnome.SettingsDaemon.Color.service: Main process exited, code=exited, status=1/FAILURE
Sep 07 14:46:56 ThinkPad-P14s-Gen-3 polkitd[806]: Unregistered Authentication Agent for unix-session:5 (system bus name :1.132, object path /org/freedesktop/PolicyKit1/AuthenticationAgent, locale en_US.UTF-8) (disconnected from bus)
Sep 07 14:46:56 ThinkPad-P14s-Gen-3 systemd[2065]: <email address hidden>: Main process exited, code=killed, status=9/KILL

It might be the main settings daemon or again Xwayland since I think a lot of GNOME code still has some X11 connection.

[Daniel van Vugt (vanvugt)]

Thank you for taking the time to report this bug and helping to make Ubuntu better. It sounds like some part of the system has crashed. To help us find the cause of the crash please follow these steps:

1. Look in /var/crash for crash files and if found run:
    ubuntu-bug YOURFILE.crash
Then tell us the ID of the newly-created bug.

2. If step 1 failed then look at https://errors.ubuntu.com/user/ID where ID is the content of file /var/lib/whoopsie/whoopsie-id on the machine. Do you find any links to recent problems on that page? If so then please send the links to us.

Please take care to avoid attaching .crash files to bugs as we are unable to process them as file attachments. It would also be a security risk for yourself.

[Martin Randau (cmmrandau)]

Cool, looking forward to finding out what's causing it. Note again that everything works when an external monitor is connected. Let me know if I can supply any logs.

[Martin Randau (cmmrandau)]

I submitted a report on the one crash log in /var/crash: https://bugs.launchpad.net/ubuntu/+source/geoclue-2.0/+bug/2034902

Not sure how geoclue plays into it.

[Daniel van Vugt (vanvugt)]

It's unlikely geoclue would bring down all the settings daemons, or gnome-shell.

Each time the problem happens please repeat the steps in comment #25...

[Martin Randau (cmmrandau)]

Well, it "crashes" EVERY time I do either of suspend/lock screen/blank screen, unless the external monitor is connected, but this does not generate a report in /var/crash. I'll try it out a few times more and see if any of the "crashes" generate a log in /var/crash.

[Martin Randau (cmmrandau)]

They have possibly identified the bug here: https://gitlab.gnome.org/GNOME/mutter/-/issues/3012

[Martin Randau (cmmrandau)]

For some reason tiling-assistant is giving additional errors now, see the attached log at 00:00:59 which is when I tried to lock the screen, e.g.:

JS ERROR: Error: Expected an object of type MtkRectangle for argument 'src2' but got type undefined
                                                       equal@file:///<email address hidden>/src/extension/utility.js:443:27
                                                       _edgeTilingPreview@file:///<email address hidden>/src/extension/moveHandler.js:564:39

edit: But uninstalling gnome-extension-ubuntu-tiling-manager does not solve the problem

[Daniel van Vugt (vanvugt)]

Each time the problem happens please repeat all the steps in comment #25.

[Martin Randau (cmmrandau)]

It very rarely generates a bug/crash report but this morning it did, from xwayland, which was identical to what I previously have reported here: https://bugs.launchpad.net/ubuntu/+source/xwayland/+bug/2034995

However, launchpad said that bug was a duplicate of a much older bug.

I reported it again here: https://bugs.launchpad.net/ubuntu/+source/xwayland/+bug/2035057

Gnome-shell and other packages have been updated since. It doesn't (yet) say it's a duplicate of an older bug.

Edit: it does now. Not sure where to go from here. The bug is very reproducible and specific: wayland, no external monitor -> lock screen or screen blank --> crash and return to gdm. If in xorg or in wayland with an external monitor --> everything works.

[Seth Arnold (seth-arnold)]

Does this bug need to remain private? If it were public it might get more attention or serve as a bug for other people to subscribe to, etc.

Please delete the CoreDump.gz attachment before making this public.

Thanks

[Julian Andres Klode (juliank)]

I'm seeing the same issue, closing the lid or pressing WIndows+L kills gnome-shell, not sure yet how to debug especially while at conference without second laptop or common wifi.

I'll go attach gdb from tty3 and then lock the screen and see what happens.

[Julian Andres Klode (juliank)]

I have attached gdb and strace and there's no SIGTERM send before it or any sort of backtrace it just gets a SIGKILL, but I don't know who is sending it.

[Daniel van Vugt (vanvugt)]

The log says it's systemd reporting the SIGKILL. Does that mean it's also systemd issuing the SIGKILL?

Random idea: Does the bug still occur without Desktop Icons NG enabled?

[Martin Randau (cmmrandau)]

Yes, it still occurs when icons are disabled.

However, it does not occur when am external monitor is connected, at least not here. The monitor is connected through an usb dock.

[Daniel van Vugt (vanvugt)]

I have multiple 23.10 machines where screen locking works correctly, no bug. So we probably need to figure out what the common feature is of machines that do experience the bug. Is it 'amdgpu' specific?

[Martin Randau (cmmrandau)]

I have an AMD Ryzen 7 PRO 6850U with Radeon Graphics (16) @ 4.768GHz. This model, Lenovo P14s (and T14) gen 3, has seen several gpu related issues with earlier kernels requiring boot parameters to avoid freezing and flickering. However, since kernel 6.3 and amd_pstate=active, these problems are gone.

Let me know what I can do to help debugging.

[Daniel van Vugt (vanvugt)]

It's more likely to be a software configuration than hardware.

[Julian Andres Klode (juliank)]

I did killsnoop and killsnoop-bpfcc but I didn't see any SIGKILL delivered to the gnome-shell process, current running theory is the kernel is killing the process directly, and sure possibly due to some GPU related code path.

I'm trying to find out how to frace the kernel function killing the process to get a kernel stack trace and see where it comes from.

[Martin Randau (cmmrandau)]

A bug with similar symptoms has recently been fixed in gnome-shell: https://gitlab.gnome.org/GNOME/mutter/-/issues/3012

But I can't say if it's related to this.

[Daniel van Vugt (vanvugt)]

That's a crash (SIGSEGV) so unlikely related to this bug. Although your original report started as a SIGSEGV it has not been reproduced so this bug is now exclusively about the SIGKILL.

[Martin Randau (cmmrandau)]

It works fine in 22.04.3 with the 6.5 kernel (linux-oem-22.04d).

[Martin Randau (cmmrandau)]

Another observation: Since it works with an external monitor connected, I tried to lock the screen and then disconnect the monitor. Not only did this not kill gnome-shell but I was also able to unlock AND lock the screen again, but only once. At the second lock attempt, gnome-shell was killed and I was returned to gdm as usual. This suggests them problem lies in some config file related to what monitor is active, e.g., monitors.xml, but I have not tried to edit that file.

[Daniel van Vugt (vanvugt)]

Please check each time that you still have this in the log:

  <email address hidden>: Main process exited, code=killed, status=9/KILL

and that it hasn't reverted to a SIGSEGV somewhere.

[Martin Randau (cmmrandau)]

It often does not generate a report. I can trigger it at will. This I did at 06:35:01 and journalctl only says this:

Sep 19 06:35:00 ThinkPad-P14s-Gen-3 update-notifier[4354]: gtk_widget_get_scale_factor: assertion 'GTK_IS_WIDGET (widget)' failed
Sep 19 06:35:01 ThinkPad-P14s-Gen-3 gsd-keyboard[3421]: Error reading events from display: Broken pipe
Sep 19 06:35:01 ThinkPad-P14s-Gen-3 update-notifier[4354]: Error reading events from display: Broken pipe
Sep 19 06:35:01 ThinkPad-P14s-Gen-3 gsd-power[3429]: Error reading events from display: Broken pipe
Sep 19 06:35:01 ThinkPad-P14s-Gen-3 xdg-desktop-por[3916]: Error reading events from display: Broken pipe
Sep 19 06:35:01 ThinkPad-P14s-Gen-3 xdg-desktop-por[4004]: Error reading events from display: Broken pipe
Sep 19 06:35:01 ThinkPad-P14s-Gen-3 gsd-wacom[3462]: Error reading events from display: Broken pipe
Sep 19 06:35:01 ThinkPad-P14s-Gen-3 snapd-desktop-i[3993]: Error reading events from display: Broken pipe
Sep 19 06:35:01 ThinkPad-P14s-Gen-3 evolution-alarm[3479]: Error reading events from display: Broken pipe
Sep 19 06:35:01 ThinkPad-P14s-Gen-3 gsd-color[3410]: Error reading events from display: Broken pipe
Sep 19 06:35:01 ThinkPad-P14s-Gen-3 gsd-media-keys[3422]: Error reading events from display: Broken pipe
Sep 19 06:35:01 ThinkPad-P14s-Gen-3 systemd[2758]: org.gnome.SettingsDaemon.Keyboard.service: Main process exited, code=exited, status=1/FAILURE
Sep 19 06:35:01 ThinkPad-P14s-Gen-3 polkitd[1226]: Unregistered Authentication Agent for unix-session:2 (system bus name :1.71, object path /org/freedesktop/PolicyKit1/AuthenticationAgent, locale en_US.UTF-8) (disconnected from bus)
Sep 19 06:35:01 ThinkPad-P14s-Gen-3 systemd[2758]: <email address hidden>: Main process exited, code=killed, status=9/KILL
Sep 19 06:35:01 ThinkPad-P14s-Gen-3 systemd[2758]: org.gnome.SettingsDaemon.Color.service: Main process exited, code=exited, status=1/FAILURE
Sep 19 06:35:01 ThinkPad-P14s-Gen-3 systemd[2758]: org.gnome.SettingsDaemon.MediaKeys.service: Main process exited, code=exited, status=1/FAILURE
Sep 19 06:35:01 ThinkPad-P14s-Gen-3 systemd[2758]: org.gnome.SettingsDaemon.Power.service: Main process exited, code=exited, status=1/FAILURE
Sep 19 06:35:01 ThinkPad-P14s-Gen-3 systemd[2758]: org.gnome.SettingsDaemon.Wacom.service: Main process exited, code=exited, status=1/FAILURE
Sep 19 06:35:01 ThinkPad-P14s-Gen-3 systemd[2758]: xdg-desktop-portal-gnome.service: Main process exited, code=exited, status=1/FAILURE
Sep 19 06:35:01 ThinkPad-P14s-Gen-3 systemd[2758]: xdg-desktop-portal-gnome.service: Failed with result 'exit-code'.
Sep 19 06:35:01 ThinkPad-P14s-Gen-3 systemd[2758]: xdg-desktop-portal-gtk.service: Main process exited, code=exited, status=1/FAILURE
Sep 19 06:35:01 ThinkPad-P14s-Gen-3 systemd[2758]: xdg-desktop-portal-gtk.service: Failed with result 'exit-code'.
Sep 19 06:35:01 ThinkPad-P14s-Gen-3 systemd[2758]: org.gnome.SettingsDaemon.Keyboard.service: Failed with result 'exit-code'.
Sep 19 06:35:01 ThinkPad-P14s-Gen-3 systemd[2758]: org.gnome.SettingsDaemon.Color.service: Failed with result 'exit-code'.
Sep 19 06:35:01 ThinkPad-P14s-Gen-3 systemd[2758]: org.gnome.SettingsDaemon.Medi...

[Daniel van Vugt (vanvugt)]

It looks like the real problem started earlier than that. Can you attach the full log?

[Daniel van Vugt (vanvugt)]

Please also repeat the steps in comment #25 each time, in case we get lucky.

[Martin Randau (cmmrandau)]

Here is a log containing a fresh boot. I press win-L to lock at Sep 19 09:27:00 where this is the first error:

Sep 19 09:27:00 ThinkPad-P14s-Gen-3 gnome-shell[3270]: JS ERROR: Extension <email address hidden>: TypeError: "AppSwitcherPopup" is read-only
                                                       destroy@file:///<email address hidden>/src/extension/altTab.js:274:9
                                                       disable@file:///<email address hidden>/extension.js:248:30
                                                       _callExtensionDisable@resource:///org/gnome/shell/ui/extensionSystem.js:202:32
                                                       _onEnabledExtensionsChanged@resource:///org/gnome/shell/ui/extensionSystem.js:624:24
                                                       async*_sessionUpdated@resource:///org/gnome/shell/ui/extensionSystem.js:808:20
                                                       ExtensionManager/<@resource:///org/gnome/shell/ui/extensionSystem.js:44:18
                                                       _callHandlers@resource:///org/gnome/gjs/modules/core/_signals.js:130:42
                                                       _emit@resource:///org/gnome/gjs/modules/core/_signals.js:119:10
                                                       _sync@resource:///org/gnome/shell/ui/sessionMode.js:216:14
                                                       pushMode@resource:///org/gnome/shell/ui/sessionMode.js:174:14
                                                       activate@resource:///org/gnome/shell/ui/screenShield.js:666:34
                                                       lock@resource:///org/gnome/shell/ui/screenShield.js:733:14
                                                       LockAsync@resource:///org/gnome/shell/ui/shellDBus.js:519:28
                                                       _handleMethodCall@resource:///org/gnome/gjs/modules/core/overrides/Gio.js:373:35
                                                       _wrapJSObject/<@resource:///org/gnome/gjs/modules/core/overrides/Gio.js:408:34
                                                       @resource:///org/gnome/shell/ui/init.js:21:20

[Martin Randau (cmmrandau)]

If I disable tiling assistant I get this (press win-L at 09:38:40):

Sep 19 09:38:31 ThinkPad-P14s-Gen-3 gnome-shell[8138]: Meta.Rectangle is deprecated, use Mtk.Rectangle instead
Sep 19 09:38:31 ThinkPad-P14s-Gen-3 gnome-shell[8138]: Meta.Rectangle is deprecated, use Mtk.Rectangle instead
Sep 19 09:38:31 ThinkPad-P14s-Gen-3 gnome-shell[8138]: Meta.Rectangle is deprecated, use Mtk.Rectangle instead
Sep 19 09:38:31 ThinkPad-P14s-Gen-3 gnome-shell[8138]: Meta.Rectangle is deprecated, use Mtk.Rectangle instead
Sep 19 09:38:37 ThinkPad-P14s-Gen-3 gnome-character[9259]: JS LOG: Characters Application exiting
Sep 19 09:38:40 ThinkPad-P14s-Gen-3 update-notifier[9048]: gtk_widget_get_scale_factor: assertion 'GTK_IS_WIDGET (widget)' failed
Sep 19 09:38:41 ThinkPad-P14s-Gen-3 update-notifier[9048]: Error reading events from display: Broken pipe
Sep 19 09:38:41 ThinkPad-P14s-Gen-3 gnome-tweaks[9196]: Error reading events from display: Broken pipe
Sep 19 09:38:41 ThinkPad-P14s-Gen-3 nautilus[8701]: Error reading events from display: Broken pipe
Sep 19 09:38:41 ThinkPad-P14s-Gen-3 gnome-control-c[9097]: Error reading events from display: Broken pipe
Sep 19 09:38:41 ThinkPad-P14s-Gen-3 xdg-desktop-por[8840]: Error reading events from display: Broken pipe
Sep 19 09:38:41 ThinkPad-P14s-Gen-3 xdg-desktop-por[8743]: Error reading events from display: Broken pipe
Sep 19 09:38:41 ThinkPad-P14s-Gen-3 gnome-terminal-[8898]: Error reading events from display: Broken pipe
Sep 19 09:38:41 ThinkPad-P14s-Gen-3 evolution-alarm[8452]: Error reading events from display: Broken pipe
Sep 19 09:38:41 ThinkPad-P14s-Gen-3 gsd-media-keys[8405]: Error reading events from display: Broken pipe
Sep 19 09:38:41 ThinkPad-P14s-Gen-3 gsd-wacom[8444]: Error reading events from display: Broken pipe
Sep 19 09:38:41 ThinkPad-P14s-Gen-3 snapd-desktop-i[8320]: Error reading events from display: Broken pipe
Sep 19 09:38:41 ThinkPad-P14s-Gen-3 gsd-power[8410]: Error reading events from display: Broken pipe
Sep 19 09:38:41 ThinkPad-P14s-Gen-3 gsd-color[8390]: Error reading events from display: Broken pipe
Sep 19 09:38:41 ThinkPad-P14s-Gen-3 gnome-text-edit[8980]: Error reading events from display: Broken pipe
Sep 19 09:38:41 ThinkPad-P14s-Gen-3 gsd-keyboard[8401]: Error reading events from display: Broken pipe
Sep 19 09:38:41 ThinkPad-P14s-Gen-3 systemd[2987]: org.gnome.SettingsDaemon.Color.service: Main process exited, code=exited, status=1/FAILURE

[Julian Andres Klode (juliank)]

Updates on my end:

Interestingly, I also got my session killed yesterday with an external screen attached, just not one it had seen before. But I was there over the weekend only, so can't continue testing there.

I also created a new user and that shows the same issues, so this is not a user configuration issue, but either system state or a general bug in the code.

As I've said before, I've done plenty of log reading and nothing is in there, just broken pipes and then the message that gnome-shell got killed. Nothing is visible when you bpf trace signals to gnome-shell PID either. I still need to use trace-cmd to ftrace the exit point of gnome-shell to see the path leading to that.

The only reasoning I can see from not getting results from both killsnoop implementations is that the kernel is killing the process itself. But reverting from 6.5 to 6.3 kernel also did not improve things.

[Daniel van Vugt (vanvugt)]

Comment #51 is bug 2034665.

Comment #52 makes me think it's Xwayland that went away. But NOT related to any log messages mentioning "Xwayland" because those are just from the login screen and harmless. It looks like the lock screen may have caused the login's Xwayland instance to crash silently or just exit, which is why a bunch of binaries start failing and one explicitly mentions ":0" (the connection to Xwayland).

[Daniel van Vugt (vanvugt)]

Found it!(?) - https://gitlab.freedesktop.org/xorg/xserver/-/issues/1579

[Daniel van Vugt (vanvugt)]

Please try adding this to /etc/environment:

  MUTTER_DEBUG_KMS_THREAD_TYPE=user

and then reboot.

[Daniel van Vugt (vanvugt)]

Workaround proposed: https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/3291

[Martin Randau (cmmrandau)]

I can confirm that it works! <:D

edit: I notice a significantly longer boot time where it seems to pause before entering gdm. Before it booted in a few seconds. NVM, it must have been something else - it boots as fast as before now again. :)

[Daniel van Vugt (vanvugt)]

Yeah I think the major kernel upgrade (or something around the same time) caused one boot to be very slow.

[Daniel van Vugt (vanvugt)]

It looks like mutter is *asking* to be killed if any system calls take too long:

  setrlimit (RLIMIT_RTTIME...

and that only seems to happen on some Ryzen systems. Still, I'm not so sure the default limit of 200ms (inherited from rtkit) is sensible if the machine would otherwise recover. It's not like the problem is happening on every frame.

[Launchpad Janitor (janitor)]

This bug was fixed in the package mutter - 45.0-3ubuntu3

---------------
mutter (45.0-3ubuntu3) mantic; urgency=medium

  * Update build autopkgtest to not run dh_auto_test (LP: #2038564)

 -- Jeremy Bícha <email address hidden> Thu, 05 Oct 2023 14:29:32 -0400

[Martin Randau (cmmrandau)]

Just updated and mutter 45.0-3ubuntu3 does not fix this error. Still crash when lock screen and works with MUTTER_DEBUG_KMS_THREAD_TYPE=user added to /etc/environment

[Nicolás Abel Carbone (nicocarbone)]

I can confirm @cmmrandau findings. mutter 45.0-3ubuntu 3 does not fix the bug but Martin's workaround works for me.

[Daniel van Vugt (vanvugt)]

Proposed a different fix:
https://salsa.debian.org/gnome-team/mutter/-/merge_requests/111

[Daniel van Vugt (vanvugt)]

Removing myself because we'll get the upstream fix in 45.1

[Mario Limonciello (superm1)]

This is the upstream fix for this issue: https://gitlab.gnome.org/GNOME/mutter/-/merge_requests/3324

[Mario Limonciello (superm1)]

The fixes are identified, I don't think this should wait until GNOME 45.1 release.
This issue is totally breaking suspend on several of my test systems.

I've backported the fixes from the GNOME 45 branch onto this PPA build:
https://launchpad.net/~superm1/+archive/ubuntu/lp2034619/+packages

I've also uploaded the mutter package into the proposed queue for SRU team to look at.

[Martin Randau (cmmrandau)]

Can confirm that the fixes work. I added Mario's ppa, upgraded and removed the line from /etc/environment. Would be nice if it was fixed in the official Ubuntu repository.

[Timo Aaltonen (tjaalton)]

Mario, could you also upload this to noble? It's open now

[Mario Limonciello (superm1)]

Sure. It wasn't opened yet when I uploaded to mantic a few days ago.

Since it's open now I've:
1. uploaded a *ubuntu4 version to noble (after a local test build with noble archive worked).
2. uploaded a *ubuntu3.1 version to mantic to make sure the version string doesn't conflict since they will be different binaries.

Please reject the *ubuntu4 version from the mantic-proposed queue and accept the *ubuntu3.1 version (if it looks good).

[Timo Aaltonen (tjaalton)]

Hello Martin, or anyone else affected,

Accepted mutter into mantic-proposed. The package will build now and be available at https://launchpad.net/ubuntu/+source/mutter/45.0-3ubuntu3.1 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-mantic to verification-done-mantic. If it does not fix the bug for you, please add a comment stating that, and change the tag to verification-failed-mantic. 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.

[Martin Randau (cmmrandau)]

How do I go about getting this upgrade if I had installed the version from Mario's ppa? I have proposed activated in sources.list and have removed the ppa. But sudo apt update && sudo apt upgrade does not show the new packages. Shall I revert from Mario's version?

[Pirouette Cacahuète (lissyx)]

Enabled proposed repo on 23.10

```
$ $ sudo apt install -t mantic-proposed mutter
Lecture des listes de paquets... Fait
Construction de l'arbre des dépendances... Fait
Lecture des informations d'état... Fait
Le paquet suivant a été installé automatiquement et n'est plus nécessaire :
  libjs-jquery-ui-theme-base
Veuillez utiliser « sudo apt autoremove » pour le supprimer.
Les paquets supplémentaires suivants seront installés :
  gir1.2-mutter-13 libmutter-13-0 mutter-common mutter-common-bin
Les NOUVEAUX paquets suivants seront installés :
  mutter
Les paquets suivants seront mis à jour :
  gir1.2-mutter-13 libmutter-13-0 mutter-common mutter-common-bin
4 mis à jour, 1 nouvellement installés, 0 à enlever et 26 non mis à jour.
Il est nécessaire de prendre 1 727 ko dans les archives.
Après cette opération, 205 ko d'espace disque supplémentaires seront utilisés.
Souhaitez-vous continuer ? [O/n]
Réception de :1 http://fr.archive.ubuntu.com/ubuntu mantic-proposed/main amd64 mutter-common all 45.0-3ubuntu3.1 [45,5 kB]
Réception de :2 http://fr.archive.ubuntu.com/ubuntu mantic-proposed/main amd64 gir1.2-mutter-13 amd64 45.0-3ubuntu3.1 [130 kB]
Réception de :3 http://fr.archive.ubuntu.com/ubuntu mantic-proposed/main amd64 libmutter-13-0 amd64 45.0-3ubuntu3.1 [1 391 kB]
Réception de :4 http://fr.archive.ubuntu.com/ubuntu mantic-proposed/main amd64 mutter-common-bin amd64 45.0-3ubuntu3.1 [49,3 kB]
Réception de :5 http://fr.archive.ubuntu.com/ubuntu mantic-proposed/universe amd64 mutter amd64 45.0-3ubuntu3.1 [111 kB]
1 727 ko réceptionnés en 0s (7 132 ko/s)
(Lecture de la base de données... 313559 fichiers et répertoires déjà installés.)
Préparation du dépaquetage de .../mutter-common_45.0-3ubuntu3.1_all.deb ...
Dépaquetage de mutter-common (45.0-3ubuntu3.1) sur (45.0-3ubuntu3) ...
Préparation du dépaquetage de .../gir1.2-mutter-13_45.0-3ubuntu3.1_amd64.deb ...
Dépaquetage de gir1.2-mutter-13:amd64 (45.0-3ubuntu3.1) sur (45.0-3ubuntu3) ...
Préparation du dépaquetage de .../libmutter-13-0_45.0-3ubuntu3.1_amd64.deb ...
Dépaquetage de libmutter-13-0:amd64 (45.0-3ubuntu3.1) sur (45.0-3ubuntu3) ...
Préparation du dépaquetage de .../mutter-common-bin_45.0-3ubuntu3.1_amd64.deb ...
Dépaquetage de mutter-common-bin (45.0-3ubuntu3.1) sur (45.0-3ubuntu3) ...
Sélection du paquet mutter précédemment désélectionné.
Préparation du dépaquetage de .../mutter_45.0-3ubuntu3.1_amd64.deb ...
Dépaquetage de mutter (45.0-3ubuntu3.1) ...
Paramétrage de mutter-common (45.0-3ubuntu3.1) ...
Paramétrage de mutter-common-bin (45.0-3ubuntu3.1) ...
Traitement des actions différées (« triggers ») pour libglib2.0-0:amd64 (2.78.0-2) ...
Traitement des actions différées (« triggers ») pour libglib2.0-0:i386 (2.78.0-2) ...
Traitement des actions différées (« triggers ») pour libc-bin (2.38-1ubuntu6) ...
Traitement des actions différées (« triggers ») pour man-db (2.11.2-3) ...
Paramétrage de libmutter-13-0:amd64 (45.0-3ubuntu3.1) ....

[Mario Limonciello (superm1)]

#75

You can turn on proposed and then use ppa-purge command to remove my PPA.

[Martin Randau (cmmrandau)]

Maybe this problem shouldn't be posted here but I can't remove the ppa version. "sudo apt install -t mantic-proposed mutter" installed the new mutter from the ubuntu repos, but the other packages are still "local" from the ppa (I think). Anyway, it's working. Would still want the ubuntu versions though. :)

martin@ThinkPad-P14s-Gen-3:~$ sudo ppa-purge ppa:superm1/lp2034619
Updating packages lists
PPA to be removed: superm1 lp2034619
Warning: Could not find package list for PPA: superm1 lp2034619

martin@ThinkPad-P14s-Gen-3:~$ sudo apt list mutter*
Listing... Done
mutter-13-tests/mantic,mantic 45.0-3ubuntu3 amd64
mutter-common-bin/now 45.0-3ubuntu4 amd64 [installed,local]
mutter-common/now 45.0-3ubuntu4 all [installed,local]
mutter/mantic-proposed,now 45.0-3ubuntu3.1 amd64 [installed]

edit:
I had deactivated Mario's ppa before running ppa-purge which is why it didn't find the files list.

martin@ThinkPad-P14s-Gen-3:~$ sudo ppa-purge ppa:superm1/lp2034619
Updating packages lists
PPA to be removed: superm1 lp2034619
Package revert list generated:
 gir1.2-mutter-13:amd64/mantic libmutter-13-0:amd64/mantic mutter/mantic
mutter-common/mantic mutter-common-bin/mantic

Updating packages lists
Reading package lists... Done
Building dependency tree... Done
Reading state information... Done
gir1.2-mutter-13 is already the newest version (45.0-3ubuntu4).
libmutter-13-0 is already the newest version (45.0-3ubuntu4).
mutter-common is already the newest version (45.0-3ubuntu4).
mutter-common-bin is already the newest version (45.0-3ubuntu4).
Selected version '45.0-3ubuntu4' (lp2034619:23.10/mantic [amd64]) for 'gir1.2-mutter-13'
Selected version '45.0-3ubuntu4' (lp2034619:23.10/mantic [amd64]) for 'libmutter-13-0'
Selected version '45.0-3ubuntu4' (lp2034619:23.10/mantic [amd64]) for 'mutter'
Selected version '45.0-3ubuntu4' (lp2034619:23.10/mantic [all]) for 'mutter-common'
Selected version '45.0-3ubuntu4' (lp2034619:23.10/mantic [amd64]) for 'mutter-common-bin'
The following packages will be upgraded:
  mutter
1 upgraded, 0 newly installed, 0 to remove and 0 not upgraded.
Need to get 144 kB of archives.
After this operation, 0 B of additional disk space will be used.
Do you want to continue? [Y/n]
Get:1 https://ppa.launchpadcontent.net/superm1/lp2034619/ubuntu mantic/main amd64 mutter amd64 45.0-3ubuntu4 [144 kB]
Fetched 144 kB in 0s (430 kB/s)
(Reading database ... 251957 files and directories currently installed.)
Preparing to unpack .../mutter_45.0-3ubuntu4_amd64.deb ...
Unpacking mutter (45.0-3ubuntu4) over (45.0-3ubuntu3.1) ...
Setting up mutter (45.0-3ubuntu4) ...
PPA purged successfully

and now

martin@ThinkPad-P14s-Gen-3:~$ sudo apt list mutter*
Listing... Done
mutter-13-tests/mantic,mantic 45.0-3ubuntu3 amd64
mutter-common-bin/now 45.0-3ubuntu4 amd64 [installed,local]
mutter-common/now 45.0-3ubuntu4 all [installed,local]
mutter/now 45.0-3ubuntu4 amd64 [installed,local]

Not sure which version I'm using. It's not mutter (45.0-3ubuntu3.1) as the previous poster.

[RavanH (ravanhagen)]

I can totally confirm this issue affecting my Thinkpad P14s with Ryzen 7 is fixed by the mutter release in Proposed. Eabling the Prodosed repo did not get me any updates but after installing mutter with "sudo apt install -t mantic-proposed mutter" all is well in town :)

Thanks so much for this!

[Ante Karamatić (ivoks)]

Another +1 on -proposed mutter. It solves the problem for me too.

[Antonio madrigal (amadrigal)]

Reporting Thinkpad T14 of mine was affected, proposed mutterfix, it fixed the reported issue.

[Launchpad Janitor (janitor)]

This bug was fixed in the package mutter - 45.0-3ubuntu4

---------------
mutter (45.0-3ubuntu4) noble; urgency=medium

  * Cherry pick 4 commits to fix SIGKILL on lockscreen. (LP: #2034619)
  * Drop kms-impl-device-Inhibit-deadline-timer-on-amdgpu.patch

 -- Mario Limonciello <email address hidden> Thu, 26 Oct 2023 10:27:44 -0500

[Halvor Lyche Strandvoll (halvors)]

Another +1 on -proposed mutter. It solves the problem for me too.

[DooMMasteR (winrootkit-w)]

For me the new mutter does not fix the issue, gnome-shell does still crash/get killed when I lock my screen.

2 other people at my office have NO ISSUES at all, despite both being on the stable release and also having AMDGPU devices (one is the exact same device I have).
Thinkpad P14s G2a

[DooMMasteR (winrootkit-w)]

Ok correction: on a colleagues machine the issue occurs when his laptop is detached from the dock (external monitors) on my machine it occurs even when docked.
the `MUTTER_DEBUG_KMS_THREAD_TYPE=user` fix/workaround works though.

[DooMMasteR (winrootkit-w)]

And another correction
sudo apt install mutter/mantic-proposed
and
sudo apt install -t mantic-proposed mutter
do different things :-)

the mutter fix works :-)

[Jiwoon Lee (metr0jw)]

+1 on -proposed mutter. I'm using Ubuntu 23.10 w/ Linux 6.5.0-10-generic kernel, ThinkPad T14 Gen 3 (AMD), it helped resolving the problem.
Many thanks!

[Johnny Jamaica (johnnyjamaica)]

Same issue here with Thinkpad P14s Gen 2 (AMD) with 23.10 installed just moments ago

I'm also unable to install the fixes on a fresh installation:

sudo apt install -t mantic-proposed mutter
Reading package lists... Done
E: The value 'mantic-proposed' is invalid for APT::Default-Release as such a release is not available in the sources

 sudo apt install mutter/mantic-proposed
Reading package lists... Done
Building dependency tree... Done
Reading state information... Done
Package mutter is not available, but is referred to by another package.
This may mean that the package is missing, has been obsoleted, or
is only available from another source
However the following packages replace it:
  mutter-common-bin

E: Release 'mantic-proposed' for 'mutter' was not found

Did I miss something?

=> mutter-common-bin is already the newest version (45.0-3ubuntu3).

[Launchpad Janitor (janitor)]

This bug was fixed in the package mutter - 45.0-3ubuntu3.1

---------------
mutter (45.0-3ubuntu3.1) mantic; urgency=medium

  * Cherry pick 4 commits to fix SIGKILL on lockscreen. (LP: #2034619)
  * Drop kms-impl-device-Inhibit-deadline-timer-on-amdgpu.patch

 -- Mario Limonciello <email address hidden> Thu, 26 Oct 2023 10:27:44 -0500

[Andreas Hasenack (ahasenack)]

The verification of the Stable Release Update for mutter has completed successfully and the package is now being released to -updates. Subsequently, the Ubuntu Stable Release Updates Team is being unsubscribed and will not receive messages about this bug report. In the event that you encounter a regression using the package from -updates please report a new bug using ubuntu-bug and tag the bug report regression-update so we can easily find any regressions.

[laurenz baba (laurenzbaba)]

I'm on stable and was also affected by this issue.
I just did the update and it seems like it fixed it for me as well.

[alex valin (alexoune)]

had the same issue on my thinkpad t16 (amd) on 23.10. I did some updates and it fixed the issue

[Luca Osvaldo Mastromatteo (lukycrociato)]

I don't think if that is the same issue, but I tried setting the automatic screen sleep to 1 minute and not always but most of the times I can't login back to the system since it does not wake up.

The system is still functional, having nomachine installing and configured trying to login from another computer, I get a successfull connection with no video, since then, the system takes like 4/5 minutes to show a black screen with a cursor first, then the lockscreen and everything works again

Ryzen 5 5600 and an AMD RX 7800 XT

[Mario Limonciello (superm1)]

This should be a different issue than the mutter issue; please open it up separately for investigation.

[Andi McClure (andi-mcc)]

I had a working system despite this bug with the MUTTER_DEBUG_KMS_THREAD_TYPE=user workaround, after upgrading to get this fix I can no longer use the laptop. It prints "[ OK ] Starting gdm.service" during boot and then hangs. No GUI comes up. I can get into vterms with ctrl-alt-fn-fkey but ofc I need a GUI. Persists across reboots. Removing the `MUTTER_DEBUG_KMS_THREAD_TYPE=user` workaround did not fix things. This happened at an inconvenient time and has caused big problems. Will file another bug I guess

[Daniel van Vugt (vanvugt)]

^^^
Bug 1717878