[nvidia] wayland does not start - blank screen with underscore for a minute

[Mateusz Łącki]

After upgrade to kinetic I can't start Gnome Desktop under Wayland.

Until recent update to the gdm3,trying to start gdm3 with "WaylandEnable=true" in /var/gdm3/custom.conf lead to blank screen with non-blinking underscore sign in top left corner of the screen (with x11 gdm3 working fine). Now the gdm3 starts well, but trying to log into "Ubuntu on Wayland" leads to the blank screen with the underscore. The computer remains bootable, no obvious error messages are IMHO present in /var/log/syslog, dmesg . After around a minute the PC comes back to the gdm3. The PC can boot into "Ubuntu" which means X11.

There are lots od "modeset errors" related to the nvidia driver. Those were present in 22.04, and AFAIK are just a nuissance.

In 22.04 Gnome desktop via Wayland worked fine, so this is a regression. I tried to log in to a new user with empty $HOME, but this changes nothing.

I am happy to attach additional logs upon request.

The attached syslog is for a single boot of the computer that:
- was started with WaylandEnable=true
- was booted into Ubuntu on Wayland (time 14:45:13 in the file), experiencing blank screen
- at 14:46:50 the timout is triggered and "gnome desktop through Wayland" shuts itself down

It seems the line:
Oct 9 14:46:45 testosteron systemd[22999]: <email address hidden>: start operation timed out. Terminating.

is very closely tied to my issue, but google does not find anything similar, so I was unable to investigate further.

----- extra info -----------
Description: Ubuntu Kinetic Kudu (development branch)
Release: 22.10

mutter:
  Installed: 43.0-1ubuntu1
  Candidate: 43.0-1ubuntu1
  Version table:
 *** 43.0-1ubuntu1 500
        500 http://archive.ubuntu.com/ubuntu kinetic/universe amd64 Packages
        100 /var/lib/dpkg/status

------ nvidia dirvers -----------
dpkg -l nvidia* | grep ii :

ii nvidia-compute-utils-515 515.65.01-0ubuntu3 amd64 NVIDIA compute utilities
ii nvidia-driver-515 515.65.01-0ubuntu3 amd64 NVIDIA driver metapackage
ii nvidia-kernel-common-515 515.65.01-0ubuntu3 amd64 Shared files used with the kernel module
ii nvidia-kernel-source-515 515.65.01-0ubuntu3 amd64 NVIDIA kernel source package
ii nvidia-prime 0.8.17.1 all Tools to enable NVIDIA's Prime
ii nvidia-settings 510.47.03-0ubuntu1 amd64 Tool for configuring the NVIDIA graphics driver
ii nvidia-utils-515 515.65.01-0ubuntu3 amd64 NVIDIA driver support binaries

ProblemType: Bug
DistroRelease: Ubuntu 22.10
Package: mutter 43.0-1ubuntu1
ProcVersionSignature: Ubuntu 5.19.0-19.19-generic 5.19.7
Uname: Linux 5.19.0-19-generic x86_64
NonfreeKernelModules: nvidia_modeset zfs zunicode nvidia zavl icp zcommon znvpair
ApportVersion: 2.23.1-0ubuntu1
Architecture: amd64
CasperMD5CheckResult: unknown
CurrentDesktop: ubuntu:GNOME
Date: Sun Oct 9 14:48:35 2022
EcryptfsInUse: Yes
SourcePackage: mutter
UpgradeStatus: Upgraded to kinetic on 2022-10-06 (2 days ago)
mtime.conffile..etc.apport.crashdb.conf: 2021-05-03T18:07:38.595923

[Mateusz Łącki (mateusz-lacki)]

Additional information: under novueau the Wayland session starts properly.

[Daniel van Vugt (vanvugt)]

I think you are right this sounds like the issue:

Oct 9 14:45:14 testosteron gnome-shell[23239]: Running GNOME Shell (using mutter 43.0) as a Wayland display server
Oct 9 14:45:14 testosteron gnome-shell[23239]: Device '/dev/dri/card0' prefers shadow buffer
Oct 9 14:45:14 testosteron gnome-shell[23239]: Added device '/dev/dri/card0' (nvidia-drm) using atomic mode setting.
Oct 9 14:45:14 testosteron gnome-shell[23239]: Enabling experimental feature 'scale-monitor-framebuffer'
Oct 9 14:45:14 testosteron gnome-shell[23239]: Enabling experimental feature 'x11-randr-fractional-scaling'
Oct 9 14:45:14 testosteron gnome-shell[23239]: Created gbm renderer for '/dev/dri/card0'
Oct 9 14:45:14 testosteron gnome-shell[23239]: Boot VGA GPU /dev/dri/card0 selected as primary
Oct 9 14:45:15 testosteron gnome-shell[23239]: Disabling DMA buffer screen sharing (implicit modifiers not supported)
Oct 9 14:45:15 testosteron gnome-shell[23239]: Using public X11 display :1, (using :2 for managed services)
Oct 9 14:45:15 testosteron gnome-shell[23239]: Using Wayland display name 'wayland-0'
...
...
...
Oct 9 14:46:45 testosteron systemd[22999]: <email address hidden>: start operation timed out. Terminating.
Oct 9 14:46:50 testosteron systemd[22999]: <email address hidden>: State 'stop-sigterm' timed out. Killing.
Oct 9 14:46:50 testosteron systemd[22999]: <email address hidden>: Killing process 23239 (gnome-shell) with signal SIGKILL.
Oct 9 14:46:50 testosteron systemd[22999]: <email address hidden>: Killing process 43235 (gst-plugin-scan) with signal SIGKILL.
Oct 9 14:46:50 testosteron systemd[22999]: <email address hidden>: Main process exited, code=killed, status=9/KILL
Oct 9 14:46:50 testosteron systemd[22999]: <email address hidden>: Failed with result 'timeout'.
Oct 9 14:46:50 testosteron systemd[22999]: Failed to start GNOME Shell on Wayland.

[Daniel van Vugt (vanvugt)]

Please:

1. Check for crashes by following:

   https://wiki.ubuntu.com/Bugs/Responses#Missing_a_crash_report_or_having_a_.crash_attachment

2. Try adding this to /etc/environment:

   MUTTER_DEBUG_FORCE_KMS_MODE=simple

   and then reboot.

[Daniel van Vugt (vanvugt)]

Also next time the problem happens, please try:

  Ctrl + Alt + F1
  Ctrl + Alt + F2

[Mateusz Łącki (mateusz-lacki)]

There is a recent gnome-shell crash that is mine:
https://errors.ubuntu.com/oops/19cbbbc2-4876-11ed-9948-fa163e993415

Other than that nothing related (at least 3 days old)

> MUTTER_DEBUG_FORCE_KMS_MODE=simple
I have tried that and it did not change anything

> Ctrl + Alt + F1
> Ctrl + Alt + F2

These do not work in the "blank screen with underscore state"

[Daniel van Vugt (vanvugt)]

Try uninstalling gstreamer1.0-vaapi in case it's the same root cause as bug 1988488 (which really should not be specific to virtual machines).

[Mateusz Łącki (mateusz-lacki)]

Hi, this allows me to log in to "Ubuntu on Wayland".

[Daniel van Vugt (vanvugt)]

Great!