Recently a system running Ubuntu 24.04.1 LTS, Gnome Shell and X11 has developed a problem resuming after suspend: on waking, no log in prompt or GUI is displayed, simply a black screen with what appears to be a non-blinking text cursor in the top left.
Sometimes this is accompanied by one or more log messages with the text Bluetooth: hci0: No dsm support to set reset delay, but I don't believe this is related to the problem as these messages are usually not present. Usually it is simply a black screen with a wide horizontal white line like an underscore or cursor, which appears on both monitors.
I have not found any action that get's the system back to a working state without losing the session and any open apps after this problem occurs.
Of the many similar questions on this site, only one appears similar in that it mentions a non-blinking text cursor. However, the solution that worked for that user was to remove nomodeset from kernel options, and that option is already not present in my case.
What I tried:
Ctrl-Alt-F1, F2, etc. have no effect so I cannot access another terminal.
SSH is running and working, so I can access and investigate the system from another machine.
Gnome Shell is running, but restarting it with
sudo killall -3 gnome-shellhas no visible effect.Killing either of the following processes, causes a login screen to appear, but the session, including all apps running when the machine was suspended, is lost:
/usr/sbin/gdm3/usr/libexec/gnome-session-binary --systemd-service --session=ubuntu
GDM is running, but attempting to restart the service with
systemctlnever completes.Attempting to restart any Nvidia services also never completes.
I can't see anything out of the ordinary in
syslog.
Other observations:
- The problem does not always occur; roughly once every three attempts it wakes without issue.
- The issue began several days after upgrading the kernel from 6.8.0.41 to 6.8.0.45 (and persists through all subsequent upgrades including the current 6.8.0-55).
- The graphics card is NVIDIA GeForce RTX 4070. The
535.183.01driver was in use for more than two months before the issue began, and the system is still using the same version.
Any advice about what might be causing this failure and how to solve it, or suggestions about how to further debug the problem, would be greatly appreciated.
