I’ve been doing some experimentation with creating a GPU VM. In the process of doing so, I installed the package xorg-x11-drv-nvidia in a standalone VM based on fedora-33. The package was installed from rpmfusion-nonfree.
After doing that, I rebooted the VM and now I am unable to run anything. I don’t know if Xorg is actually running, but I cannot get any application to run anymore.
My expectation was that nothing would change, since the qubes Xorg instance should be using the configuration from /etc/X11/xorg-qubes.conf which always specifies the qubes drivers. In other words, I expected the nvidia drivers to exist, but not affect the existing configuration at all.
How can I debug this? What I want to find out is what changed in the boot process to prevent Xorg from running with the same configuration as before?
This is a standalone VM, and I can easily get back to the original state by using qvm-volume rollback.
My issue is that I want to know what causes the the VM to not start properly so that I can fix it. In other words, I want to be in a state where the Nvidia driver package is installed in the VM, but I can still start it.
Since I posted the original question, I did some more tests and used qvm-run --pass-io from dom0 to determine that the Xorg process is started properly with the correct config, which means that my original guess as to what went wrong was incorrect.
Clearly something is not running correctly,but I cannot figure out what it is that fails.