Hello Qubes forum!
I’ve been working on getting a GPU passthrough up and running on and off for several months now. For the most part, I’ve been able to muddle my way through things using the guide written by neowutran. My PC currently contains two GPUs, one of which is passed through to a VM (pciback driver in dom0)
I do not think my current issue specifically concerns GPU passthrough, though it is impeding me from making further progress on that front: My issue is that when the secondary GPU has a monitor attached during boot, X11 fails to start. I can get to a virtual terminal and log into qubes, but trying to manually ‘startx’ gives me an error of “unable to find valid frame buffer device”. But this only happens when a monitor is plugged in to the ‘passed through’ gpu. Even with the second GPU present, if I don’t have anything plugged into it, Qubes/X11 will start.
1 or 2 monitors plugged into GPU1, no monitors in GPU2 == Qubes and X boot like normal
1 or 2 monitors plugged into GPU1, 1 monitor in GPU2 == X fails to start, but Qubes starts in CLI
No monitor in GPU1, 1 or 2 monitors in GPU2 == no graphics at all (this is expected behaviour)
Errors in Xorg log:
WW - Falling back to old probe method for modesetting
EE - Unable to find valid frame buffer device
WW - Falling back to old probe method for fbdev
EE - Screen 0/1 deleted because of no matching config section
EE - Cannot run in framebuffer mode. Please specify busIDs for all framebuffer devices
I thought that Qubes would completely disregard the second GPU since it’s being passed through. Does the above seem like normal behavior? Would blacklisting the GPU2 driver perhaps help? Any ideas how to work around this issue? I’m not very knowledgeable about Xorg, unfortunately, so I’d appreciate any suggestions.