Hi all!
I’ve got a laptop with two GPUs one integrated and one discrete. I would like to passthrough the discrete to an HVM.
I’ve got an external monitor that will be dedicated to that HVM.
I have the following problem:
-
when the PCI devices (VGA-compliant and corresponding audio) are hidden via
rd.qubes.hide_pci
, the monitor is not detected when plugging in the HDMI cable, nor when plugged in before booting Qubes OS. -
when the PCI devices are not hidden, the monitor is detected as expected when I plug in the HDMI cable, but attempting to start the HVM crashes Qubes OS
I believe the monitor needs to be detected for the passthrough to suceed because attempting to generate a configuration in the HVM via X :1 -configure
fails saying that there is no monitor corresponding to :1
(and :0
is already configured, but looks like the laptop’s monitor).
Am I misunderstanding something about the basics of GPU passthrough, or is that behaviour somewhat surprising?
Context: