I’m trying really hard to find a way for Qubes to be the solution to my next PC build (hardware assembly has been delayed a bit, so I have a little time, but not a lot); I’d much rather go the Qubes route than the Linux/Xen route. But once I commit the time to figure out how to do one solution, I’ll not likely start all over again to change direction. So I’d like to do it right the first time.
I absolutely do not want to dual boot to get a windows VM with GPU passthrough running. I could live with just one, could live with adding in a usb card for that VM’s sole use, could live with a hardware KVM to direct my keyboard and mouse to that VM, and can push the “inputs” button on the monitor when I want to see the GPU output.
Xen can do passthrough as is, though the learning curve for its configuration is daunting.
I understand the resistance Joanna et al have to the concept, but I’m determined to get one such VM running with passthrough.
It seems to me that if Xen is sufficiently visible under Qubes that I can issue XI commands, then I can do what I need without help from the Qubes devs.
So my question is: Is this so? How visible is Xen to manipulation under qubes? Does it use the config files documented in Xen forum posts etc, or XI commands, or something similar sufficient that I can set hide values for BDF and such?