[Xen-users] gfx-passthrough in qubesOS

[Doesnt Work <doesnt.work@xxxxxx>]

Hello XEN-users!

I'm running a QubesOS 4.0, a security focused xen-derived operating system.

I'm trying to start a windows-VM with an Nvidia-gfx attached using xl
command line tools. I believe I've almost got it, but the nvidia-driver
doesn't correctly detect it.

What I've done so far: enabled pciback, successfully launched vm with
gfx attached as pci-device and played around with all settings in my
win7.cfg I could think of.
However, whenever I start the vm, windows detects the gfx as "3D graphic
acceleration", not nvidia-device. (which is, I believe, why the driver
doesn't detect it.)

I found a guide how to pass through gfx with kvm which, the internet
claims, is applicable to my model:
https://gist.github.com/Misairu-G/616f7b2756c488148b7309addc940b28
An important step there is to specify the sub-vendor and sub-device id.
There is even a link how the libvirt-setup should look like:
https://gist.github.com/anonymous/500f1edf89d6f22c40bd2cbbdec6490b (I
believe the relevant part is at the very end, the qemu:commandline block)

Is it possible to specify these ids using xl-commandline tools as well?
Or am I completely wrong about the problem-source?

sincerely
  GammaSQ

_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxxxxxxxxx
https://lists.xenproject.org/mailman/listinfo/xen-users