Hello people. After a few weeks, I was finally able to passthrough my GPU to my standalone Windows VM. Everthing seams to be working fine, but when I try to install the Nvidia drivers with GeForce Experience, it takes about 7 minutes and show an error message:
This Nvidia driver is not compatible with this version of Windows
After consulting some Nvidia guides, updating Windows, and asking Gemini, it replied me with:
If your Windows VM is configured with Secure Boot enabled (common in some Qubes setups), you might need to disable it temporarily for the Nvidia driver installation to proceed. Refer to the Qubes documentation on Secure Boot for specific instructions [invalid URL removed]. Remember to re-enable Secure Boot after successful driver installation for security reasons.
I’ve been searching all over Qubes forum and website, and didn’t find a thing about secure boot on a qube.
Pre-passthrough observations:
To be able to attach my GPU devices on the Windows qube, I’ve forced dom0 to use Intel Integrated Graphics by creating a /etc/X11/xorg.conf.d/20-intel.conf
configuration file:
Section "Device"
Identifier "Intel Graphics"
Driver "intel"
Option "BusID" "PCI:MY:INTELGPU.BUSID
EndSection
Section "Screen"
Identifier "Default Screen"
Monitor "eDP-1"
Device "Intel Graphics"
EndSection
similarly to this guide, to make the Nvidia card available to the qube.
After that, I’ve followed this other guide to identify GPU IOMMU group, and add them on Qube’s qube devices tab configuration.
Note
A PCI bridge was indicated as part of the same IOMMU group as the Nvidia GPU, but since it gave the error Non-endpoint pci devices cannot be assigned to guests
, I didn’t attach it to the qube.
Questions
Anyways, my two main questions are:
1. How can disable and enable secure boot on a qube?
2. Is there an equivalent to lspci
on Windows so I can be sure the GPU can be seen by the VM in the first place?
Thanks