Hello all!
I have been trying to pass through AMD integrated GPU with virtualGL by following these guides, but I’ve been running into trouble.
Seamless GPU passthrough on Qubes OS with VirtualGL
Another guide I followed was “Create a gaming HVM”, unable to link it because new users are limited to 2 links per post.
I successfully passed the GPU onto the standalone HVM and I am able to see it by running lspci, but glxinfo doesn’t recognize it. I can use vglrun to run programs but those programs still use software rendering, it doesn’t recognize the GPU either. Dmesg log showed me this.
amdgpu: Unable to locate a BIOS ROM
amdgpu: Fatal error during GPU init
I googled those errors and found a thread from 4 years ago with something similar.
AMD iGPU passthrough attempt
One user told that he was somehow able to solve the issue, and I got my hopes up. Quote from him below.
Update: Using the 3.5G TOLUD patch for the stub domain seems to have fixed this issue for me for the moment.
I looked up the TOLUD patch from 2018 and found that it has been integrated into Qubes since 4.2, the version that I’m using, yet the iGPU issue is still here, presumably because the iGPU passthrough fix was just a side-effect. The directories of the original script from 2018 don’t exist anymore in Qubes, making me unable to run the script.
What should I do? Is there an easy solution to this? Can you point me in the right direction? Thanks in advance!