Updated dom0, now nvidia drivers don't work, how to troubleshoot?

I have limited intelligence

Despite this, somehow I managed to install nvidia drivers into debian-12-xfce and get them to work after disabling the bus that the nvidia card is one.

I updated dom0 and now it doesn’t work at all. It’s so frustrating. I don’t even know how to approach this problem and it was a struggle to figure out before. I didn’t not ultimately “understand” what I did, I just eventually got it to work.

Can you someone please suggest what to check? I likes qubes and computers but I’m not a tech whiz. This is really frustrating, it took a long time to get it to work to begin with.

could this be a result of the kernel update and if i use a different kernel image the problem will go away?

this is probably going to take days to figure out and it’s just so frustrating.


Maybe you need to build the drivers for new kernel.
Did you use the kernel provided by dom0 for your debian-12-xfce-nvidia template?
Or in-VM kernel from debian template itself?
Try to rebuild the drivers in debian-12-xfce-nvidia template for new kernel.

i don’t even remember how i did it. i think i installed the drivers right into the template from either the nvidia website or from an app repository, because they got added in

would i need to rebuild the drivers if i have a new kernel? i don’t even understand that

i am not sure if dom0’s kernal updated or debian-xfce-12 updated. i just know i have a new cool usb icon in the tray and now i can’t use the nvidia chips

You can check the commands that you used in template using history command.
I guess you can re-run the nvidia drivers installer.
Or if the drivers are installed using DKMS you can run DKMS to rebuild them for new kernel.

I don’t think the DKMS method works for dom0 unfortunately, the only working guide I found was building without DKMS. Hence, a manual recompilation is required after each dom0 kernel update… :confused:

I think @dispuser is talking about GPU passthrough to VM and not about GPU in dom0.

1 Like

i think it is GPU-passthough?

I have a nvidia card inside my system and a regular graphics card. I disabled something before using OMUIIU? It was hard to do.

I don’t know if I should be checking that to see if it’s still disabled? The GPU still attaches to the VM with the debian template without error messages.

Do I need to go into the template and remove the nvidia drivers and reinstall them? Could it be that when Qubes updated the template it messed up the drivers?

i don’t know what DKMS means or how to check this

You can configure the template to use in-VM kernel like this:

Then after kernel update in the future if the nvidia drivers used DKMS then the nvidia drivers will be built for new kernel automatically during template update.

You can try to change the kernel and maybe the nvidia drivers will be rebuilt during this and nvidia will work for you in the qube again.
If it won’t work then try to reinstall the driver in template.

i uninstalled nvidia-drivers, reinstalled nvidia-drivers, then installed a file i saw listed that said dkms in it (nvidia-drivers-dkms?)

it runs now

it makes no sense at all

You can read about DKMS and how to rebuild the module for new kernel here (if it’ll be needed for you in the future):

im not sure i have the skills to know what i did to make this work or to understand arch linux. im new to linux mostly