Gpu pass through

so i’m trying to use proxmox and pass through a gpu to a windows Vm

after i changed the vm to use uefi and q35 and set pass through device i get this message

Starting VM 101 failed: can’t lock file ‘/var/lock/qemu-server/lock-101.conf’ - got timeout

and i cant change any of the settings for the VM

cpu r9 3900x
gpu rx5700

What does the command
ls -l /var/lock/qemu-server/lock-101.conf
give you?

-rw-r–r-- 1 root root 0 Jun 27 14:18 /var/lock/qemu-server/lock-101.conf

That looks about right. Are there other files in that directory, if so, what ownership and permissions to they have and do they work in their respective VMs?

i’m not sure how i would find that out. i did make a second hdd into a directory. and that is were i’m using the vms

Hello :wave:

Your post has been moved to the vfio subcategory for better assistance.

Use the above to get a full list of the contents of the directory.

The fact you’ve used a 2nd HDD to house/store the VMs shouldn’t be an issue, but perhaps the way you created it might be (part of) the problem. What exactly did you do?

here is that output,
-rw-r–r-- 1 root root 0 Jun 27 13:48 lock-100.conf
-rw-r–r-- 1 root root 0 Jun 27 14:18 lock-101.conf
-rw-r–r-- 1 root root 0 Jun 27 13:48 lock–1.conf

if i remember corectly i just went in that node went to disks and used the intialize disk with GPT

OK, I’m assuming the other files also relate to a VM and those are running correctly, right?

Post the results of this command:
cat /etc/fstab

the vms ran fine till i tried to pass the gpu.

/dev/pve/root / ext4 errors=remount-ro 0 1
UUID=4EA4-E09A /boot/efi vfat defaults 0 1
/dev/pve/swap none swap sw 0 0
proc /proc proc defaults 0 0
[email protected]:~# #
[email protected]:~# /dev/pve/root / ext4 errors=remount-ro 0 1
-bash: /dev/pve/root: Permission denied
[email protected]:~# UUID=4EA4-E09A /boot/efi vfat defaults 0 1
-bash: /boot/efi: Is a directory
[email protected]:~# /dev/pve/swap none swap sw 0 0
-bash: /dev/pve/swap: Permission denied
[email protected]:~# proc /proc proc defaults 0 0
-bash: proc: command not found

Any have an idea

i still cant seem to figure this out. any help would be much appreciated

I’ve run out of ideas, sorry. Perhaps others can chime in?

So you have 3 vm’s there.
The passthrough vm does not start, it gives an error.
If you remove the GPU from the VM, does it start then?

I presume the -100 , -101 and -1 relate to the VM machine names?

1 Like

yes for both questions

ok so i finally got it to work but now i have another problem. now i can’t get the vm to restart. i have to do a hard shut down of the server. problem is i can’t finish installing drivers because of it. is this the amd reset bug?

just so everone knows i got it working after adding vfio_iommu_type1 to grub instead of modules


Interesting. And thanks for posting a follow up, it’s always appreciated.

Still cant figure out the reset. But that is how i finally got it to pass through