/dev/dri missing

[Miyel]

I've installed Proxmox 8 on a NUC and installed a vm with Home Assistant OS, Docker with Portainer and a Frigate stack. Now I want to passtrough the GPU but I'm stuck on the /dev/dri folder missing:
root@Proxmox:~# ls -l /dev/dri
ls: cannot access '/dev/dri': No such file or directory

I've followed all kinds of tutorials, forum posts and discussed it with ChatGPT. All of them are leading to the same steps and I've tried them all but in the end nothing changes.

The GPU is there and not being passed through a other system:
root@Proxmox:~# lspci -nnk | grep -A3 -i vga
00:02.0 VGA compatible controller [0300]: Intel Corporation Iris Plus Graphics G7 (Ice Lake) [8086:8a51] (rev 07)
DeviceName: Onboard - Video
Subsystem: Intel Corporation Iris Plus Graphics G7 (Ice Lake) [8086:2212]
Kernel modules: i915

During installation the display went black, setting "nomodeset" fixed this. However this was not in the /etc/default/grub file to be found.

Can someone point this NOOB in the right direction?

 

[Impact]

There is no driver loaded. Does it change if you boot wirh a monitor connected?
Please share

cat /proc/cmdline
grep -sREi "blacklist|softdep" /etc/modprobe.d/

# Might be useful too
journalctl -b

 

[leesteken]

If you go for PCI(e) passthrough of the GPU to a VM, of course you wlll lose the /dewv/dri on the Proxmox host because you dedicate it to the VM. If you wanted to pass the dev-nodes to a container, then don't use PCI(e) passthrough. Lots of information about this in the Proxmox manual and Wiki and this forum but it really depends on whether it's a VM or a container.

I've followed all kinds of tutorials, forum posts and discussed it with ChatGPT. All of them are leading to the same steps and I've tried them all but in the end nothing changes.

I can't really do anything about that or provide help as there is no concrete information (but a lot of potential for disastrous changes to your Proxmox). Maybe start over and read the offical Proxmox manual and Wiki, instead of out-of-date information from the internet?

 

[Miyel]

@Impact
cat /proc/cmdline:
BOOT_IMAGE=/boot/vmlinuz-6.8.12-9-pve root=/dev/mapper/pve-root ro nomodeset quiet intel_iommu=on

grep -sREi "blacklist|softdep" /etc/modprobe.d/
/etc/modprobe.d/pve-blacklist.conf:blacklist nvidiafb

journalctl -b
Jun 03 21:17:44 Proxmox kernel: Linux version 6.8.12-9-pve (build@proxmox) (gcc (Debian 12.2.0-14) 12.2.0, GNU ld (GNU Binutils for Debian) 2.40) #1 SMP PREE>
Jun 03 21:17:44 Proxmox kernel: Command line: BOOT_IMAGE=/boot/vmlinuz-6.8.12-9-pve root=/dev/mapper/pve-root ro nomodeset quiet intel_iommu=on
Jun 03 21:17:44 Proxmox kernel: KERNEL supported cpus:
Jun 03 21:17:44 Proxmox kernel: Intel GenuineIntel
Jun 03 21:17:44 Proxmox kernel: AMD AuthenticAMD
Jun 03 21:17:44 Proxmox kernel: Hygon HygonGenuine
Jun 03 21:17:44 Proxmox kernel: Centaur CentaurHauls
Jun 03 21:17:44 Proxmox kernel: zhaoxin Shanghai
Jun 03 21:17:44 Proxmox kernel: x86/split lock detection: #AC: crashing the kernel on kernel split_locks and warning on user-space split_locks
Jun 03 21:17:44 Proxmox kernel: BIOS-provided physical RAM map:
Jun 03 21:17:44 Proxmox kernel: BIOS-e820: [mem 0x0000000000000000-0x000000000009efff] usable
Jun 03 21:17:44 Proxmox kernel: BIOS-e820: [mem 0x000000000009f000-0x00000000000fffff] reserved
Jun 03 21:17:44 Proxmox kernel: BIOS-e820: [mem 0x0000000000100000-0x000000006f098fff] usable
Jun 03 21:17:44 Proxmox kernel: BIOS-e820: [mem 0x000000006f099000-0x000000006f099fff] ACPI NVS
Jun 03 21:17:44 Proxmox kernel: BIOS-e820: [mem 0x000000006f09a000-0x000000006f0c0fff] usable
Jun 03 21:17:44 Proxmox kernel: BIOS-e820: [mem 0x000000006f0c1000-0x000000006f0c1fff] reserved
Jun 03 21:17:44 Proxmox kernel: BIOS-e820: [mem 0x000000006f0c2000-0x000000006f2c3fff] usable
Jun 03 21:17:44 Proxmox kernel: BIOS-e820: [mem 0x000000006f2c4000-0x0000000071790fff] reserved
Jun 03 21:17:44 Proxmox kernel: BIOS-e820: [mem 0x0000000071791000-0x0000000071814fff] ACPI data
Jun 03 21:17:44 Proxmox kernel: BIOS-e820: [mem 0x0000000071815000-0x000000007188efff] ACPI NVS
Jun 03 21:17:44 Proxmox kernel: BIOS-e820: [mem 0x000000007188f000-0x0000000071e6cfff] reserved
Jun 03 21:17:44 Proxmox kernel: BIOS-e820: [mem 0x0000000071e6d000-0x0000000071efefff] type 20
Jun 03 21:17:44 Proxmox kernel: BIOS-e820: [mem 0x0000000071eff000-0x0000000071efffff] usable
Jun 03 21:17:44 Proxmox kernel: BIOS-e820: [mem 0x0000000071f00000-0x0000000079ffffff] reserved
Jun 03 21:17:44 Proxmox kernel: BIOS-e820: [mem 0x000000007ae00000-0x000000007fffffff] reserved
Jun 03 21:17:44 Proxmox kernel: BIOS-e820: [mem 0x00000000c0000000-0x00000000cfffffff] reserved
Jun 03 21:17:44 Proxmox kernel: BIOS-e820: [mem 0x00000000fe000000-0x00000000fe010fff] reserved
Jun 03 21:17:44 Proxmox kernel: BIOS-e820: [mem 0x00000000fec00000-0x00000000fec00fff] reserved
Jun 03 21:17:44 Proxmox kernel: BIOS-e820: [mem 0x00000000fed00000-0x00000000fed00fff] reserved
Jun 03 21:17:44 Proxmox kernel: BIOS-e820: [mem 0x00000000fed20000-0x00000000fed7ffff] reserved
Jun 03 21:17:44 Proxmox kernel: BIOS-e820: [mem 0x00000000fee00000-0x00000000fee00fff] reserved
Jun 03 21:17:44 Proxmox kernel: BIOS-e820: [mem 0x00000000ff000000-0x00000000ffffffff] reserved
Jun 03 21:17:44 Proxmox kernel: BIOS-e820: [mem 0x0000000100000000-0x000000047fffffff] usable
Jun 03 21:17:44 Proxmox kernel: NX (Execute Disable) protection: active
Jun 03 21:17:44 Proxmox kernel: APIC: Static calls initialized

@leesteken
As far as I understand I'm not trying to, nor have I, passed through the GPU to a VM. It's a fresh install and I haven't tweaked much yet. I also rolled back the changes made. The GPU needs to be passed trough to the Docker container (not a vm).
Unfortunately a lot of info is outdated yes, that's why I tried to follow the official information but it's so excessively and dificult to understand (for beginners) that I'm bound to some tutorials.

 

[leesteken]

cat /proc/cmdline:
BOOT_IMAGE=/boot/vmlinuz-6.8.12-9-pve root=/dev/mapper/pve-root ro nomodeset quiet intel_iommu=on

intel_iommu=on is totally unnecessary on kernel version 6.8 or higher: https://pve.proxmox.com/pve-docs/pve-admin-guide.html#_configuration_14

As far as I understand I'm not trying to, nor have I, passed through the GPU to a VM.

Then why is there no Kernel driver in use: in this output:

root@Proxmox:~# lspci -nnk | grep -A3 -i vga
00:02.0 VGA compatible controller [0300]: Intel Corporation Iris Plus Graphics G7 (Ice Lake) [8086:8a51] (rev 07)
DeviceName: Onboard - Video
Subsystem: Intel Corporation Iris Plus Graphics G7 (Ice Lake) [8086:2212]
Kernel modules: i915

Maybe the iGPU is too new for the current driver? Or did you blacklist it? Or did the output or the grep command get cut off?

The GPU needs to be passed trough to the Docker container (not a vm).

Running Docker in a container is highly not recommended: https://pve.proxmox.com/pve-docs/pve-admin-guide.html#chapter_pct
Why are you doing stuff like intel_iommu=on and preventing drivers from loading for the iGPU (which you would do for a PCI(e) passthrough to a VM)when you want to pass dev-nodes to a container?
Maybe search the forum about Docker and GPU passthrough to a container to find a concrete example? I have no personal experience with this, sorry.

 

[mache]

I have passed my vga to a vm with libreelec KODI in order to use the hdmi output to a tv, but if i want to use for example jellyfin with hw transcoding what should i do ?

00:02.0 VGA compatible controller [0300]: Intel Corporation Alder Lake-UP3 GT1 [UHD Graphics] [8086:46b3] (rev 0c)
DeviceName: Onboard - Video
Kernel driver in use: i915
Kernel modules: i915, xe