I have an optimus laptop, and after the update to KDE6 optimus-manager stopped working. I needed a second display, and all my display outputs are on the Nvdia GPU, so I needed to switch. I tried many different X11 configs, envycontrol then more X11 configs, but I couldn’t get it working right, it would only be the internal display or the external one, not both. after a few hours I gave up and tried optimus-manager again. This time I checked the error log and it was failing to load the nvidia module, I tried loading it manually but I got a “No such device” error, which is where the title of the post comes in. My GPU has disappeared from linux, it won’t show up in lspci, lshw, nvidia-smi, or anything else it should. The only reference to the thing in dmesg I can find are :

[    0.216410] pci 0000:01:00.0: [10de:1ba1] type 00 class 0x030000
[    0.216419] pci 0000:01:00.0: reg 0x10: [mem 0xde000000-0xdeffffff]
[    0.216427] pci 0000:01:00.0: reg 0x14: [mem 0xc0000000-0xcfffffff 64bit pref]
[    0.216435] pci 0000:01:00.0: reg 0x1c: [mem 0xd0000000-0xd1ffffff 64bit pref]
[    0.216440] pci 0000:01:00.0: reg 0x24: [io  0xe000-0xe07f]
[    0.216445] pci 0000:01:00.0: reg 0x30: [mem 0xdf000000-0xdf07ffff pref]
[    0.216460] pci 0000:01:00.0: Enabling HDA controller
[    0.257300] pci 0000:01:00.0: vgaarb: bridge control possible
[    0.257300] pci 0000:01:00.0: vgaarb: VGA device added: decodes=io+mem,owns=none,locks=none
[    0.270521] pci 0000:01:00.1: D0 power state depends on 0000:01:00.0

and then nothing, it doesn’t even seem to try to load the nvidia module. I tried booting into windows and it shows up there fine, so the GPU didn’t randomly die.
As far as I can tell I’ve rolled back everything I did in my histfile until it stopped working, The only thing I could think is I upgraded my kernel to (6.7.9) from (6.6.10), could that have caused it? I also tried adding pcie_port_pm=off to the kernel params from the archwiki, but still nothing. I’m just at a loss here, anyone have any ideas?

EDIT: I’m using the nvidia-dkms package
EDIT2: one kernel downgrade later and it’s still not appearing, so thats not it.
EDIT3: fixed, see comments

  • Nimmo@lem.nimmog.ukEnglish
    7·
    8 months ago

    I have the same issue on my desktop. I’d assumed it was something I’d done (it usually is) but I had to admit defeat and resort to switching to booting into a backup OS so that I could get on with all the tasks I need to get done but I’m assuming it was a problem with the Nvidia-dkms package that’ll be resolved in time as people have reported similar issues in the past.

    • taaz@biglemmowski.winEnglish
      5·
      8 months ago

      I think I had this occur to me once and it was something really dumb but I can’t remember what.

      @thomasdouwes@sopuli.xyz just for the sake of trying everything, you could rebuild the dkms and initrams, then reboot:

      dkms autoinstall -F -a kernel-6.8.5-arch1 # change the kernel version according what you have now (read from uname -a)
mkinitcpio -P

      E: Exhaustive of what I would try

      • check if drivers and modprobe blacklist make sense (this one is broad and requires digging into arch wiki but the optimus laptop I had required blacklisting some drivers from early loading afaik)
      • fiddle with re-scans and power states in the sys bus PCI folders for the GPU
      • check that my mkinitcpio makes sense, additionally look for .pacnew (/etc/mkinitcpio.conf.pacnew) and see if the changes might affect the system
      • downgrade kernel - already tried
      • downgrade dkms packages
      • update BIOS and firmwares from windows
      • cold boot the laptop (shutdown, remove AC and battery, leave it cold for few seconds)
      • on windows, look into ROG Armoury/MSI Center for any kind of toggles that could have impact on the GPUs (iGPU/dGPU) stuff like power states, optimizations etc)
      • Thomas DouwesOP
        5·
        8 months ago

        Looks like you where right about the udev rules earlier, I ran a pacman command to find all untracked files in /usr and I found /usr/lib/udev/rules.d/50-remove-nvidia.rules was there. Contents:

        # Automatically generated by EnvyControl

# Remove NVIDIA USB xHCI Host Controller devices, if present
ACTION=="add", SUBSYSTEM=="pci", ATTR{vendor}=="0x10de", ATTR{class}=="0x0c0330", ATTR{power/control}="auto", ATTR{remove}="1"

# Remove NVIDIA USB Type-C UCSI devices, if present
ACTION=="add", SUBSYSTEM=="pci", ATTR{vendor}=="0x10de", ATTR{class}=="0x0c8000", ATTR{power/control}="auto", ATTR{remove}="1"

# Remove NVIDIA Audio devices, if present
ACTION=="add", SUBSYSTEM=="pci", ATTR{vendor}=="0x10de", ATTR{class}=="0x040300", ATTR{power/control}="auto", ATTR{remove}="1"

# Remove NVIDIA VGA/3D controller devices
ACTION=="add", SUBSYSTEM=="pci", ATTR{vendor}=="0x10de", ATTR{class}=="0x03[0-9]*", ATTR{power/control}="auto", ATTR{remove}="1"

        looks like EnvyControl left some extra files after uninstalling.
        Personally, I think it’s pretty weird that it put runtime files in /usr/lib, if they where in /etc I would have found them quickly.
        The GPU is back on the bus now and I can run optimus-manager to get my extra screen. Thank you for the help troubleshooting this issue.

      • Thomas DouwesOP
        1·
        8 months ago

        I don’t seem to have an -F on my dkms? when I ran that it without, it didn’t rebuild all the DKMS modules for some reason, just bbswitch and evdi

        • taaz@biglemmowski.winEnglish
          2·
          8 months ago

          ah the -F might be wrong then actually, I was playing with custom kernels recently and my dkms is a mess, wouldn’t worry about that option

        • Thomas DouwesOP
          1·
          8 months ago

          dkms status doesn’t even list half of my DKMS modules for some reason

    • Thomas DouwesOP
      1·
      8 months ago

      hmm, maybe downgrading nvidia-dkms might work? I’ll try that

      • Nimmo@lem.nimmog.ukEnglish
        2·
        8 months ago

        I considered trying that then read about some kernel versions not being compatible with certain versions of the dkms package so decided to give up, go to bed and deal with the issue later. (This was Saturday night and sadly time hasn’t permitted me to start investigating again yet)