Driver doesn't initialize correctly on kernel 6.5.2 (archlinux)

[francois26]

I upgraded to kernel 6.5.2 (6.5.2-arch1-1) today, using nvidia 535.104.05-5
I have some optimus laptop (intel + nvidia card).
My card is NVIDIA Corporation GA104M [Geforce RTX 3070 Ti Laptop GPU] (rev a1)

the kernel logs with nvidia in it say:

[ 0.000000] Command line: initrd=\intel-ucode.img initrd=\initramfs-linux.img cryptdevice=/dev/nvme0n1p5:cryptroot root=/dev/mapper/cryptroot rootflags=subvol=@ rw quiet nvidia_drm.modeset=1 resume=/dev/nvme0n1p6
[ 0.043179] Kernel command line: initrd=\intel-ucode.img initrd=\initramfs-linux.img cryptdevice=/dev/nvme0n1p5:cryptroot root=/dev/mapper/cryptroot rootflags=subvol=@ rw quiet nvidia_drm.modeset=1 resume=/dev/nvme0n1p6
[ 22.119347] nvidia: module license ‘NVIDIA’ taints kernel.
[ 22.119351] nvidia: module license taints kernel.
[ 22.253526] nvidia-nvlink: Nvlink Core is being initialized, major device number 509
[ 22.254104] nvidia 0000:01:00.0: enabling device (0000 → 0003)
[ 22.254203] nvidia 0000:01:00.0: vgaarb: changed VGA decodes: olddecodes=io+mem,decodes=none:owns=none
[ 22.376281] nvidia_uvm: module uses symbols nvUvmInterfaceDisableAccessCntr from proprietary module nvidia, inheriting taint.
[ 22.453603] nvidia-uvm: Loaded the UVM driver, major device number 507.
[ 22.747672] nvidia-modeset: Loading NVIDIA Kernel Mode Setting Driver for UNIX platforms 535.104.05 Sat Aug 19 00:59:57 UTC 2023
[ 22.750762] [drm] [nvidia-drm] [GPU ID 0x00000100] Loading driver
[ 23.182199] [drm:nv_drm_load [nvidia_drm]] ERROR [nvidia-drm] [GPU ID 0x00000100] Failed to allocate NvKmsKapiDevice
[ 23.182260] [drm:nv_drm_probe_devices [nvidia_drm]] ERROR [nvidia-drm] [GPU ID 0x00000100] Failed to register device

and logs about nvrm are:

[ 22.303229] NVRM: loading NVIDIA UNIX x86_64 Kernel Module 535.104.05 Sat Aug 19 01:15:15 UTC 2023
[ 23.182079] NVRM: GPU 0000:01:00.0: RmInitAdapter failed! (0x24:0x72:1436)
[ 23.182137] NVRM: GPU 0000:01:00.0: rm_init_adapter failed, device minor number 0

If I use the nvidia-open driver, I get more error messages, I suspect though it would be linked to this.

nvidia-bug-report.log.gz (320.3 KB)

is it something known? Has some configuration changed at some place?

[francois26]

I got it fixed by installing the dkms versions of the nvidia(-open) package so I guess it’s more some packaging issue. I hope not too much time was spent on this

[hys]

I was having the same issue but I fixed the problem by downgrading my “nvidia” package to 535.104.05-2.

It seems that the package version of “nvidia” is 535.104.05-5 and “nvidia-dkms” is 535.104.05-1. I am wondering if there is a bug in the latest nvidia driver.

[francois26]

My understanding on that is that the driver version is 535.104.05 but since the driver must be compatible with the kernel, they recompile the same version with the kernel everytime there is a new kernel, that’s why they add this -1 … -5 at the end of the name.
For the dkms, since it is somewhat rebuilt on the user computer with the user kernel, no need for new versions with each kernel, that’s why it is still at version 1.

I think there was some mismatch with the latest update, which cannot really exist using the dkms version