Nvidia error when upgrading to kernel-6.12.4-100.fc40.x86_64

Ask Fedora
, ,
1

I ran dnf today, and got some Nvidia errors. Has anyone else run into this?

There seem to be no reports on nvidia’s site.

I have yet to reboot, so everything is still running OK.

FYI: I’m still on fc40, since the upgrade from fc39 to fc40 was a nightmare I haven’t yet recovered from – I had to wipe my disk, install fc40, and reload my data from backups.

Here are today’s errors from /var/lib/dkms/nvidia/555.58.02/build/make.log:

DKMS make.log for nvidia/555.58.02 for kernel 6.12.4-100.fc40.x86_64 (x86_64)
Mon Dec 16 09:20:14 AM EST 2024
make[1]: Entering directory '/usr/src/kernels/6.12.4-100.fc40.x86_64'

/var/lib/dkms/nvidia/555.58.02/build/nvidia-drm/nvidia-drm-drv.c:207:6: error: ‘const struct drm_mode_config_funcs’ has no member named ‘output_poll_changed’
  207 |     .output_poll_changed = nv_drm_output_poll_changed,
      |      ^~~~~~~~~~~~~~~~~~~
/var/lib/dkms/nvidia/555.58.02/build/nvidia-drm/nvidia-drm-drv.c:207:28: error: initialization of ‘struct drm_atomic_state * (*)(struct drm_device *)’ from incompatible pointer type ‘void (*)(struct drm_device *)’ [-Wincompatible-pointer-types]
  207 |     .output_poll_changed = nv_drm_output_poll_changed,
      |                            ^~~~~~~~~~~~~~~~~~~~~~~~~~
/var/lib/dkms/nvidia/555.58.02/build/nvidia-drm/nvidia-drm-drv.c:207:28: note: (near initialization for ‘nv_mode_config_funcs.atomic_state_alloc’)

make[2]: *** [/usr/src/kernels/6.12.4-100.fc40.x86_64/Makefile:1976: /var/lib/dkms/nvidia/555.58.02/build] Error 2
make[1]: *** [Makefile:236: __sub-make] Error 2
make[1]: Leaving directory '/usr/src/kernels/6.12.4-100.fc40.x86_64'
make: *** [Makefile:89: modules] Error 2

Thanks!
Fred

3

It seems I have similar issue on Fedora 40 Kinoite

Dec 16 14:42:59 fedora rpm-ostree(akmod-nvidia.post)[94794]:   gcc -Wp,-MMD,/tmp/akmodsbuild.plkwH5m1/BUILD/nvidia-kmod-560.35.03/_kmod_build_6.12.4-100.fc40.x86_64/nvidia-drm/.nvidia-drm-gem-dma-buf.o.d -nostdinc -I./arch/x86/include >
Dec 16 14:42:59 fedora rpm-ostree(akmod-nvidia.post)[94794]: make[2]: *** [/usr/src/kernels/6.12.4-100.fc40.x86_64/Makefile:1976: /tmp/akmodsbuild.plkwH5m1/BUILD/nvidia-kmod-560.35.03/_kmod_build_6.12.4-100.fc40.x86_64] Error 2
Dec 16 14:42:59 fedora rpm-ostree(akmod-nvidia.post)[94794]: make[1]: *** [Makefile:236: __sub-make] Error 2
Dec 16 14:42:59 fedora rpm-ostree(akmod-nvidia.post)[94794]: make: *** [Makefile:89: modules] Error 2
Dec 16 14:42:59 fedora rpm-ostree(akmod-nvidia.post)[94794]: error: Bad exit status from /var/tmp/rpm-tmp.wmWORl (%build)
4

You will need nvidia to fix their software.

Alternatively you could use the rpmfusion packaged nvidia drivers that should not have this issue.

Do you have a reason not to use the rpmfusion drivers?

5

I think I had installed it via rpmfusion actually following this guide Howto/NVIDIA - RPM Fusion

only additional change I added Envycontrol using envycontrol/README.md at main · bayasdev/envycontrol · GitHub

so I’m not sure why is trying to compile it

6

After installing F40 with the default rpmfusion Nvidia drivers, my system was nearly completely unable to display graphics at all. The Gnome screen appeared, but there was a 5-15 second lag between mouse movements and keypresses and seeing the result on the screen (yes, I used a stopwatch). This was intolerable. (I’m not very patient.)

Everything worked fine when I booted to the command line, but that’s suboptimal.

Note: F39 and earlier did not present these problems on exactly the same hardware.

So I downloaded NVIDIA-Linux-x86_64-555.58.02.run from Nvidia and installed that, after which everything worked in F40.

I have yet to dare to try F41 …

7

Same issue here, Fedora 40 with Nvidia drivers from RPM fusion. Kernel 6.12.4 does not work with Nvidia drivers.

8

The drivers from rpmfusion-nonfree-updates-testing solved it for me

yum upgrade --enablerepo=*test* *nvidia*
2 Likes
9

I’m not using the nvidia driver, but the problem may be the same with the broadcom-wl driver and kernel 6.12:

Does anyone of you have the log file from akmodsbuild?

10

/var/cache/akmods/nvidia/560.35.03-1-for-6.12.4-100.fc40.x86_64.failed.log

11

This fixed it for me

12

thank you. it was more of a personal interest. there are already some patches prepared in the rpm fusion git repo if you look for the errors in the log file. so the solution here for those who need nvidia is to use the test repo for now. if you look at all the commits of the source code, you can only take your hat off to the maintainers who have to make adjustments under time pressure for every new kernel and driver version. huge respect.

13

I have no problem with the nvidia drivers and the 6.12.4 kernel.
I am using the 565.77 drivers from the rpmfusion-nonfree-updates-testing repo, and everything just worked for me perfectly.

14

How can I do that with rpm-ostree?

15

Seems like a lot of people having some kind of nvidia promleim with kernel 6.12.4 in Fedora 40. Check how i fixed it.

16

fixed today for me, seems like Fedora upgraded to 565. Good Job!

  rpm-ostree db diff
ostree diff commit from: booted deployment (27261bc27fa835435e27f9aabb8c09549e2e670bf53c99d9ea5b3cb71b955d75)
ostree diff commit to:   pending deployment (88a81f31911b67cd10c5627087ff486ffb3d45d4c3ced60be6b125015413d5dc)
Upgraded:
  akmod-nvidia 3:560.35.03-1.fc40 -> 3:565.77-1.fc40
  kernel 6.11.11-200.fc40 -> 6.12.4-100.fc40
  kernel-core 6.11.11-200.fc40 -> 6.12.4-100.fc40
  kernel-devel 6.11.11-200.fc40 -> 6.12.4-100.fc40
  kernel-devel-matched 6.11.11-200.fc40 -> 6.12.4-100.fc40
  kernel-headers 6.11.3-200.fc40 -> 6.12.4-100.fc40
  kernel-modules 6.11.11-200.fc40 -> 6.12.4-100.fc40
  kernel-modules-core 6.11.11-200.fc40 -> 6.12.4-100.fc40
  kernel-modules-extra 6.11.11-200.fc40 -> 6.12.4-100.fc40
  kmod-nvidia 3:560.35.03-1.fc40 -> 3:565.77-1.fc40
  libnfsidmap 1:2.8.1-2.rc2.fc40 -> 1:2.8.1-3.rc2.fc40
  libsoup3 3.4.4-5.fc40 -> 3.4.5-1.fc40
  nfs-utils 1:2.8.1-2.rc2.fc40 -> 1:2.8.1-3.rc2.fc40
  nvidia-modprobe 3:560.35.03-1.fc40 -> 3:565.77-1.fc40
  nvidia-persistenced 3:560.35.03-1.fc40 -> 3:565.77-1.fc40
  nvidia-settings 3:560.35.03-1.fc40 -> 3:565.77-1.fc40
  python3-boto3 1.35.77-1.fc40 -> 1.35.80-1.fc40
  python3-botocore 1.35.77-1.fc40 -> 1.35.80-1.fc40
  vim-data 2:9.1.906-1.fc40 -> 2:9.1.919-1.fc40
  vim-minimal 2:9.1.906-1.fc40 -> 2:9.1.919-1.fc40
  xorg-x11-drv-nvidia 3:560.35.03-5.fc40 -> 3:565.77-3.fc40
  xorg-x11-drv-nvidia-cuda 3:560.35.03-5.fc40 -> 3:565.77-3.fc40
  xorg-x11-drv-nvidia-cuda-libs 3:560.35.03-5.fc40 -> 3:565.77-3.fc40
  xorg-x11-drv-nvidia-kmodsrc 3:560.35.03-5.fc40 -> 3:565.77-3.fc40
  xorg-x11-drv-nvidia-libs 3:560.35.03-5.fc40 -> 3:565.77-3.fc40
  xorg-x11-drv-nvidia-power 3:560.35.03-5.fc40 -> 3:565.77-3.fc40
17

Does it by any chance recognize a non-existent second monitor on gnome display settings? I managed to install using the .run file last month, and had this happen to me. I’ve since went back to using akmods because of the updates.