GPU not working after update

Ask Fedora
, , ,
1

Hi,

whenever I update, my gpu with CUDA is not working anymore at least with two app, minimon and dsnote.

Here are the log given by journalctl -b -g nvidia :

Summary

nov. 19 17:10:33 fedora kernel: Command line: BOOT_IMAGE=(hd2,gpt2)/ostree/fedora-00f8aae766ed59572f61a205588ddeef7cad345b424d251ef35ee256ec9f085a/vmlinuz-6.17.8-200.fc42.x86_64 ostree=/ostree/boot.1/fedora/00f8aae766ed59572f61a205588ddeef7cad345b424d251ef>
nov. 19 17:10:33 fedora kernel: Kernel command line: BOOT_IMAGE=(hd2,gpt2)/ostree/fedora-00f8aae766ed59572f61a205588ddeef7cad345b424d251ef35ee256ec9f085a/vmlinuz-6.17.8-200.fc42.x86_64 ostree=/ostree/boot.1/fedora/00f8aae766ed59572f61a205588ddeef7cad345b42>
nov. 19 17:10:33 fedora dracut-cmdline[270]: Using kernel command line parameters: rd.driver.pre=btrfs BOOT_IMAGE=(hd2,gpt2)/ostree/fedora-00f8aae766ed59572f61a205588ddeef7cad345b424d251ef35ee256ec9f085a/vmlinuz-6.17.8-200.fc42.x86_64 ostree=/ostree/boo>
nov. 19 17:10:38 lo kernel: nvidia: loading out-of-tree module taints kernel.
nov. 19 17:10:38 lo kernel: nvidia: module license ‘NVIDIA’ taints kernel.
nov. 19 17:10:38 lo kernel: nvidia: module verification failed: signature and/or required key missing - tainting kernel
nov. 19 17:10:38 lo kernel: nvidia: module license taints kernel.
nov. 19 17:10:39 lo kernel: nvidia-nvlink: Nvlink Core is being initialized, major device number 234
nov. 19 17:10:39 lo kernel: nvidia 0000:01:00.0: vgaarb: VGA decodes changed: olddecodes=io+mem,decodes=none:owns=io+mem
nov. 19 17:10:39 lo kernel: input: HDA NVidia HDMI/DP,pcm=3 as /devices/pci0000:00/0000:00:01.1/0000:01:00.1/sound/card0/input26
nov. 19 17:10:39 lo kernel: input: HDA NVidia HDMI/DP,pcm=7 as /devices/pci0000:00/0000:00:01.1/0000:01:00.1/sound/card0/input27
nov. 19 17:10:39 lo kernel: input: HDA NVidia HDMI/DP,pcm=8 as /devices/pci0000:00/0000:00:01.1/0000:01:00.1/sound/card0/input28
nov. 19 17:10:39 lo kernel: input: HDA NVidia HDMI/DP,pcm=9 as /devices/pci0000:00/0000:00:01.1/0000:01:00.1/sound/card0/input29
nov. 19 17:10:39 lo kernel: NVRM: loading NVIDIA UNIX x86_64 Kernel Module 580.95.05 Tue Sep 23 10:11:16 UTC 2025
nov. 19 17:10:39 lo kernel: nvidia_uvm: module uses symbols nvUvmInterfaceDisableAccessCntr from proprietary module nvidia, inheriting taint.
nov. 19 17:10:39 lo audit[1]: SERVICE_START pid=1 uid=0 auid=4294967295 ses=4294967295 subj=system_u:system_r:init_t:s0 msg=‘unit=nvidia-powerd comm=“systemd” exe=“/usr/lib/systemd/systemd” hostname=? addr=? terminal=? res=success’
nov. 19 17:10:39 lo systemd[1]: Started nvidia-powerd.service - nvidia-powerd service.
nov. 19 17:10:39 lo nvidia-powerd[911]: nvidia-powerd version:2.0 (build 1)
nov. 19 17:10:39 lo kernel: nvidia-modeset: Loading NVIDIA Kernel Mode Setting Driver for UNIX platforms 580.95.05 Tue Sep 23 09:41:17 UTC 2025
nov. 19 17:10:40 lo kernel: [drm] [nvidia-drm] [GPU ID 0x00000100] Loading driver
nov. 19 17:10:40 lo systemd[1]: nvidia-powerd.service: Deactivated successfully.
nov. 19 17:10:40 lo audit[1]: SERVICE_STOP pid=1 uid=0 auid=4294967295 ses=4294967295 subj=system_u:system_r:init_t:s0 msg=‘unit=nvidia-powerd comm=“systemd” exe=“/usr/lib/systemd/systemd” hostname=? addr=? terminal=? res=success’
nov. 19 17:10:40 lo kernel: [drm] Initialized nvidia-drm 0.0.0 for 0000:01:00.0 on minor 0
nov. 19 17:10:40 lo kernel: nvidia 0000:01:00.0: vgaarb: deactivate vga console
nov. 19 17:10:40 lo kernel: fbcon: nvidia-drmdrmfb (fb0) is primary device
nov. 19 17:10:40 lo kernel: nvidia 0000:01:00.0: [drm] fb0: nvidia-drmdrmfb frame buffer device
nov. 19 17:10:40 lo systemd[1]: nvidia-fallback.service - Fallback to nouveau as nvidia did not load was skipped because of an unmet condition check (ConditionPathExists=!/sys/module/nvidia).
nov. 19 17:10:40 lo systemd[1]: nvidia-fallback.service - Fallback to nouveau as nvidia did not load was skipped because of an unmet condition check (ConditionPathExists=!/sys/module/nvidia).
nov. 19 17:13:08 lo systemd[1366]: Starting app-nvidia\x2dsettings\x2duser@autostart.service - nvidia-settings…
nov. 19 17:13:08 lo systemd[1366]: Started app-nvidia\x2dsettings\x2duser@autostart.service - nvidia-settings.
nov. 19 17:13:09 lo systemd[1366]: app-nvidia\x2dsettings\x2duser@autostart.service: Main process exited, code=exited, status=1/FAILURE
nov. 19 17:13:09 lo systemd[1366]: app-nvidia\x2dsettings\x2duser@autostart.service: Failed with result ‘exit-code’.
nov. 19 17:13:09 lo systemd[1366]: app-nvidia\x2dsettings\x2duser@autostart.service: Consumed 204ms CPU time, 72.3M memory peak.
nov. 19 17:13:13 lo cosmic-applet-minimon[2385]: Connection to NVML failed, reason: a libloading error occurred: libnvidia-ml.so.1: cannot open shared object file: No such file or directory
nov. 19 17:13:13 lo cosmic-applet-minimon[2385]: No Nvidia GPUs found
nov. 19 17:13:13 lo cosmic-applet-minimon[2385]: 01:00.0 → NVIDIA GP107 (GeForce GTX 1050 Ti) (10de1c82) (rev a1)
nov. 19 17:19:06 lo sudo[12663]: lo : TTY=pts/0 ; PWD=/var/home/lo ; USER=root ; COMMAND=/usr/sbin/modprobe nvidia

and after rollback, with the working configuration :

Summary

nov. 19 17:21:29 fedora kernel: Command line: BOOT_IMAGE=(hd2,gpt2)/ostree/fedora-bec4e003bd3baf8fa5c61a0ab62df500d6c821876b435248065c7aa52468559a/vmlinuz-6.16.10-200.fc42.x86_64 ostree=/ostree/boot.0/fedora/bec4e003bd3baf8fa5c61a0ab62df500d6c821876b435248>
nov. 19 17:21:29 fedora kernel: Kernel command line: BOOT_IMAGE=(hd2,gpt2)/ostree/fedora-bec4e003bd3baf8fa5c61a0ab62df500d6c821876b435248065c7aa52468559a/vmlinuz-6.16.10-200.fc42.x86_64 ostree=/ostree/boot.0/fedora/bec4e003bd3baf8fa5c61a0ab62df500d6c821876>
nov. 19 17:21:29 fedora dracut-cmdline[288]: Using kernel command line parameters: rd.driver.pre=btrfs BOOT_IMAGE=(hd2,gpt2)/ostree/fedora-bec4e003bd3baf8fa5c61a0ab62df500d6c821876b435248065c7aa52468559a/vmlinuz-6.16.10-200.fc42.x86_64 ostree=/ostree/bo>
nov. 19 17:21:33 lo kernel: nvidia: loading out-of-tree module taints kernel.
nov. 19 17:21:33 lo kernel: nvidia: module license ‘NVIDIA’ taints kernel.
nov. 19 17:21:33 lo kernel: nvidia: module verification failed: signature and/or required key missing - tainting kernel
nov. 19 17:21:33 lo kernel: nvidia: module license taints kernel.
nov. 19 17:21:34 lo kernel: nvidia-nvlink: Nvlink Core is being initialized, major device number 235
nov. 19 17:21:34 lo kernel: nvidia 0000:01:00.0: vgaarb: VGA decodes changed: olddecodes=io+mem,decodes=none:owns=io+mem
nov. 19 17:21:34 lo kernel: input: HDA NVidia HDMI/DP,pcm=3 as /devices/pci0000:00/0000:00:01.1/0000:01:00.1/sound/card0/input21
nov. 19 17:21:34 lo kernel: input: HDA NVidia HDMI/DP,pcm=7 as /devices/pci0000:00/0000:00:01.1/0000:01:00.1/sound/card0/input22
nov. 19 17:21:34 lo kernel: input: HDA NVidia HDMI/DP,pcm=8 as /devices/pci0000:00/0000:00:01.1/0000:01:00.1/sound/card0/input23
nov. 19 17:21:34 lo kernel: input: HDA NVidia HDMI/DP,pcm=9 as /devices/pci0000:00/0000:00:01.1/0000:01:00.1/sound/card0/input24
nov. 19 17:21:34 lo kernel: NVRM: loading NVIDIA UNIX x86_64 Kernel Module 580.82.09 Fri Aug 29 17:44:50 UTC 2025
nov. 19 17:21:34 lo kernel: nvidia_uvm: module uses symbols nvUvmInterfaceDisableAccessCntr from proprietary module nvidia, inheriting taint.
nov. 19 17:21:34 lo audit[1]: SERVICE_START pid=1 uid=0 auid=4294967295 ses=4294967295 subj=system_u:system_r:init_t:s0 msg=‘unit=nvidia-powerd comm=“systemd” exe=“/usr/lib/systemd/systemd” hostname=? addr=? terminal=? res=success’
nov. 19 17:21:34 lo systemd[1]: Started nvidia-powerd.service - nvidia-powerd service.
nov. 19 17:21:34 lo nvidia-powerd[897]: nvidia-powerd version:2.0 (build 1)
nov. 19 17:21:35 lo kernel: nvidia-modeset: Loading NVIDIA Kernel Mode Setting Driver for UNIX platforms 580.82.09 Fri Aug 29 17:11:30 UTC 2025
nov. 19 17:21:35 lo kernel: [drm] [nvidia-drm] [GPU ID 0x00000100] Loading driver
nov. 19 17:21:35 lo systemd[1]: nvidia-powerd.service: Deactivated successfully.
nov. 19 17:21:35 lo audit[1]: SERVICE_STOP pid=1 uid=0 auid=4294967295 ses=4294967295 subj=system_u:system_r:init_t:s0 msg=‘unit=nvidia-powerd comm=“systemd” exe=“/usr/lib/systemd/systemd” hostname=? addr=? terminal=? res=success’
nov. 19 17:21:36 lo kernel: [drm] Initialized nvidia-drm 0.0.0 for 0000:01:00.0 on minor 0
nov. 19 17:21:36 lo kernel: nvidia 0000:01:00.0: vgaarb: deactivate vga console
nov. 19 17:21:36 lo kernel: fbcon: nvidia-drmdrmfb (fb0) is primary device
nov. 19 17:21:36 lo kernel: nvidia 0000:01:00.0: [drm] fb0: nvidia-drmdrmfb frame buffer device
nov. 19 17:21:36 lo systemd[1]: nvidia-fallback.service - Fallback to nouveau as nvidia did not load was skipped because of an unmet condition check (ConditionPathExists=!/sys/module/nvidia).
nov. 19 17:21:36 lo systemd[1]: nvidia-fallback.service - Fallback to nouveau as nvidia did not load was skipped because of an unmet condition check (ConditionPathExists=!/sys/module/nvidia).
nov. 19 17:25:03 lo systemd[1363]: Starting app-nvidia\x2dsettings\x2duser@autostart.service - nvidia-settings…
nov. 19 17:25:03 lo systemd[1363]: Started app-nvidia\x2dsettings\x2duser@autostart.service - nvidia-settings.
nov. 19 17:25:04 lo systemd[1363]: app-nvidia\x2dsettings\x2duser@autostart.service: Main process exited, code=exited, status=1/FAILURE
nov. 19 17:25:04 lo systemd[1363]: app-nvidia\x2dsettings\x2duser@autostart.service: Failed with result ‘exit-code’.
nov. 19 17:25:04 lo systemd[1363]: app-nvidia\x2dsettings\x2duser@autostart.service: Consumed 209ms CPU time, 76.9M memory peak.
nov. 19 17:25:08 lo cosmic-applet-minimon[2433]: 01:00.0 → NVIDIA GP107 (GeForce GTX 1050 Ti) (10de1c82) (rev a1)
nov. 19 17:25:08 lo cosmic-applet-minimon[2433]: Found GPU. Name: NVIDIA GeForce GTX 1050 Ti. UUID: GPU-1de5a7c2-16ef-8a3f-720e-f71d5e12657

As I need the GPU, I can not update the system.

Do you know how to fix this ?
Thank you :slight_smile:

3

According to those two lines it appears the nvidia driver was loaded.

However, the first also shows that the driver you have loaded is quite out-of-date. The driver that works and that I have not seen any problem reports for is the 580.95.05 which is available from the rpmfusion-nonfree repo.
The latest drivers (580.105.08) from the rpmfusion-nonfree-nvidia-driver repo or the rpmfusion-nonfree-updates repo has some reported bugs.

I believe the 580.105.08 driver has an issue with ultra-wide monitors but works normally for 16:9 resolutions. It works fine for me with a 4k 16:9 monitor (3840x2160)

1 Like
4

Aren’t these lines an indication that the nvidia driver was not successfully loaded and that the system used the nouveau driver instead ?

5

Absolutely not!

This line (which is part of that message) shows that the path /sys/module/nvidia exists, whiich caused the fallback to nouveau to fail. This means the nvidia driver was already loaded as is shown by the message that I quoted just before.

In other words, if that path did not exist then the system would fall back to the nouveau driver.

6

Thank you for you answers.

As a matter of fact the GPU is found and use by some software/app like gnome view (flatpak) but not found by others, like dsnote (flatpak) or cosmic minimon-applet (rpm).

I have no idea how come.

Here are screenshot from nvidia-smi (the first one, with 580.95.05, is the non working with some app one) :

Screenshot_2025-11-20_17-43-15
Screenshot_2025-11-20_17-43-15908×622 65.5 KB

Screenshot_2025-11-20_17-40-22
Screenshot_2025-11-20_17-40-22926×640 80.9 KB

Maybe I’ll wait for new updates to fix this, in particular for the different involved repository to keep up so that each drivers and librairies or whatever, versions to matches with one another…

7

have you run flatpak update?

$ flatpak list |grep org.freedesktop.Platform.GL.nvidia
nvidia-580-105-08	org.freedesktop.Platform.GL.nvidia-580-105-08		1.4	system

this has to be in sync with the installed nvidia drivers.

8

They are in sync in my actual setup, but might not be so after system updates.

Maybe the org.freedesktop flatpak have not yet been updated to 580.95.05 in the repo and that I have to wait for that to happens. Indeed I was wrong to state that minimon is rpm, it is actually a flatpak, so it seems that flatpaks are indeed the ones that does not recognize the GPU.

9

again, you have to run flatpak update.
Why do you think the installed version on my system is 580-105-08?

so update nvidia drivers, wait for the kernel modules to be built. Reboot, update flatpaks.

10

Thanks, you’re right, flatpak updates will somewhat adjust to the new driver versions if I understand what you are saying.

So I’ll try to update rpm-ostree then update flatpak and reboot to see how it goes.

I am not sure though when the kernel module will be rebuild, if it is just after rpm-ostree update, or after the first boot in the new tree.

11

not sure when they are built on a rpm-ostree system. Check with nvidia-smi that the new driver is working and then run flatpak update. This should pull the matching nvidia userland libraries.

1 Like
12

Thanks alot. That was it. The flatpak did adjust to the new driver version and the flatpak app now find the GPU.

2 Likes
13

Note that you edited the solution into the initial post.
It repeats what was said by Mark in the post you marked as the solution so becomes confusing.

To keep the conversation and events in time sequence it would seem better if you left things in the order they occur so readers are not confused by out-of-order comments.

Glad a solution was identified. :+1:

1 Like
14

ok, I removed the edit to avoid confusion.

Here is the summary of the problem+solution :

Problem : after fedora update, the flatpak installed apps do find the GPU while the driver is ok for the other app as shown by the command : nvidia-smi

Both versions sent back by this command must be the same and they are no more after the update : flatpak list |grep org.freedesktop.Platform.GL.nvidia

Solution : flatpak drivers or whatever have to be update separately AFTER fedora update to keep up with the new driver version. After reboot, check with “nvidia-smi” that the driver and GPU are ok, then use : “flatpak update”, then reboot.

3 Likes
15

I might note, that although I dislike using the software app (it is both gui and still uses the older dnf4 backend) it does have the benefit when doing updates in that it updates both the rpm and the flatpak packages. When doing updates on the cli using dnf (which is now dnf5) it is also necessary to update with flatpak separately.

EDIT
Ignore this. I missed the fact that this is an atomic system and uses rpm-ostree so the dnf commands are not valid.

16

I doubt that this would make any difference in this case. Flatpak update downloads the appropriate nvidia version that matches the currently active driver. It’s always a two step process.

17

Also with an rpm-ostree (atomic) system, the flatpaks have to be updated separately.

1 Like