Hello kind people,
I recently installed Fedora 39 KDE on my laptop, which has a dedicated Nvidia Gtx 1650ti GPU. As I planned to game a fair bit on it, I installed the proprietary Nvidia drivers through rpmfusion and signed them to work with safe boot; both Steam and Heroic Launcher worked flawlessly with every game I tried on default settings; I didn’t even need to switch to X11 as the games I play worked perfectly on Wayland. After a couple days of normal use (installing flatpacks and updating software mainly), almost every game I try to launch since on both platforms gives the following error:
Additionally, the error pops up on both Wayland and X11 sessions (so I assume it’s not a Wayland-only issue). Searching online I just found some rough fixes like disabling the DXVK/VKD3D options on Heroic and setting PROTON_USE_WINED3D11=1 %command% as a launch option on Steam; these of course aren’t long term solutions (the games work but don’t behave correctly), though they are essentially doing the same thing I think. From my understanding the underlying problem involves Vulkan rendering, but I can’t for the life of me fathom why they were working in the beginning and then just stop functioning properly.
Now, I’m a beginner user (switched full time on Linux just some weeks ago), and don’t have the abilities to dig deeper into this, so I’m asking for help here to not become insane.
I don’t really know which logs would be most useful to inspect, so I will produce and send them upon request. Thank you all
You stated that you have done updates.
Doing so while a kernel update is involved has the potential to corrupt the drivers should a reboot be performed too soon after the update is completed.
Please try this and let us know the result.
sudo dnf remove kmod-nvidia-$(uname -r) to remove the problematic drivers
then
sudo akmods --force to rebuild and reinstall the drivers.
I note that the cmdline does not contain the often needed option nvidia-drm.modeset=1.
You might cause the grub menu to display for boot with the command sudo grub2-editenv - unset menu_auto_hide. When you reboot the grub menu should now be displayed.
(You can cause the grub menu to be hidden again with sudo grub2-editenv - set menu_auto_hide=1)
Once the grub menu is displayed it is easy to press the e key to edit the commands for booting for temporary testing.
To test if this option may be helpful you can edit the line that begins with linux and add that option (nvidia-drm.modeset=1) to that line then continue booting.
If it solves the issue with monitor resolution then you could make it permanent by editing the file /etc/default/grub, add that option to the line that begins with GRUB_CMDLINE_LINUX, then run the command sudo grub2-mkconfig -o /boot/grub2/grub.cfg. This should make that option a part of every subsequent boot.
I’m very sorry to report that setting nvidia-drm.modeset=1 from grub did not solve the issue. Should I enable permanently the boot command anyways?
Yesterday I also noticed that the issue temporarily solved itself before shutting down the system for the day, but it did so without my intervention whatsoever, while starting the laptop today reverted it to the faulty state. I can include the log of one of the games launching with the error if it is useful in pinpointing the issue.
Good day, I tried booting fedora with kernel 6.7 and the issue persisted. Booting with the previous version (6.5) didn’t load the Nvidia drivers (classic “Kernel module missing” error) and loaded nouveau, in which case the games started up without errors. Considering the nouveau drivers were loaded, I don’t think this really worked. Should I try again after disabling safe boot?
Please check the status of secureboot. mokutil --sb-state should show that clearly.
Also show us the actual kernel in use. uname -r You show a 6.8.4 kernel in the inxi output, you mention a 6.7 kernel in the post just above, and you mention a 6.5 kernel in the same post. F39 has progressed from the 6.5.6 kernel with release in November to the 6.8.4 kernel release in the last week, with a lot of kernel updates over the past 5 months. If you still have the 6.5.6 kernel installed I have to wonder why you have not been routinely performing updates.
We find it difficult to follow symptoms if the system is not running consistently in the same (and most recent) environment, and problems are often solved (or introduced) with system updates. Changing the kernels used to boot presents problems of its own in troubleshooting.
Please boot to the latest kernel then we can perform troubleshooting with the system updated and in a known config. If you cannot be patient while troubleshooting like that then it is almost impossible to analyze the situation.
My apologies for not explaining in a clearer manner. All my considerations since the post has been written were done while on the latest kernel (6.8). I tried booting with older kernels as suggested by user @ledeni just to see if the issue persisted, as explored in my most recent answer: it was not meant to create confusion. Everything I described so far happens on the latest and most recent kernel version that I adopted since it rolled out; from what I can tell, I have performed system updates as soon as they were notified to be available from the update utility.
Output of mokutil --sb-state reports that secure boot is enabled, and uname -r shows the following:
Then I would suggest that you add the option noted above as a permanent addition to the kernel command line by editing /etc/default/grub as indicated above and updating the grub.cfg file.
Also check that the nvidia drivers are loaded with lsmod | grep -iE "nvidia|nouveau"
