System hanging on Nvidia graphics with kernels after 5.19.16

Starting with Kernel 5.19.16, the system hangs in LUKS password screen (right after choosing Kernel in Grub). The system works fine with 5.19.15.

The system will boot if I connect the monitor to the integrated HDMI/Display port on the motherboard (which uses AMD integrated driver), but then will fail to load the Graphical desktop environment, because that is set to use Nvidia.

Another interesting observation is that when using the AMD graphics, the kernel 5.19.15 does not switch to a high resolution, while the newer kernels all switch to a high resolution. So a theory can be: the newer kernels try to switch to high resolution but fail on Nvidia.

Some other observations: I have installed, but when I run dkms status I get:

nvidia/520.61.05, 5.19.15-201.fc36.x86_64, x86_64: installed
nvidia/520.61.05, 5.19.16-200.fc36.x86_64, x86_64: installed
nvidia/520.61.05, 6.0.5-200.fc36.x86_64, x86_64: installed
nvidia/520.61.05, 6.1.0-0.rc2.20221027gitb229b6ca5abb.23.vanilla.1.fc36.x86_64, x86_64: installed

Please note that this shows a newer driver (520.61.05) than the one I downloaded (520.56.06) and there is no newer file on NVidia website.

When I try to re-run the installer, it complains that the NVidia module is already loaded (even if I boot to run level 3). But if I want to uninstall the driver it complains that it is not installed.

This crash also happens on rawhide and vanilla kernels.

Operating System: Fedora Linux 36
KDE Plasma Version: 5.25.5
KDE Frameworks Version: 5.99.0
Qt Version: 5.15.6
Kernel Version: 5.19.15-201.fc36.x86_64 (64-bit)
Graphics Platform: X11
Processors: 32 × AMD Ryzen 9 7950X 16-Core Processor
Memory: 30.5 GiB of RAM
Graphics Processor: NVIDIA Graphics Device/PCIe/SSE2 (Rtx 4090, Gigabyte)
Manufacturer: ASUS (ROG Strix X670e-e)

I also have the same problem with Fedora 37 live, booting from usb: it starts correctly when monitor is connected to the integrated AMD graphics, but not when connected to Nvidia.

I note that you installed nvidia from the nvidia .run file.
We cannot support the nvidia drivers installed in that manner since they are not tested nor configured to be 100% compatible with fedora.

Alternatively we can suggest that you remove the driver installed from the .run file and instead install the driver from rpmfusion. Those rpms are tweaked and tested to function with fedora. They also, since installed using a fedora compatible repo, are designed to automatically receive updates of driver version and to automatically be updated on the system with every kernel update so the user does not have to recompile every time a new kernel update is recieved.

Instructions for configuring the repos are here, and then the drivers can be installed with a simple sudo dnf install akmod-nvidia xorg-x11-drv-nvidia-cuda. Note that the drivers installed directly from nvidia must be removed first to ensure no conflicts.


Hello Jeff.

Thank you. I’m now on Fedora 37. X11 works fairly well, but Wayland goes black screen right after login. Is there any additional package to install for Wayland?


I see you are using an RTX 4090 which is a very new GPU. It is quite possible that the nvidia driver in rpmfusion is not yet updated to fully support that card with the newer kernels, which you see as issues with wayland. As long as nvidia supports it for linux it does not take long for the drivers to be updated at rpmfusion once released by nvidia.

You may continue using xorg temporarily and wait for the driver update from rpmfusion. Then try wayland again to see if it is fixed.

1 Like

Thank you Jeff.

I can confirm that the open source NVidia driver is working smoothly in general. After some updates I was also able to boot in Wayland. The system had a weird crash, however. So I’m back to X11 for some time.