rope/nixos
1
1
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

mediaserver: explicitly load NVIDIA kernel modules at boot

Browse source 
On a headless system, services.xserver.videoDrivers doesn't guarantee
the NVIDIA modules load. Add nvidia, nvidia_modeset, nvidia_uvm, and
nvidia_drm to boot.kernelModules explicitly.

nvidia_uvm is particularly critical — without it, CUDA initialisation
fails with CUDA_ERROR_NO_DEVICE, which is exactly what broke Jellyfin
NVENC transcoding.

https://claude.ai/code/session_016jJU8ZtWLSnJQBdbMr5pxK
This commit is contained in:
Claude 2026-04-15 07:28:57 +00:00
parent cac4bda2ab
commit a42d906758
No known key found for this signature in database
1 changed files with 4 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

5
hosts/hardware/FredOS-Mediaserver.nix
View file

@ -7,7 +7,10 @@
boot.initrd.availableKernelModules = [ "ahci" "xhci_pci" "ata_generic" "ehci_pci" "usbhid" "usb_storage" "sd_mod" "rtsx_usb_sdmmc" ];
boot.initrd.kernelModules = [ ];
boot.kernelModules = [ "kvm-intel" ];
# nvidia_uvm is required for CUDA (used by NVENC/NVDEC in Jellyfin).
# The other nvidia modules are loaded via services.xserver.videoDrivers but
# nvidia_uvm is not pulled in automatically on a headless system.
boot.kernelModules = [ "kvm-intel" "nvidia" "nvidia_modeset" "nvidia_uvm" "nvidia_drm" ];
boot.extraModulePackages = [ ];
fileSystems."/" = {