nvidia: Process '/sbin/modprobe nvidia-uvm' failed with exit code 1.

[Todd Taft]

On 20.04, I'm getting repeated errors on my system when the nvdidia drivers are installed. This sequence of errors is logged multiple times per second:

Oct 7 01:45:29 kangaroo systemd-udevd[499101]: nvidia: Process '/sbin/modprobe nvidia-drm' failed with exit code 1.
Oct 7 01:45:29 kangaroo kernel: [26851.518031] nvidia-nvlink: Nvlink Core is being initialized, major device number 510
Oct 7 01:45:29 kangaroo kernel: [26851.518035] NVRM: The NVIDIA probe routine was not called for 1 device(s).
Oct 7 01:45:29 kangaroo kernel: [26851.518590] NVRM: This can occur when a driver such as:
Oct 7 01:45:29 kangaroo kernel: [26851.518590] NVRM: nouveau, rivafb, nvidiafb or rivatv
Oct 7 01:45:29 kangaroo kernel: [26851.518590] NVRM: was loaded and obtained ownership of the NVIDIA device(s).
Oct 7 01:45:29 kangaroo kernel: [26851.518590] NVRM: Try unloading the conflicting kernel module (and/or
Oct 7 01:45:29 kangaroo kernel: [26851.518590] NVRM: reconfigure your kernel without the conflicting
Oct 7 01:45:29 kangaroo kernel: [26851.518590] NVRM: driver(s)), then try loading the NVIDIA kernel module
Oct 7 01:45:29 kangaroo kernel: [26851.518590] NVRM: again.
Oct 7 01:45:29 kangaroo kernel: [26851.518591] NVRM: No NVIDIA devices probed.
Oct 7 01:45:29 kangaroo kernel: [26851.518746] nvidia-nvlink: Unregistered the Nvlink Core, major device number 510

Multiple systemd services report failures:
root@kangaroo:~# systemctl list-units --state=failed
  UNIT LOAD ACTIVE SUB DESCRIPTION
● nvidia-persistenced.service loaded failed failed NVIDIA Persistence Daemon
● plymouth-read-write.service loaded failed failed Tell Plymouth To Write Out Runtime Data
● systemd-udev-settle.service loaded failed failed udev Wait for Complete Device Initialization

ProblemType: Bug
DistroRelease: Ubuntu 20.04
Package: xserver-xorg-video-nvidia-465 (not installed)
ProcVersionSignature: Ubuntu 5.10.0-1049.51-oem 5.10.52
Uname: Linux 5.10.0-1049-oem x86_64
ApportVersion: 2.20.11-0ubuntu27.20
Architecture: amd64
CasperMD5CheckMismatches: ./pool/main/l/linux-hwe-5.11/linux-modules-extra-5.11.0-34-generic_5.11.0-34.36~20.04.1_amd64.deb
CasperMD5CheckResult: fail
Date: Thu Oct 7 01:43:15 2021
InstallationDate: Installed on 2021-10-06 (1 days ago)
InstallationMedia: Ubuntu-Server 20.04.3 LTS "Focal Fossa" - Release amd64 (20210914)
ProcEnviron:
 TERM=xterm-256color
 PATH=(custom, no user)
 LANG=en_US.UTF-8
 SHELL=/bin/bash
SourcePackage: nvidia-graphics-drivers-470
UpgradeStatus: No upgrade log present (probably fresh install)

[rasomc (rasomc)]

I experienced a similar issue in Ubuntu 22.04 with a newer driver version.

Please find the discussions & findings here:
https://forums.developer.nvidia.com/t/unknown-error-from-cuda-11-7-ubuntu-22-04-64bit/220384/16

In a nutshell:
Based on the modinfo output, nvidia as well as nvidia-uvm were signed with the same key but nvidia-uvm got rejected during secure boot, the other nvidia modules were loaded as expected.
Disabling secure boot solves the issue but that's not the way it should be.

I hope the info helps.