# Description: systemd-udevd will now indefinitely try to run nvidia-modprobe when nvidia-modprobe exits with an error. I have tracked it down to this commit: https://gitlab.archlinux.org/archlinux/packaging/packages/nvidia-utils/-/commit/4fe55fc13e8bb00a839ae09c4566974e1633cece I am already binding a VFIO dummy driver on boot, as described here: https://wiki.archlinux.org/title/PCI_passthrough_via_OVMF#Isolating_the_GPU This causes nvidia-modprobe to always exit with exit code 1 and systemd-udevd will spawn a lot of worker processes, hogging down the system and actually causing noticeable lag. Downgrading nvidia-utils to version 545.29.06-1 is a workaround. # Additional info: * package version(s): 545.29.06-3 * config and/or log files: [systemd-udevd.log](/uploads/2da36c20c8a2c47f154e97917067c4d6/systemd-udevd.log) * link to upstream bug report, if any: - # Steps to reproduce: Bind the VFIO dummy driver on boot and reboot. Maybe loading the nouveau driver will show the same behaviour, but I have not tested this.