https://bugs.kde.org/show_bug.cgi?id=521814
--- Comment #9 from [email protected] --- Created attachment 193621 --> https://bugs.kde.org/attachment.cgi?id=193621&action=edit journalctl_kwin_udev_remove_trigger I tried the following : ❯ sudo udevadm trigger --verbose --action=remove --subsystem-match=drm --sysname-match=renderD129 --sysname-match=card0 /sys/devices/pci0000:00/0000:00:1b.4/0000:05:00.0/drm/card0 /sys/devices/pci0000:00/0000:00:1b.4/0000:05:00.0/drm/renderD129 and kwin_wayland did not release the gpu. Did some additional testing : With nvidia_drm.modeset=0 kernel parameter, kwin_wayland uses 2.3 MiB of VRAM at boot and after the udev remove trigger, and always holds the secondary gpu. The udev remove trigger does not change anything in vram used. Without nvidia_drm.modeset=0, kwin_wayland, uses 10.1 MiB of VRAM at boot. After executing the udev remove trigger via : ❯ sudo udevadm trigger --verbose --action=remove --subsystem-match=drm --sysname-match=renderD129 --sysname-match=card0 kwin reacts to the udev remove trigger in some way, and uses 2.3 MiB of VRAM after its execution, but still holds the gpu. It seems that after the udev remove trigger, I become in the same state I am when booting with nvidia_drm.modeset=0. I attached in a txt file the journalctl logs for kwin_wayland when booting with and without nvidia_drm.modeset=0, and running the udev remove trigger command. No additional logs are generated when running the udev remove trigger command with and without nvidia_drm.modeset=0. -- You are receiving this mail because: You are watching all bug changes.
