System
-
Distribution: Fedora Linux 43 KDE Plasma Desktop Edition
-
Session: KDE Plasma / Wayland
-
GPU: NVIDIA GeForce GTX 960, GM206, PCI ID
10de:1401 -
NVIDIA driver: proprietary
580.173.02 -
Kernel when the failure occurred:
7.1.4-104.fc43.x86_64 -
Display resolution: 1920×1080
Problem
The display froze while the computer was idle and a screensaver had been running for an extended period. There was no suspend/resume, monitor hot-plug, scheduled job, or user activity at the time.
The machine remained accessible over SSH, but the graphical display was completely frozen.
The first relevant kernel message was:
Jul 27 09:38:00 kernel: nvidia-modeset: ERROR: GPU:0: Idling display engine timed out: 0x0000957d:0:0:1255
Shortly afterward:
kwin_wayland: The main thread was hanging temporarily!
kwin_wayland: Pageflip timed out! This is a bug in the nvidia-drm kernel driver
The NVIDIA display-engine timeout then repeated multiple times, including:
nvidia-modeset: ERROR: GPU:0: Idling display engine timed out: 0x0000927c:3:0:1264
nvidia-smi did not respond. A SysRq blocked-task dump showed both the NVIDIA modeset thread and nvidia-smi in uninterruptible D state:
task:nvidia-modeset/ state:D
task:nvidia-smi state:D
No NVRM: Xid message was logged, and there was no “GPU has fallen off the bus” or PCIe AER error.
Running nvidia-bug-report.sh normally produced only the report header, apparently because one of its GPU-query operations blocked. Running:
sudo nvidia-bug-report.sh --safe-mode
successfully produced the attached report while the GPU/display stack was still hung.
A normal reboot did not complete in a reasonable amount of time, and I eventually had to use a power button.
nvidia-bug-report.log.gz (170.4 KB)
The attached safe-mode bug report was captured during the failure.