GPU HANG: ecode 9:1:85dffffb on kernel 5.13.8
To whom it may concern,
I've met some problem on OpenGL applications.
The screen freezes, but the sound and interaction are normal, except that the screen output does not change. You can switch to the kernel terminal with the Ctrl-Alt-Fx key combination, but even after the process that caused the problem is killed, the screen display still freezes, and does not recover.
Kernel has message like this:
[ 111.900829] Asynchronous wait on fence 0000:00:02.0:kwin_x11[2517]:e88 timed out (hint:intel_atomic_commit_ready [i915])
[ 114.621866] i915 0000:00:02.0: [drm] Resetting rcs0 for preemption time out
[ 114.621883] i915 0000:00:02.0: [drm] valley_x64[3756] context reset due to GPU hang
[ 114.651360] i915 0000:00:02.0: [drm] GPU HANG: ecode 9:1:85dffffb, in valley_x64 [3756]
[ 115.549923] i915 0000:00:02.0: [drm] Resetting rcs0 for preemption time out
[ 115.549949] i915 0000:00:02.0: [drm] valley_x64[3756] context reset due to GPU hang
[ 115.582029] i915 0000:00:02.0: [drm] GPU HANG: ecode 9:1:87f97cf9, in valley_x64 [3756]
Following these steps can always reproduce this bug on my machine:
- Download and extract the Valley Benchmark program
Valley Benchmark is a free performance testing software. And a video game called Sid Meier’s Civilization VI can trigger this bug, too. - Run it.
My system information:
OS: openSUSE Tumbleweed x86_64
Host: 20QNA00TCD ThinkPad P53
Kernel: 5.13.8-1-default
DE: Plasma 5.22.4
WM: KWin
CPU: Intel i5-9400H (8) @ 2.500GHz
GPU: Intel CoffeeLake-H GT2 (UHD Graphics 630)
Resolution: 1920x1080
Display connector: eDP (Laptop Builtin Screen)
Log and error dump are here: report-5.13.8-1-default.zip
I'm still waiting for the drm-tip kernel to compile on OBS. As long as it finished, I will test this again with drm-tip kernel.
Edited by Yao Nianqing