GPU HANG: ecode 7:1:62426161
Since a few weeks I'm experiencing GPU HANG issues, aparrently always originating from accelerated contexts (chromium, zoom, ...). I finally managed to grab a crash dump which I am attaching here.
How to reproduce
Use machine normally, with GPU accelerated applications. I managed to trigger this in chromium, but also for example in a Zoom conference. At some point (haven't been able to pinpoint the exact action yet...) the whole display (including mouse cursor etc) will freeze for a couple of seconds, and then become responsible again. From then on, graphics artifacts are visible in different applications, which change appearance when scrolling for example.
This can happen within minutes or days of usage, and restarting the app that caused the HANG does not fix it (machine needs to be rebooted).
In dmesg, I find messages such as this:
[113756.316347] i915 0000:00:02.0: [drm] GPU HANG: ecode 7:1:62426161, in chromium [55298]
[113756.316353] GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace.
[113756.316355] Please file a _new_ bug report at https://gitlab.freedesktop.org/drm/intel/issues/new.
[113756.316357] Please see https://gitlab.freedesktop.org/drm/intel/-/wikis/How-to-file-i915-bugs for details.
[113756.316359] drm/i915 developers can then reassign to the right component if it's not a kernel issue.
[113756.316361] The GPU crash dump is required to analyze GPU hangs, so please always attach it.
[113756.316364] GPU crash dump saved to /sys/class/drm/card0/error
[113756.316521] i915 0000:00:02.0: [drm] Resetting chip for stopped heartbeat on rcs0
[113756.418269] i915 0000:00:02.0: [drm] chromium[55298] context reset due to GPU hangOther Information
- Kernel version:
Linux xxx 5.10.14-arch1-1 #1 SMP PREEMPT Sun, 07 Feb 2021 22:42:17 +0000 x86_64 GNU/Linux - Chromium version:
Version 88.0.4324.150 (Official Build) Arch Linux (64-bit) - Machine: Lenovo X240, happens with the internal display (but also when having an external display attached on DP-2-3, the Docking-Station VGA port)
This problem has been occurring since a couple of kernel versions, and I have recently also gotten rid of xf86-video-intel (which is apparently often causing issues and is not recommended?), but it did not make a difference.