Created attachment 114822 [details] /sys/class/drm/card0/error A Qt application suddently hung and stopped redrawing, a few seconds later my screen was black quickly and I got this in my dmesg: kern :info : [Apr 1 21:34] [drm] stuck on render ring kern :info : [ +0.001517] [drm] GPU HANG: ecode 0:0x95fdfffd, in systemd-logind [424], reason: Ring hung, action: reset kern :info : [ +0.000005] [drm] GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace. kern :info : [ +0.000003] [drm] Please file a _new_ bug report on bugs.freedesktop.org against DRI -> DRM/Intel kern :info : [ +0.000003] [drm] drm/i915 developers can then reassign to the right component if it's not a kernel issue. kern :info : [ +0.000002] [drm] The gpu crash dump is required to analyze gpu hangs, so please always attach it. kern :info : [ +0.000003] [drm] GPU crash dump saved to /sys/class/drm/card0/error kern :info : [ +5.990725] [drm] stuck on render ring kern :info : [ +0.002213] [drm] GPU HANG: ecode 0:0x85effffd, in systemd-logind [424], reason: Ring hung, action: reset kern :err : [ +0.000138] [drm:i915_context_is_banned] *ERROR* gpu hanging too fast, banning! This is on Archlinux, Linux 3.18.6-1-ARCH, X.org 1.17.1-3, xf86-video-intel 2.99.917-4.
That's puzzling. The hang was just in some standard sub-pixel text rendering, all the surfaces are valid and the state looks intact. Then once the GPU died, it died for good. Can you reproduce this? (Is the Qt application a good test case?) Could you then try a few different kernels just to see if it a regression there?
I can't reproduce it, no. I'm using that application (qutebrowser) for months (I'm the developer, heh :D), but this is the first time I've seen this. Could this be somehow related to the system being low on (or out of) memory? qutebrowser/QtWebKit has a bug where it uses gigabytes of memory on a certain page - the OOM killer got invoked 12 minutes before that bug, and it's possible I was on that page again when it happened (though I don't remember, and there was no OOM killer this time).... So yeah, seems unlikely, but that's the only thing coming to mind.
Seems unlikely we'll be able to fix the bug if it is impossible to reproduce. Please reopen if you can reproduce it again.
I've never seen it again either - so I understand. Thanks for your effort!
Use of freedesktop.org services, including Bugzilla, is subject to our Code of Conduct. How we collect and use information is described in our Privacy Policy.