Created attachment 96087 [details]
GPU crash dump
I had this GPU crash on a Debian Linux PC:
Mar 20 09:33:45 yuggs kernel: [ 42.772483] [drm] stuck on render ring
Mar 20 09:33:45 yuggs kernel: [ 42.772488] [drm] GPU crash dump saved to /sys/class/drm/card0/error
Mar 20 09:33:45 yuggs kernel: [ 42.772489] [drm] GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace.
Mar 20 09:33:45 yuggs kernel: [ 42.772490] [drm] Please file a _new_ bug report on bugs.freedesktop.org against DRI -> DRM/Intel
Mar 20 09:33:45 yuggs kernel: [ 42.772491] [drm] drm/i915 developers can then reassign to the right component if it's not a kernel issue.
Mar 20 09:33:45 yuggs kernel: [ 42.772491] [drm] The gpu crash dump is required to analyze gpu hangs, so please always attach it.
Mar 20 09:33:45 yuggs kernel: [ 42.775090] [drm:i915_set_reset_status] *ERROR* render ring hung inside bo (0x2898000 ctx 1) at 0x2898220
xorg : 1:7.7+6
xserver-xorg-core : 2:1.15.0-2
xserver-xorg-video-intel : 2:2.21.15-2+b1
Attached the crash dump file and dmesg
Created attachment 96088 [details]
Before writing it off as another blorp hang probably fixed in 10.0 or 10.1, have a look at the state and wonder.
Chris, did you have a specific observation about the state you'd like to point out?
It does strike me as a bit odd that the PIPE_CONTROL which supposedly hung is in the batch, and appears to be the start of a valid workaround sequence. i.e. this /is/ the thing we've been applying to fix problems.
Please reopen if you can still reproduce with Mesa 17.0.