[snb] GPU hang in 'Heroes of Hammerwatch'
I am seeing some hard GPU hangs here while playing (and I don't think it's related with a specific game).
For example, twice I got hangs while playing Heroes of Hammerwatch:
[sáb jun 27 01:59:11 2020] i915 0000:00:02.0: GPU HANG: ecode 6:1:85fffffc, in HWR [27910]
[sáb jun 27 01:59:11 2020] GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace.
[sáb jun 27 01:59:11 2020] Please file a _new_ bug report at https://gitlab.freedesktop.org/drm/intel/issues/new.
[sáb jun 27 01:59:11 2020] Please see https://gitlab.freedesktop.org/drm/intel/-/wikis/How-to-file-i915-bugs for details.
[sáb jun 27 01:59:11 2020] drm/i915 developers can then reassign to the right component if it's not a kernel issue.
[sáb jun 27 01:59:11 2020] The GPU crash dump is required to analyze GPU hangs, so please always attach it.
[sáb jun 27 01:59:11 2020] GPU crash dump saved to /sys/class/drm/card0/error
[sáb jun 27 01:59:11 2020] i915 0000:00:02.0: Resetting chip for stopped heartbeat on rcs0
[sáb jun 27 01:59:11 2020] i915 0000:00:02.0: HWR[27910] context reset due to GPU hang
dmesg.txt and /sys/class/drm/card0/error error.txt
Then another crash:
[sáb jun 27 22:44:50 2020] i915 0000:00:02.0: GPU HANG: ecode 6:1:85fffffd, in HWR [67839]
[sáb jun 27 22:44:50 2020] GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace.
[sáb jun 27 22:44:50 2020] Please file a _new_ bug report at https://gitlab.freedesktop.org/drm/intel/issues/new.
[sáb jun 27 22:44:50 2020] Please see https://gitlab.freedesktop.org/drm/intel/-/wikis/How-to-file-i915-bugs for details.
[sáb jun 27 22:44:50 2020] drm/i915 developers can then reassign to the right component if it's not a kernel issue.
[sáb jun 27 22:44:50 2020] The GPU crash dump is required to analyze GPU hangs, so please always attach it.
[sáb jun 27 22:44:50 2020] GPU crash dump saved to /sys/class/drm/card0/error
[sáb jun 27 22:44:50 2020] i915 0000:00:02.0: Resetting chip for stopped heartbeat on rcs0
[sáb jun 27 22:44:50 2020] i915 0000:00:02.0: HWR[67839] context reset due to GPU hang
[dom jun 28 22:03:55 2020] i915 0000:00:02.0: GPU HANG: ecode 6:1:85fffffd, in HWR [157553]
[dom jun 28 22:03:55 2020] i915 0000:00:02.0: Resetting chip for stopped heartbeat on rcs0
[dom jun 28 22:03:55 2020] i915 0000:00:02.0: HWR[157553] context reset due to GPU hang
[dom jun 28 22:04:04 2020] i915 0000:00:02.0: GPU HANG: ecode 6:1:95feffb8, in HWR [157553]
[dom jun 28 22:04:04 2020] i915 0000:00:02.0: Resetting chip for stopped heartbeat on rcs0
[dom jun 28 22:04:04 2020] i915 0000:00:02.0: HWR[157553] context reset due to GPU hang
[dom jun 28 22:04:07 2020] i915 0000:00:02.0: GPU HANG: ecode 6:1:94eeffbc, in HWR [157553]
[dom jun 28 22:04:07 2020] i915 0000:00:02.0: Resetting chip for stopped heartbeat on rcs0
[dom jun 28 22:04:07 2020] i915 0000:00:02.0: HWR[157553] context reset due to GPU hang
[dom jun 28 22:04:07 2020] HWR[157529]: segfault at 80 ip 00000000004436ec sp 00007ffd9a4db570 error 6 in HWR[400000+59b000]
[dom jun 28 22:04:07 2020] Code: 48 83 c4 08 c3 90 90 90 90 90 90 90 90 90 53 89 f3 48 83 ec 10 8b 07 bf 80 4f 41 01 48 8d 74 24 0c 89 44 24 0c e8 a4 ff ff ff <01> 98 80 00 00 00 c7 80 e0 00 00 00 00 00 00 00 48 83 c4 10 5b c3
[dom jun 28 22:04:53 2020] i915 0000:00:02.0: GPU HANG: ecode 6:1:97eeffbc, in Renderer [4944]
[dom jun 28 22:04:53 2020] i915 0000:00:02.0: Resetting chip for stopped heartbeat on rcs0
[dom jun 28 22:04:53 2020] i915 0000:00:02.0: Renderer[4944] context reset due to GPU hang
With the dmesg2.txt and /sys/class/drm/card0/error error2.txt
And now, playing Team Fortress 2:
[ter jun 30 18:09:41 2020] i915 0000:00:02.0: GPU HANG: ecode 6:1:f3ff7ffc, in hl2_linux [203723]
[ter jun 30 18:09:41 2020] GPU hangs can indicate a bug anywhere in the entire gfx stack, including userspace.
[ter jun 30 18:09:41 2020] Please file a _new_ bug report at https://gitlab.freedesktop.org/drm/intel/issues/new.
[ter jun 30 18:09:41 2020] Please see https://gitlab.freedesktop.org/drm/intel/-/wikis/How-to-file-i915-bugs for details.
[ter jun 30 18:09:41 2020] drm/i915 developers can then reassign to the right component if it's not a kernel issue.
[ter jun 30 18:09:41 2020] The GPU crash dump is required to analyze GPU hangs, so please always attach it.
[ter jun 30 18:09:41 2020] GPU crash dump saved to /sys/class/drm/card0/error
[ter jun 30 18:09:41 2020] i915 0000:00:02.0: Resetting chip for stopped heartbeat on rcs0
[ter jun 30 18:09:41 2020] i915 0000:00:02.0: hl2_linux[203723] context reset due to GPU hang
dmesg3.txt and error3.txt
System information:
$ uname -m
x86_64
$ uname -r
5.7.0-1-amd64
It's a Debian unstable system, up-to-date. Do you need any more info, please?
Edited by Chris Wilson