I noticed a really nasty bug in the graphics drivers lately.
While playing games based on the source engine, I get GPU hangs.
Here is the relevant info.
OpenGL: Mesa DRI Intel(R) HD Graphics 520 (Skylake GT2) x86/MMX/SSE2 3.0 Mesa 17.3.0-devel (3.0.0)
Linux walaryne-HP-Notebook 4.10.0-27-generic #30~16.04.2-Ubuntu SMP (It says Ubuntu but it's Mint.)
Thu Jun 29 16:07:46 UTC 2017 x86_64 x86_64 x86_64 GNU/Linux
Relevant error from Counter Strike: Source:
intel_do_flush_locked failed: Input/output error
DMESG Relevant error(s):
[ 445.882919] [drm] GPU HANG: ecode 9:0:0x85dffffb, in hl2_linux , reason: Hang on render ring, action: reset
[ 445.882962] drm/i915: Resetting chip after gpu hang
[ 445.883020] [drm] RC6 off
[ 445.897614] [drm] GuC firmware load skipped
[ 457.834654] drm/i915: Resetting chip after gpu hang
[ 457.834740] [drm] RC6 off
[ 457.848586] [drm] GuC firmware load skipped
[ 458.024493] hl2_linux: segfault at 0 ip (null) sp 00000000ffd2453c error 14 in hl2_linux[8048000+1000]
I sincerely hope this helps fix the problem.
I did attempt to run the game with RC6 enabled and disabled,
including frame buffer compression, neither made any effect.
Steps to reproduce:
1. Launch a Source based game (CS:S, CS:GO, TF2, etc.)
2. Play as normal.
3. Crash seems to occur 5 to 10 minutes through gameplay.
This is my first time submitting a bug report, I hope it looks right.
I've tried to reproduce this on Kabylake GT2 system using 4.13.0-rc4+ kernel and Mesa 17.3.0-devel and could not reproduce the bug. I played CS:GO for quite a long time, not sure if visual settings matter (?) FWIW I do have the guc & dmc firmwares installed.
Okay so, to clarify. Someone changed i915 to i965 which my system is not using, second, it's a core i3.
(In reply to hazardoustheradioactive from comment #2)
> Okay so, to clarify. Someone changed i915 to i965 which my system is not
> using, second, it's a core i3.
Okay so I looked it up and saw i965 is mesa embedded, disregard the i915 thing.
(In reply to Tapani Pälli from comment #1)
> I've tried to reproduce this on Kabylake GT2 system using 4.13.0-rc4+ kernel
> and Mesa 17.3.0-devel and could not reproduce the bug. I played CS:GO for
> quite a long time, not sure if visual settings matter (?) FWIW I do have the
> guc & dmc firmwares installed.
Thanks, but I believe this is Skylake specific.
Created attachment 134153 [details]
GPU error dump
First 832 lines of /sys/class/drm/card0/error
I have the same issue with my Intel(R) Core(TM) i7-6600U CPU (GPU: Skylake GT2). See the attached GPU error dump.
Hello, this is most likely fixed in Mesa master by:
Author: Jason Ekstrand <firstname.lastname@example.org>
Date: Wed Nov 29 16:22:42 2017 -0800
i965: Disable regular fast-clears (CCS_D) on gen9+
This partially reverts commit 3e57e9494c2279580ad6a83ab8c065d01e7e634e
which caused a bunch of GPU hangs on several Source titles. To date, we
have no clue why these hangs are actually happening. This undoes the
final effect of 3e57e9494c227 and gets us back to not hanging. Tested
with Team Fortress 2.
Reviewed-by: Kenneth Graunke <email@example.com>
It should be fixed in Mesa 17.3.0 as well (once that's released).
Please reopen if you still experience hangs. We've confirmed the fix on Counter Strike: Global Offensive, Left 4 Dead 2, Team Fortress 2, and Portal 2. We didn't explicitly test Counter Strike: Source, but it ought to be fixed as well.