Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

MSAA related crashes with Intel Kabylake gpu crashes on Ubuntu 17.10 #1509

Closed
nightsky30 opened this issue Aug 26, 2017 · 14 comments
Closed

Comments

@nightsky30
Copy link

nightsky30 commented Aug 26, 2017

Your system information

1ab13c40031b146ffa7ec30a15c9f847-d0fb9a358abf936a446ebe45804aafb3de9bee68.zip

  • Have you checked for system updates?: Yes

Please describe your issue in as much detail as possible:

I recently updated from Xubuntu 17.04 to Xubuntu 17.10 Artful Aardvark Dev Release. I did this knowing things might be buggy. I wanted to do some testing.

  • CSGO freezes and crashes when joining servers after the map load while the screen says "Retrieving Game Data..."
  • I can hear other players on their mics, and in-game sound effects before the crash
  • I have verified my game files multiple times
  • CSGO was playing fine without issue prior to the system update

Here are my crash .dmp files:

csgo-crash-dump.tar.gz

Steps for reproducing this issue:

  1. Run the current release of Ubuntu 17.10 and Steam
  2. Start CSGO
  3. Attempt to play a casual game by selecting a game type (Hostage)
  4. Freeze and Crash
@kisak-valve
Copy link
Member

Hello @nightsky30, unfortunately, my minidump reader could not tell where these crashes are occurring, but your description of this issue hints that it may be video / video driver related. Can you check dmesg for what library the game is crashing in?

@nightsky30
Copy link
Author

Hi Kisak,

Yes, I can check dmesg. I will say that I tweaked the video settings, and the game doesn't crash if I use the settings in this screenshot. I tested a few hours with no issues:

screenshot_2017-08-26_20-36-55

Which makes me think it is much the same issue as TF2 with Intel drivers?? It is funny that I hadn't modified the video settings prior to it starting to crash. I only did the distribution upgrade to 17.10. But the lowering of screen res, texture detail, anti-alising, and switching to single core seems to fix it. Now which one of those breaks if switched on?? :)

@nightsky30
Copy link
Author

nightsky30 commented Aug 27, 2017

I enabled Multicore Rendering and played a long while before this crash. Here's the latest with dmesg and .dmp:

assert_20170827102405_1.tar.gz

Also the GPU Crash Dump:

assert_20170827102405_1_GPU_Crash_Dump.txt

@kisak-valve
Copy link
Member

Thanks @nightsky30, CS:GO is tripping over itself after a gpu reset. You should also bring this issue to the attention of your driver vendor if it has not been already.

@nightsky30
Copy link
Author

Ok, thanks Kisak.

Submitted here:
https://bugs.freedesktop.org/show_bug.cgi?id=102435

@nightsky30
Copy link
Author

I noted this in the linked bug, but:

FYI, it seems I've cornered the cause of the crashing. If I set Multisampling Anti-Aliasing Mode to None, every time I join a server and view the first in game [Continue] banner screen, the game will crash. If I set Multisampling Anti-Aliasing Mode to 2xMSAA, the game will load and play just fine.

@kisak-valve kisak-valve changed the title [Linux] Freeze and Crash upon joining server and retrieving game data MSAA related crashes with Intel Skylake gpu crashes on Ubuntu 17.10 Sep 1, 2017
@nightsky30 nightsky30 changed the title MSAA related crashes with Intel Skylake gpu crashes on Ubuntu 17.10 MSAA related crashes with Intel Kabylake gpu crashes on Ubuntu 17.10 Sep 1, 2017
@kerrickstaley
Copy link

I created another bug report on this issue. It has more crash reports if that helps: #1529

Confirming what others have said: this issue is only present with Mesa 17.2.0 with MSAA disabled in the game settings. The issue is not present with Mesa 17.1.8 or with MSAA set to "2x MSAA" in the game settings.

@kisak-valve
Copy link
Member

This issue bisects to:

commit 3e57e9494c2279580ad6a83ab8c065d01e7e634e
Author: Jason Ekstrand <jason.ekstrand@intel.com>
Date:   Wed Jun 21 21:35:07 2017 -0700

    i965: Enable regular fast-clears (CCS_D) on gen9+
    
    The set of formats which supports CCS_E is actually fairly small on
    gen9.  However, everything that supports fast-clears on gen8 also
    supports fast-clears on gen9+.  The one very annoying exception is
    that blending is broken for non-0/1 clear colors with sRGB formats.
    In order to solve that problem, we do a resolve to get rid of the
    clear color.  Another option would be to just not fast-clear with
    non-0/1 clear colors however non-0/1 + blending + sRGB is uncommon
    enough that this shouldn't be a significant performance problem.
    
    This appears to help gl_manhattan31_off by about 2%.
    
    Reviewed-by: Topi Pohjolainen <topi.pohjolainen@intel.com>

Assigning an Intel dev.

@Cheaterman
Copy link

Confirmed, using Mesa 17.1.x for the time being seems to be a valid workaround. Thanks for everything @kisak-valve !

@tpalli
Copy link

tpalli commented Nov 2, 2017

FWIW I've confirmed the bisected commit on KBL GT2 (with drm-tip kernel, 4.14.0-rc7+).

@jljusten
Copy link

jljusten commented Nov 3, 2017

It also affects Skylake. We have a bug open for it at https://bugs.freedesktop.org/show_bug.cgi?id=102435.

@ml-
Copy link

ml- commented Dec 2, 2017

Fixed (today) with this commit: https://cgit.freedesktop.org/mesa/mesa/commit/?id=ee57b15ec764736e2d5360beaef9fb2045ed0f68

@nightsky30
Copy link
Author

Tested MSAA set to None with the latest unstable padoka PPA which was built after the commit, and it appears to be fixed:

1:17.4git171201235300.d93fabba~padoka0
https://launchpad.net/~paulo-miguel-dias/+archive/ubuntu/mesa

@kisak-valve
Copy link
Member

This is now fixed upstream in 17.2.7 and 17.3.0.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

No branches or pull requests

7 participants