Bug 111986 - 5700 XT hangs entire system in games
Summary: 5700 XT hangs entire system in games
Status: NEW
Alias: None
Product: DRI
Classification: Unclassified
Component: DRM/AMDgpu (show other bugs)
Version: unspecified
Hardware: x86-64 (AMD64) Linux (All)
: high critical
Assignee: Default DRI bug account
QA Contact:
URL:
Whiteboard:
Keywords:
Depends on:
Blocks:
 
Reported: 2019-10-13 02:29 UTC by Daniel Suarez
Modified: 2019-10-22 17:25 UTC (History)
1 user (show)

See Also:
i915 platform:
i915 features:


Attachments
Apitrace (85.55 KB, text/plain)
2019-10-13 04:06 UTC, Daniel Suarez
no flags Details
Apitrace GFXBench (5.38 KB, text/plain)
2019-10-13 04:07 UTC, Daniel Suarez
no flags Details
Apitrace The Witcher 2 (36.14 KB, text/plain)
2019-10-13 04:07 UTC, Daniel Suarez
no flags Details
Xorg log (74.02 KB, text/plain)
2019-10-13 17:50 UTC, Daniel Suarez
no flags Details
Xorg log 2 (81.90 KB, text/plain)
2019-10-14 17:54 UTC, Daniel Suarez
no flags Details
Got by running sudo journalctl -b > dmesg.log (389.55 KB, text/x-log)
2019-10-14 19:27 UTC, Daniel Suarez
no flags Details
Correct journalctl (374.75 KB, text/x-log)
2019-10-15 00:24 UTC, Daniel Suarez
no flags Details

Note You need to log in before you can comment on or make changes to this bug.
Description Daniel Suarez 2019-10-13 02:29:01 UTC
Hello there!

I just installed my 5700 XT in my system, running Manjaro in testing branch with Mesa 19.2.1, 5.3.6-1-MANJARO and LLVM 9.0.0.

Issue has appeared in native game The Witcher 1 and Space Engineers, in both these instances a few moments (in The Witcher 1 it once took a few minutes) the video freezes, while audio continues in the background. I cannot alt-tab or enter TTY, I am forced to force shutdown. 

I have not been able to reproduce the hang issue while playing YouTube or other normal tasks. 

I am not sure how to provide logs since this hangs my entire system and I am forced to shutdown, which journalctl filter do I use to provide useful information? 

Thank you, and sorry if I am not providing anything useful, it is my first time submitting a bug here.
Comment 1 Daniel Suarez 2019-10-13 02:46:29 UTC
(In reply to Daniel Suarez from comment #0)


I was able to reproduce the issue with gfxbench, it hanged during the tessellation test at the beginning, I was still able to hear a YouTube video I had playing in the background.
Comment 2 Daniel Suarez 2019-10-13 04:06:46 UTC
Created attachment 145720 [details]
Apitrace

I think I was able to get an apitrace by doing 

DEBUGGER="apitrace trace" steam steam://rungameid/20920 > ApiTrace.txt 

For The Witcher 2 and Space Engineers respectably and then 

DEBUGGER="apitrace trace" gfxbench 

For Gfxbench, I include the results of these since I am able to recover them since they save after I am forced to shutdown my PC
Comment 3 Daniel Suarez 2019-10-13 04:07:22 UTC
Created attachment 145721 [details]
Apitrace GFXBench
Comment 4 Daniel Suarez 2019-10-13 04:07:37 UTC
Created attachment 145722 [details]
Apitrace The Witcher 2
Comment 5 Daniel Suarez 2019-10-13 04:29:49 UTC
Update: The same issue just occurred with just Firefox being opened

Apologies for so many comments, if any additional information is needed please do not hesitate to ask
Comment 6 Daniel Suarez 2019-10-13 17:50:01 UTC
Created attachment 145728 [details]
Xorg log

I switched over to mesa-git (along with all the other packages recommended to accompany it mentioned in the Arch wiki) and the same exact issue occurs, this is the log found in "/var/log/Xorg.0.log.old", grabbed it right after booting in after having been forced to shutdown when my PC froze when running GFXBench
Comment 7 Pierre-Eric Pelloux-Prayer 2019-10-14 09:54:31 UTC
Thanks for your bug report.

It would be really helpful to have the content of dmesg when your system hangs.
You can get it from another computer using ssh or you could extract it from /var/log/dmesg* files after a reboot.

You can also check some of the existing bugs and see if they are the same as yours:
- https://bugs.freedesktop.org/show_bug.cgi?id=111986
- https://bugzilla.kernel.org/show_bug.cgi?id=205169
Comment 8 Daniel Suarez 2019-10-14 17:54:15 UTC
(In reply to Pierre-Eric Pelloux-Prayer from comment #7)
> Thanks for your bug report.
> 
> It would be really helpful to have the content of dmesg when your system
> hangs.
> You can get it from another computer using ssh or you could extract it from
> /var/log/dmesg* files after a reboot.
> 
> You can also check some of the existing bugs and see if they are the same as
> yours:
> - https://bugs.freedesktop.org/show_bug.cgi?id=111986
> - https://bugzilla.kernel.org/show_bug.cgi?id=205169

Thank you so much for taking the time to reply to my bug report, I really wanted to provide that but unfortunately it doesn't exist, I even did "ls -ls /var/log/" to make sure it wasn't dolphin hiding the file, but a dmesg file/folder doesn't exist. Closest thing I see is "Xorg.0.log.old" which I will attach just incase 

As for bug report 205169, I do not believe so since my system hangs regardless if a game or application is fullscreen or not.
Comment 9 Daniel Suarez 2019-10-14 17:54:48 UTC
Created attachment 145735 [details]
Xorg log 2
Comment 10 Daniel Suarez 2019-10-14 19:27:57 UTC
Created attachment 145736 [details]
Got by running sudo journalctl -b > dmesg.log

Thanks to the explanation by someone I believe I understand now what you meant. I went ahead and reproduced the hang by running gfxbench and when the system hanged at around 3:23 I force shutdown seconds later, once I booted in I ran "sudo journalctl -b > dmesg.log" and I am sending you the output now
Comment 11 Pierre-Eric Pelloux-Prayer 2019-10-14 20:17:36 UTC
(In reply to Daniel Suarez from comment #10)
> Created attachment 145736 [details]
> Got by running sudo journalctl -b > dmesg.log
> 
> Thanks to the explanation by someone I believe I understand now what you
> meant. I went ahead and reproduced the hang by running gfxbench and when the
> system hanged at around 3:23 I force shutdown seconds later, once I booted
> in I ran "sudo journalctl -b > dmesg.log" and I am sending you the output now

You need to use "sudo journalctl -b -1 > dmesg.log" to get the logs from the previous boot (= the one that hanged).
Comment 12 Daniel Suarez 2019-10-15 00:24:13 UTC
(In reply to Pierre-Eric Pelloux-Prayer from comment #11)
> (In reply to Daniel Suarez from comment #10)
> > Created attachment 145736 [details]
> > Got by running sudo journalctl -b > dmesg.log
> > 
> > Thanks to the explanation by someone I believe I understand now what you
> > meant. I went ahead and reproduced the hang by running gfxbench and when the
> > system hanged at around 3:23 I force shutdown seconds later, once I booted
> > in I ran "sudo journalctl -b > dmesg.log" and I am sending you the output now
> 
> You need to use "sudo journalctl -b -1 > dmesg.log" to get the logs from the
> previous boot (= the one that hanged).

I am deeply sorry, I attach the correct one now
Comment 13 Daniel Suarez 2019-10-15 00:24:36 UTC
Created attachment 145744 [details]
Correct journalctl
Comment 14 Pierre-Eric Pelloux-Prayer 2019-10-15 17:20:34 UTC
The last lines of your log indicates a gfx hang.

You can try to set the environment variable AMD_DEBUG to the "nongg" value to see if it helps.

The apitrace you attached are the log output from apitrace, not the real trace.
The trace should be next to the binary of the game (try to search for .trace file in $HOME/.steam/steam/steamapps/common).
Comment 15 Sabbie 2019-10-22 16:12:10 UTC
I'm having the same issue on a 5700 (non-xt).

It seems to be this bug:

https://bugs.freedesktop.org/show_bug.cgi?id=111481#c9
Comment 16 Daniel Suarez 2019-10-22 17:25:41 UTC
(In reply to Sabbie from comment #15)
> I'm having the same issue on a 5700 (non-xt).
> 
> It seems to be this bug:
> 
> https://bugs.freedesktop.org/show_bug.cgi?id=111481#c9

Yeah, apologies I have not provided the logs asked I've had a few hard days due to college entrance shit, but it's the exact same issue as the one listed in the bug report you linked, this is completely unacceptable from Amd honestly


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.