Bug 1220043

Summary: amdgpu DRM/radeonsi Mesa: X and Wayland die when using Prism Launcher, sometimes disabling user input entirely.
Product: [openSUSE] openSUSE Tumbleweed Reporter: roke beedell <rokejulianlockhart+1674683091>
Component: X.OrgAssignee: Gfx Bugs <gfx-bugs>
Status: RESOLVED NORESPONSE QA Contact: Gfx Bugs <gfx-bugs>
Severity: Major    
Priority: P3 - Medium    
Version: Current   
Target Milestone: ---   
Hardware: x86-64   
OS: openSUSE Tumbleweed   
URL: https://discussion.fedoraproject.org/t/random-indefinite-system-hang/110020#related-issues-3
See Also: https://github.com/PrismLauncher/PrismLauncher/issues/2139#issue-2141081956
https://bugzilla.redhat.com/show_bug.cgi?id=2293505
https://bugs.kde.org/show_bug.cgi?id=488989
Whiteboard:
Found By: Community User Services Priority:
Business Priority: Blocker: ---
Marketing QA Status: --- IT Deployment: ---
Attachments: Video demonstrating the issue.
Video demonstrating the issue.

Description roke beedell 2024-02-18 23:09:43 UTC 
Per https://forums.opensuse.org/t/where-to-report-xwayland-crash-the-cause-matters-right/172543/4, I've had two different types of results after a crash of X, albeit always caused when playing *Minecraft: Java Edition* via https://flathub.org/apps/org.prismlauncher.PrismLauncher, whereby Minecraft slowly becomes laggier, and eventually crashes:

1. The game became progressively slower, to the point where the game-reported 75 FPS felt like barely 30;

   2. Common:

         1. Firstly, the display freezes. Soon after, it becomes black. Then, the PC fans momentarily cease to operate, and the display becomes artefacted as the fans resume.
         2. Although the system responds (Alt+SysReq+B performs systemctl reboot) Alt+F3 doesn't switch from the display server to a TTY. journalctl --boot=-1 -e produces the log posted to https://github.com/PrismLauncher/PrismLauncher/issues/2139#issue-2141081956.

   2. Uncommon:

         1. The X server actually crashed (see the last line of the log at Forge 1.19.2 Server Log [#H1b3oft] - mclo.gs).
         2. Discord and Prism became 10% opacity.
         3. When I logged-out, I remained in TTY3 rather than TTY2, and switching manually to TTY2 demonstrated that the handover hadn’t occurred. I had to use TTY4 to systemctl reboot.

This doesn't occur on my brother's installation of Windows 11. In fact, I know of solely one *potential* report by anyone else of this.
Comment 1 Stefan Dirsch 2024-02-19 09:12:13 UTC 
I found a Xwayland backtrace with radeonsi Mesa driver involved and a lot of error messages of amdgpu kernel driver. Probably it's a combination or one of those not working correctly.
Comment 2 roke beedell 2024-02-20 18:16:35 UTC 
(In reply to Stefan Dirsch from comment #1)
> I found a Xwayland backtrace with radeonsi Mesa driver involved and a lot of
> error messages of amdgpu kernel driver. Probably it's a combination or one
> of those not working correctly.

Per https://bugs.kde.org/show_bug.cgi?id=481564#c4, perhaps it is?
Comment 3 Stefan Dirsch 2024-05-21 14:50:10 UTC 
Could this be retested with latest TW please?
Comment 4 roke beedell 2024-05-24 13:56:34 UTC 
(In reply to Stefan Dirsch from comment #3)
> Could this be retested with latest TW please?

I'm using https://download.fedoraproject.org/pub/fedora/linux/releases/40/Spins/x86_64/iso/Fedora-KDE-Live-x86_64-40-1.14.iso currently, so I can't - I don't have the space for second OS. Unless someone else who was able to reproduce this can do so, it'll have to wait until I next use OSTW on this HW configuration.
Comment 5 Stefan Dirsch 2024-05-24 14:09:22 UTC 
Ok. I can understand that you switched to another distribution with such issues. Feel free to reopen in case you try TW again and can still reproduce the issue. Thanks for you feedback! Closing as "NORESPONSE" ...
Comment 6 roke beedell 2024-05-24 14:34:03 UTC 
(In reply to Stefan Dirsch from comment #5)
> Ok. I can understand that you switched to another distribution with such
> issues. Feel free to reopen in case you try TW again and can still reproduce
> the issue. Thanks for you feedback! Closing as "NORESPONSE" ...

Thank you.
Comment 7 roke beedell 2024-06-20 21:34:05 UTC 
(In reply to Stefan Dirsch from comment #5)
> Ok. I can understand that you switched to another distribution with such
> issues. Feel free to reopen in case you try TW again and can still reproduce
> the issue. Thanks for you feedback! Closing as "NORESPONSE" ...

For posterity, I'd like to note that this continues in a vaguely lesser state with Fedora 40. There might be a few notifications on this issue as I edit it slightly so that I can collate this with other reports at disparate places to ascertain what the commonalities are that cause it to arise.
Comment 8 roke beedell 2024-06-20 22:39:00 UTC 
Created attachment 875622 [details]
Video demonstrating the issue.

https://imgur.com/YYUOD3q
Comment 9 roke beedell 2024-06-20 22:39:19 UTC 
Created attachment 875623 [details]
Video demonstrating the issue.

https://imgur.com/hLoDOqR