| Summary: | kwin_wayland crash when screen goes blank for a while with Plasma 6.0.90 | ||
|---|---|---|---|
| Product: | [Plasma] kwin | Reporter: | ollilein |
| Component: | wayland-generic | Assignee: | ollilein |
| Status: | RESOLVED FIXED | ||
| Severity: | crash | CC: | agurenko, akselmo, julien.dlq, kde, kostadinshishmanov, postix |
| Priority: | NOR | ||
| Version First Reported In: | git-stable-Plasma/6.1 | ||
| Target Milestone: | --- | ||
| Platform: | Gentoo Packages | ||
| OS: | Linux | ||
| Latest Commit: | https://invent.kde.org/plasma/kwin/-/commit/ea83fc0052d9622e8ebe96a7de884ec486fab382 | Version Fixed In: | |
| Sentry Crash Report: | |||
| Attachments: |
Backtrace of plasmashell crashing
Backtrace from crashing kwin_wayland Backtrace from kwin-6.0.90.1-r1 kwin_wayland-6.0.90.1-r1-new kwin backtrace Patch for kwin-6.0.90.1 Patch for kde-plasma/kwin-6.0.90.1-r2 |
||
|
Description
ollilein
2024-05-28 17:21:25 UTC
Hi, can you share a backtrace for the crash with debug symbols? Thanks! https://wiki.gentoo.org/wiki/Project:Quality_Assurance/Backtraces Created attachment 169959 [details]
Backtrace of plasmashell crashing
Today i recompiled my system with cflags "-march=znver3 -O2 -mtune=znver3 -pipe -g -ggdb" and FEATURES="parallel-install parallel-fetch fixlafiles nostrip" so that everything is compiled with debug symbols.
I startet gdb with "gdb -q /usr/bin/plasmashell"
But in the logfile there comes only the output "no stack".
These are the last lines in gdb before plasmashell crashes:
Thread 0x7fff956006c0 (LWP 2693) exited]
[Thread 0x7fff8b4006c0 (LWP 2692) exited]
[Thread 0x7fff72a006c0 (LWP 2591) exited]org.kde.pulseaudio: No object for name "alsa_output.pci-0000_09_00.1.hdmi-stereo"
[New Thread 0x7fffa2a006c0 (LWP 2685)]
[New Thread 0x7fff8b4006c0 (LWP 2686)]
[New Thread 0x7fff956006c0 (LWP 2687)]
[Thread 0x7fff956006c0 (LWP 2687) exited]
[Thread 0x7fff8b4006c0 (LWP 2686) exited]
[Thread 0x7fffa2a006c0 (LWP 2685) exited]
[New Thread 0x7fffa2a006c0 (LWP 2691)]
[New Thread 0x7fff8b4006c0 (LWP 2692)]
[New Thread 0x7fff956006c0 (LWP 2693)]
org.kde.pulseaudio: No object for name "alsa_output.pci-0000_09_00.1.hdmi-stereo"
The Wayland connection broke. Did the Wayland compositor die?
[Thread 0x7fff734006c0 (LWP 2590) exited]
[Thread 0x7fff73e006c0 (LWP 2589) exited]
[Thread 0x7fff7e0006c0 (LWP 2588) exited]
[Thread 0x7fff7ea006c0 (LWP 2587) exited]
[Thread 0x7fff7f4006c0 (LWP 2586) exited]
[Thread 0x7fff70a006c0 (LWP 2585) exited]
[Thread 0x7fff714006c0 (LWP 2576) exited]
[Thread 0x7fff71e006c0 (LWP 2572) exited]
[Thread 0x7fff7d0006c0 (LWP 2571) exited]
[Thread 0x7fffc56006c0 (LWP 2532) exited]
[Thread 0x7fffc60006c0 (LWP 2531) exited]
[Thread 0x7fffc6a006c0 (LWP 2530) exited]
[Thread 0x7fffc74006c0 (LWP 2528) exited]
[Thread 0x7fffcf4006c0 (LWP 2525) exited]
[Thread 0x7fffcfe006c0 (LWP 2524) exited]
[Thread 0x7fffe4c006c0 (LWP 2523) exited]
[Thread 0x7fffd7e006c0 (LWP 2519) exited]
[Thread 0x7fffdd2006c0 (LWP 2505) exited]
[Thread 0x7fffe74006c0 (LWP 2491) exited]
[Thread 0x7fffe7e006c0 (LWP 2490) exited]
[Thread 0x7fffed6006c0 (LWP 2489) exited]
[Thread 0x7ffff0373a00 (LWP 2486) exited]
[Thread 0x7fffa2a006c0 (LWP 2691) exited]
[New process 2486]
[Inferior 1 (process 2486) exited with code 0377]
I hope it helps.
I forgot something, dmesg says: [ 172.436605] kwin_wayland[1197]: segfault at 58 ip 00007f2619081ab4 sp 00007fff52b0c408 error 4 in libkwin.so.6.0.90[7f2618f4d000+55b000] likely on CPU 31 (core 15, socket 0) [ 172.436614] Code: b2 7c ef ff 66 90 5b c3 66 66 2e 0f 1f 84 00 00 00 00 00 0f 1f 00 f3 0f 1e fa c6 47 5c 01 c3 90 66 0f 1f 44 00 00 f3 0f 1e fa <8b> 47 58 ff 4f 54 85 c0 75 06 80 7f 5c 00 75 0c c3 66 66 2e 0f 1f Created attachment 169960 [details]
Backtrace from crashing kwin_wayland
Since i found out that "kwin_wayland" is crashing i made a backtrace with "gdb -q /usr/bin/kwin_wayland".
I think that the Problem is "kwin_wayland"
Did host or nested kwin_wayland crash? How do you run plasma? (In reply to Vlad Zahorodnii from comment #5) > Did host or nested kwin_wayland crash? How do you run plasma? Both are crashing, i had running plasma_wayland in gdb and normal in the plasmashell. Both crashed at the same time, but i could only get a backtrace for the one which was startet by gdb. I start plasma through my .bashrc with the following command: if [ "$(tty)" = "/dev/tty1" ] then exec /usr/bin/dbus-run-session /usr/bin/startplasma-wayland fi and then use systemd autologin for tty1. I do this because sddm is making trouble on my PC. Created attachment 170065 [details]
Backtrace from kwin-6.0.90.1-r1
I updated kwin from 6.0.90 to 6.0.90.1-r1 (gentoo version) and it keeps crashing sometimes.
>I start plasma through my .bashrc with the following command:
Not using a display manager is not supported. There is a lot the display manager does that you're not doing.
It's also the cause of your issue, kwin is trying to start nested inside another kwin that isn't running because.
If any issue persists with LightDM/GDM/SDDM please reopen this bug
I now started with sddm, and the bug differs a bit, kwin_wayland keeps crashing sometimes but plasmashell is now restartet automatically, which i had to do manually before. Created attachment 170080 [details]
kwin_wayland-6.0.90.1-r1-new
I should say that i start kwin_wayland through the konsole as a second process with gdb, so there is another instance running from "native" plasmashell.
So it keeps crashing.
Created attachment 170249 [details]
kwin backtrace
I am also running into this I believe (using sddm). Everything on my system is built with -ggdb3, so the backtrace might be more useful than just -ggdb.
A possibly relevant merge request was started @ https://invent.kde.org/plasma/kwin/-/merge_requests/5865 Git commit 0c5ee4789215eefd838a6b97783a05ac6cd01433 by Xaver Hugl. Committed on 10/06/2024 at 11:46. Pushed by zamundaaa into branch 'master'. core/renderbackend: check for the renderloop being deleted Output frames can outlive the output they were created for, so the render loop might also be deleted by the time the output frame is destroyed or presented M +4 -2 src/core/renderbackend.cpp M +1 -1 src/core/renderbackend.h https://invent.kde.org/plasma/kwin/-/commit/0c5ee4789215eefd838a6b97783a05ac6cd01433 Git commit ea83fc0052d9622e8ebe96a7de884ec486fab382 by Xaver Hugl. Committed on 10/06/2024 at 12:01. Pushed by zamundaaa into branch 'Plasma/6.1'. core/renderbackend: check for the renderloop being deleted Output frames can outlive the output they were created for, so the render loop might also be deleted by the time the output frame is destroyed or presented (cherry picked from commit 0c5ee4789215eefd838a6b97783a05ac6cd01433) Co-authored-by: Xaver Hugl <xaver.hugl@gmail.com> M +4 -2 src/core/renderbackend.cpp M +1 -1 src/core/renderbackend.h https://invent.kde.org/plasma/kwin/-/commit/ea83fc0052d9622e8ebe96a7de884ec486fab382 Created attachment 170340 [details] Patch for kwin-6.0.90.1 I made a patch from https://invent.kde.org/plasma/kwin/-/commit/0c5ee4789215eefd838a6b97783a05ac6cd01433 for the "native" Version of kwin 6.0.90.1 Created attachment 170341 [details]
Patch for kde-plasma/kwin-6.0.90.1-r2
I made a patch for the gentoo patched Version kde-plasma/kwin-6.0.90.1-r2
I tried the patch for the gentoo patched kwin-6.0.90.1-r2, and it look like there are no more crashes. I tested for 40 minutes with every minute screen blank and kwin hasn't crashed dureing this test. So i would say it is fixed with this patch, but i cannot say for sure for the "native" version 6.0.90.1 because i can't test it at the moment, but i think it should also be crash free. |