qubes-issues
qubes-issues copied to clipboard
KDE plasma desktop stopped working after upgrading linux kernel to 5.19.9
Qubes OS release
qubes: |
R4.1
xen: |
4.14.5
...
cpu: |
12th Gen Intel(R) Core(TM) i7-12800H
Brief summary
After enabling qubes-dom0-current-testing
repo and downloading kernel-latest
package version 5.19.9
, KDE plasma doesn't work.
Steps to reproduce
- Install
kernel-latest
package version5.18.16
. - Install KDE plasma. Ensure it works with the kernel version mentioned above.
- Enable
qubes-dom0-current-testing
repo - Download
kernel-latest
package version5.19.9
from that repo. - Reboot to the new kernel try to load KDE plasma.
Expected behavior
KDE plasma visually works the same way as with kernel version 5.18.16
.
Actual behavior
KDE plasma is completely non-functional. It is observed as a blank screen with only certain things visible: mouse cursor, network connection notifications, windows of applications that auto-start. After 1-2 minutes the following warning appears:
Below some logs from journalctl
Sep 18 19:30:12 dom0 systemd-coredump[11449]: Process 11071 (kwin_x11) of user 1000 dumped core.
Stack trace of thread 11443:
#0 0x000074ea548c67d5 raise (libc.so.6 + 0x3c7d5)
#1 0x000074ea548c6860 __restore_rt (libc.so.6 + 0x3c860)
#2 0x000074ea44027029 n/a (n/a + 0x0)
#3 0x000074ea30c19334 llvm_pipeline_generic.constprop.0 (swrast_dri.so + 0x619334)
#4 0x000074ea30c19a81 llvm_middle_end_run (swrast_dri.so + 0x619a81)
#5 0x000074ea30bd64c2 vsplit_run_ushort (swrast_dri.so + 0x5d64c2)
#6 0x000074ea30bce2c7 draw_vbo (swrast_dri.so + 0x5ce2c7)
#7 0x000074ea30c92c43 llvmpipe_draw_vbo (swrast_dri.so + 0x692c43)
#8 0x000074ea3073f468 st_draw_vbo (swrast_dri.so + 0x13f468)
#9 0x000074ea30984929 _mesa_validated_drawrangeelements (swrast_dri.so + 0x384929)
#10 0x000074ea30984c7c _mesa_DrawElements (swrast_dri.so + 0x384c7c)
#11 0x000074ea521b07ab _ZN16QSGBatchRenderer8Renderer17renderMergedBatchEPKNS_5BatchE (libQt5Quick.so.5 + 0x1b07ab)
#12 0x000074ea521b51ed _ZN16QSGBatchRenderer8Renderer13renderBatchesEv (libQt5Quick.so.5 + 0x1b51ed)
#13 0x000074ea521b5ac2 _ZN16QSGBatchRenderer8Renderer6renderEv (libQt5Quick.so.5 + 0x1b5ac2)
#14 0x000074ea5219e2a4 _ZN11QSGRenderer11renderSceneERK11QSGBindable (libQt5Quick.so.5 + 0x19e2a4)
#15 0x000074ea5219e773 _ZN11QSGRenderer11renderSceneEj (libQt5Quick.so.5 + 0x19e773)
#16 0x000074ea521fe985 _ZN23QSGDefaultRenderContext15renderNextFrameEP11QSGRendererj (libQt5Quick.so.5 + 0x1fe985)
#17 0x000074ea52267109 _ZN19QQuickWindowPrivate16renderSceneGraphERK5QSizeS2_ (libQt5Quick.so.5 + 0x267109)
#18 0x000074ea5220c22b _ZN15QSGRenderThread13syncAndRenderEP6QImage (libQt5Quick.so.5 + 0x20c22b)
#19 0x000074ea5220d0c7 _ZN15QSGRenderThread3runEv (libQt5Quick.so.5 + 0x20d0c7)
#20 0x000074ea530c6680 _ZN14QThreadPrivate5startEPv (libQt5Core.so.5 + 0xc6680)
#21 0x000074ea52883432 start_thread (libpthread.so.0 + 0x9432)
#22 0x000074ea5498b6d3 __clone (libc.so.6 + 0x1016d3)
Stack trace of thread 11094:
#0 0x000074ea5498086f __poll (libc.so.6 + 0xf686f)
#1 0x000074ea542bb38a _xcb_conn_wait (libxcb.so.1 + 0xd38a)
#2 0x000074ea542bd71a xcb_wait_for_event (libxcb.so.1 + 0xf71a)
#3 0x000074ea4d534c68 _ZN14QXcbEventQueue3runEv (libQt5XcbQpa.so.5 + 0x64c68)
#4 0x000074ea530c6680 _ZN14QThreadPrivate5startEPv (libQt5Core.so.5 + 0xc6680)
#5 0x000074ea52883432 start_thread (libpthread.so.0 + 0x9432)
#6 0x000074ea5498b6d3 __clone (libc.so.6 + 0x1016d3)
Stack trace of thread 11142:
#0 0x000074ea5288a1b8 pthread_cond_timedwait@@GLIBC_2.3.2 (libpthread.so.0 + 0x101b8)
#1 0x000074ea530cbcfa _ZN14QWaitCondition4waitEP6QMutex14QDeadlineTimer (libQt5Core.so.5 + 0xcbcfa)
#2 0x000074ea530cbdfb _ZN14QWaitCondition4waitEP6QMutexm (libQt5Core.so.5 + 0xcbdfb)
#3 0x000074ea530c9f35 _ZN17QThreadPoolThread3runEv (libQt5Core.so.5 + 0xc9f35)
#4 0x000074ea530c6680 _ZN14QThreadPrivate5startEPv (libQt5Core.so.5 + 0xc6680)
#5 0x000074ea52883432 start_thread (libpthread.so.0 + 0x9432)
#6 0x000074ea5498b6d3 __clone (libc.so.6 + 0x1016d3)
Xfce seems to work as before the kernel upgrade.
I have a similar issue. But my issue only exists when there's Option "AccelMethod" "none"
in xorg.conf.d
. Maybe you look at /etc/X11/xorg.conf.d/
to see if there's special configurations?
Right, I had to make a customization there like suggested by a comment from another issue in order to solve graphical artifacts problem. Sounds like the file requires further tweaking.
Tried "none", "sna", "uxa"
values for Option "AccelMethod"
, all led to the same behavior.
I don't have any customisation and KDE works as usual with kernel-latest.
Tried
"none", "sna", "uxa"
values forOption "AccelMethod"
, all led to the same behavior.
It becomes stable too, if I delete that user-made .conf file. However you need that file to avoid severe graphical glitches, right? So you need a different way to address the issue.
I don't have any customisation and KDE works as usual with kernel-latest.
@unman do you have Alder Lake CPU?
It becomes stable too, if I delete that user-made .conf file. However you need that file to avoid severe graphical glitches, right?
@SurFlurer for me even without that file KDE is completely broken on 5.19.9 kernel, so that graphical glitches problem becomes irrelevant.
Today my Qubes installation received a dom0 update, mostly xen
-related packages. Now Qubes behaves as expected even with the latest 5.19.9
kernel.
That's https://github.com/QubesOS/qubes-vmm-xen/commit/046904b most likely. Good to hear it fixed this issue too.