Bug 451976

Can't see any KDE code in the backtrace, but there may be something in #16 0x000055a4acc12dc9 in  (), which is currently missing.

Do you see an crashes for kwin_wayland too? Or only Plasma? The command-line `coredumpctl` utility can tell you.

kwin_wayland seems to work just fine. In fact, I get a Konsole window open w/o Plasma showing up. Starting /usr/bin/startplasma-wayland from Konsole causes the same crash to happen.

Additionally, I think this is related to NVIDIA drivers, because when I limited prime to intel-only (sudo prime-select intel), Plasma started just fine.

I am going to try to update NVIDIA drivers and try to produce more meaningful stack traces. And also will attach output of /usr/bin/startplasma-wayland

I wasn't able to find dbgsym packages for the requested files
* /usr/bin/plasmashell
* /lib/x86_64-linux-gnu/libQt5Core.so.5
* /lib/x86_64-linux-gnu/libQt5WaylandClient.so.5

However, I got core dumps from coredumpctl. At the startup two coredumps are created:

TIME                          PID  UID  GID SIG     COREFILE EXE                                    SIZE
Tue 2022-03-29 12:37:34 CEST 1685 1000 1000 SIGABRT present  /usr/bin/plasmashell                  14.0M
Tue 2022-03-29 12:37:37 CEST 1278 1000 1000 SIGABRT present  /usr/bin/ksplashqml                    2.5M

I am attaching core dump and info from the plasmashell one.  Also attaching output of running /usr/bin/startplasma-wayland.

The suspicious lines from running startplasma-wayland are:
qt.qpa.wayland: Wayland does not support QWindow::requestActivate()
[destroyed object]: error 7: importing the supplied dmabufs failed
The Wayland connection experienced a fatal error: Protocol error

Created attachment 147816 [details]
Plasmashell core info (from coredumpctl)

Created attachment 147817 [details]
Output of running /usr/bin/startplasma-wayland

Here is the core file: https://drive.google.com/file/d/1qRVisFEFY1Jxb8bJN7NT7b1IFuPA6EeX/view?usp=sharing

Can't attach it here directly, since it is too big (14M).

Just for the sake of reproduction, here are the versions of packages which provide the binaries:

/usr/bin/plasmashell: plasma-workspace: 4:5.24.3-0ubuntu1
/lib/x86_64-linux-gnu/libQt5Core.so.5: libqt5core5a: 5.15.2+dfsg-15
/lib/x86_64-linux-gnu/libQt5WaylandClient.so.5: libqt5waylandclient5: 5.15.2-4

We don't need the whole coredump file; we need yo to use that coredump file to get a text backtrace that has full debug symbols. Can you do that?

Unfortunately I am unable to find packages with debug symbols for the requested binaries (does Ubuntu still provide these?). Therefore I am unable to produce a better backtrace.

E.g. When I search for plasma-workspace (package providing /usr/bin/plasmashell), I get plasma-workspace-data [installed], plasma-workspace-dev (headers, no debug symbols), plasma-workspace-wallpapers and plasma-workspace-wayland [installed]. No *-dbg or *-dbgsym package.

I provided the core dump and package versions in case someone has access to debug symbol packages.

Please disregard my previous comment (just learned that one has to add ddebs repo to get -dbgsym packages). Here is the bettet stacktrace:

Application: Plasma (plasmashell), signal: Aborted

[KCrash Handler]
#4  __pthread_kill_implementation (no_tid=0, signo=6, threadid=140492689308096) at ./nptl/pthread_kill.c:44
#5  __pthread_kill_internal (signo=6, threadid=140492689308096) at ./nptl/pthread_kill.c:78
#6  __GI___pthread_kill (threadid=140492689308096, signo=signo@entry=6) at ./nptl/pthread_kill.c:89
#7  0x00007fc7050e1476 in __GI_raise (sig=sig@entry=6) at ../sysdeps/posix/raise.c:26
#8  0x00007fc7050c77f3 in __GI_abort () at ./stdlib/abort.c:79
#9  0x00007fc705583bb3 in qt_message_fatal (message=<synthetic pointer>..., context=...) at global/qlogging.cpp:1914
#10 QMessageLogger::fatal (this=this@entry=0x7ffe396eabe0, msg=msg@entry=0x7fc7037ef0b8 "The Wayland connection experienced a fatal error: %s") at global/qlogging.cpp:893
#11 0x00007fc703763354 in QtWaylandClient::QWaylandDisplay::checkError (this=<optimized out>) at ./src/client/qwaylanddisplay.cpp:211
#12 0x00007fc70376406a in QtWaylandClient::QWaylandDisplay::flushRequests (this=0x557fcb547600) at ./src/client/qwaylanddisplay.cpp:222
#13 0x00007fc7057e3e58 in doActivate<false> (sender=0x557fcb5642c0, signal_index=4, argv=0x7ffe396eace0) at kernel/qobject.cpp:3898
#14 0x00007fc7057dd137 in QMetaObject::activate (sender=sender@entry=0x557fcb5642c0, m=m@entry=0x7fc705a4a840 <QAbstractEventDispatcher::staticMetaObject>, local_signal_index=local_signal_index@entry=1, argv=argv@entry=0x0) at kernel/qobject.cpp:3946
#15 0x00007fc7057a8f77 in QAbstractEventDispatcher::awake (this=this@entry=0x557fcb5642c0) at .moc/moc_qabstracteventdispatcher.cpp:149
#16 0x00007fc70580594b in QEventDispatcherGlib::processEvents (this=0x557fcb5642c0, flags=...) at kernel/qeventdispatcher_glib.cpp:430
#17 0x00007fc7057aae6b in QEventLoop::exec (this=this@entry=0x7ffe396eae00, flags=..., flags@entry=...) at ../../include/QtCore/../../src/corelib/global/qflags.h:69
#18 0x00007fc7057b33f4 in QCoreApplication::exec () at ../../include/QtCore/../../src/corelib/global/qflags.h:121
#19 0x0000557fcab5edce in main (argc=<optimized out>, argv=<optimized out>) at ./shell/main.cpp:238
[Inferior 1 (process 1351) detached]

I also have this problem on openSUSE Tumbleweed.
Heare is my bug report on openSUSE bugzilla https://bugzilla.opensuse.org/show_bug.cgi?id=1196975, maybe it will be usefull.
If you need, I can also try to provide some debug/stacktrace/etc.. .

It seems that installing qtwayland from the PPA below changed the behaviour. Neither Plasma nor ksplashqml ckash anymore. But they do not show on the screen either

https://launchpad.net/~ci-train-ppa-service/+archive/ubuntu/4829

Ok cool. Sounds like it was a packaging issue. Feel free to report additional bugs separately.

(In reply to Nate Graham from comment #13)
> Ok cool. Sounds like it was a packaging issue. Feel free to report
> additional bugs separately.

Sorry for asking, but will it be fixed on openSUSE Tumbleweed/Leab15.4?
Or I should create here a new separate report about openSUSE?

vasya.ukraine, that PPA I mentioned contains QtWayland with KDE patches. Debian also has plans to integrate them: https://salsa.debian.org/qt-kde-team/qt/qtwayland/-/merge_requests/5#note_305550

So look around, maybe notify OpenSuse maintainers aboutthese patches. I imagine it is doable.

That being said, Plasma still does not work for me on NVIDIA drivers, but the problem seems to be different.

Слава Україні ✌️

Created attachment 148323 [details]
New crash information added by DrKonqi

plasmashell (5.24.4) using Qt 5.15.3

- What I was doing when the application crashed:
Launching Kde Plasma Wayland session resulting in an instant crash.  With the only error box showing at first mentioning Ksplash.qml.  NOTE this is with a AMD APU.  It might not be an Nvidia issue!   That said I do Have an Nvidia GPU for passthrough gaming and another one for using CUDA code and NVENC. d

- Unusual behavior I noticed:
It is such a hard and total crash that I cannot even CTRL-Alt Del and log out.  Were I not a very able user this would be a HUGE problem.   NOTE before I did an upgrade of Kubuntu 21.10 to 22.04 I had a working KDE Plasma Wayland environment.  I don't mean just barely working I mean ... other than a few apps not optimized for Kwin Wayland like Zoom I could use it as a default.   Furthermore. 

Everything works in KDE Plasma for X11 on this 22.04 system. 
Wayland works just fine in Gnome on this 22.04 system. 

This was an upgrade in place so I thought maybe I should try to nuke all my GUI's reboot, sudo apt update, etc etc  then apt install Kde again.   Still wayland gives me no joy.  

- Custom settings of the application:

To ensure there was not an issue of special settings uninstalled my Kde and gnome to the point where I booted to command line.  Then reinstalled this issue still occurs.  This could be a KDE problem or a Kubuntu problem.  Kubuntu being a main way people use KDE kinda makes this a KDE problem too.

-- Backtrace (Reduced):
#4  __pthread_kill_implementation (no_tid=0, signo=6, threadid=140710538758592) at ./nptl/pthread_kill.c:44
#5  __pthread_kill_internal (signo=6, threadid=140710538758592) at ./nptl/pthread_kill.c:78
#6  __GI___pthread_kill (threadid=140710538758592, signo=signo@entry=6) at ./nptl/pthread_kill.c:89
#7  0x00007ff9bde52476 in __GI_raise (sig=sig@entry=6) at ../sysdeps/posix/raise.c:26
#8  0x00007ff9bde387f3 in __GI_abort () at ./stdlib/abort.c:79

Created attachment 148663 [details]
New crash information added by DrKonqi

plasmashell (5.24.4) using Qt 5.15.3

- What I was doing when the application crashed:
Starting a plasma wayland session after upgrading Ubuntu 21.04 to 22.04

- Custom settings of the application:
* This happens when I use the 5.15.0-27-lowlatency kernel image, but I also have a 5.15.0-5.4-liquorix-amd64 kernel image which runs fine. So it might either be a 5.15.0-27 issue, or somehow the liquorix kernel features work as a workaround.
* On a Nvidia GeForce GTX 1050 Ti Mobile graphics card and using the 470.103.01 proprietary Nvidia driver.

-- Backtrace (Reduced):
#4  __pthread_kill_implementation (no_tid=0, signo=6, threadid=140341219895744) at ./nptl/pthread_kill.c:44
#5  __pthread_kill_internal (signo=6, threadid=140341219895744) at ./nptl/pthread_kill.c:78
#6  __GI___pthread_kill (threadid=140341219895744, signo=signo@entry=6) at ./nptl/pthread_kill.c:89
#7  0x00007fa3c0c68476 in __GI_raise (sig=sig@entry=6) at ../sysdeps/posix/raise.c:26
#8  0x00007fa3c0c4e7f3 in __GI_abort () at ./stdlib/abort.c:79