After package update (dnf upgrade) on 13/09/2024 plasma/kde on F40 is no longer usable because plasmashell crashes. My system is running with nvidia driver as I need CUDA.
Work-around: I currently use XFCE (and lightdm) instead of plasma/kde.
More details on in the (automated) bug report at 2312359 – [abrt] plasma-workspace: _sub_I_65535_0.0(): ksplashqml killed by SIGILL .
Any suggestions how to fix that?
Are you using the rpmfusion nvidia drivers?
Fyi the bug report seems to be for the ksplash crashing not plasmashell.
When I try to log in into plasma/kde, I get a konsole window and an error message about something wrong with plasmashell. Beside konsole, there is no sign of a ‘desktop environment’ (all black, no task bar, no background tile, …).
I use the nvidia driver from fedora-multimedia of negativo17.org . See details below.
$ sudo dnf list installed "*nvidia*"
akmod-nvidia.x86_64 3:560.35.03-1.fc40 @fedora-multimedia
kmod-nvidia-6.10.7-200.fc40.x86_64.x86_64 3:560.35.03-1.fc40 @@commandline
kmod-nvidia-6.10.8-200.fc40.x86_64.x86_64 3:560.35.03-1.fc40 @@commandline
kmod-nvidia-6.10.9-200.fc40.x86_64.x86_64 3:560.35.03-1.fc40 @@commandline
libnvidia-cfg.x86_64 3:560.35.03-3.fc40 @fedora-multimedia
libnvidia-ml.x86_64 3:560.35.03-3.fc40 @fedora-multimedia
libva-nvidia-driver.x86_64 1:0.0.12-1.fc40 @fedora-multimedia
nvidia-driver.x86_64 3:560.35.03-3.fc40 @fedora-multimedia
nvidia-driver-cuda.x86_64 3:560.35.03-3.fc40 @fedora-multimedia
nvidia-driver-cuda-libs.x86_64 3:560.35.03-3.fc40 @fedora-multimedia
nvidia-driver-libs.x86_64 3:560.35.03-3.fc40 @fedora-multimedia
nvidia-gpu-firmware.noarch 20240909-1.fc40 @updates
nvidia-kmod-common.noarch 3:560.35.03-1.fc40 @fedora-multimedia
nvidia-modprobe.x86_64 3:560.35.03-1.fc40 @fedora-multimedia
nvidia-persistenced.x86_64 3:560.35.03-1.fc40 @fedora-multimedia
xorg-x11-drv-nvidia-kmodsrc.x86_64 3:560.35.03-3.fc40 @rpmfusion-nonfree-updates
xorg-x11-nvidia.x86_64 3:560.35.03-3.fc40 @fedora-multimedia
$ sudo dnf list installed "*mesa*"
Installierte Pakete
mesa-dri-drivers.i686 1:24.1.6-1.fc40 @fedora-multimedia
mesa-dri-drivers.x86_64 1:24.1.6-1.fc40 @fedora-multimedia
mesa-filesystem.i686 1:24.1.6-1.fc40 @fedora-multimedia
mesa-filesystem.x86_64 1:24.1.6-1.fc40 @fedora-multimedia
mesa-libEGL.i686 1:24.1.6-1.fc40 @fedora-multimedia
mesa-libEGL.x86_64 1:24.1.6-1.fc40 @fedora-multimedia
mesa-libGL.i686 1:24.1.6-1.fc40 @fedora-multimedia
mesa-libGL.x86_64 1:24.1.6-1.fc40 @fedora-multimedia
mesa-libGLU.x86_64 9.0.3-4.fc40 @fedora
mesa-libgbm.i686 1:24.1.6-1.fc40 @fedora-multimedia
mesa-libgbm.x86_64 1:24.1.6-1.fc40 @fedora-multimedia
mesa-libglapi.i686 1:24.1.6-1.fc40 @fedora-multimedia
mesa-libglapi.x86_64 1:24.1.6-1.fc40 @fedora-multimedia
mesa-libxatracker.x86_64 1:24.1.6-1.fc40 @fedora-multimedia
mesa-va-drivers.i686 1:24.1.6-1.fc40 @fedora-multimedia
mesa-va-drivers.x86_64 1:24.1.6-1.fc40 @fedora-multimedia
mesa-vulkan-drivers.i686 1:24.1.6-1.fc40 @fedora-multimedia
mesa-vulkan-drivers.x86_64 1:24.1.6-1.fc40 @fedora-multimedia
it seems that you have mixed negativo’s fedora-multimedia repo with the rpmfusion-nonfree-updates repo, as you can see here:
xorg-x11-drv-nvidia-kmodsrc.x86_64 3:560.35.03-3.fc40
@rpmfusion-nonfree-updates
xorg-x11-nvidia.x86_64 3:560.35.03-3.fc40
@fedora-multimedia
Why isn’t the rpmfusion drivers good enough?
You are using the mesa packages from negativo instead of the fedora software as well as the nvidia drivers from negativo (and mixed with rpmfusion there).
Many users have had problems when using the negativo repos.
As requested, I’ve switched to rpmfusion (and also done akmods --force --rebuild and dracut --regenerate-all --force --verbose and a reboot):
$ sudo dnf list installed "*nvidia*"
Installierte Pakete
akmod-nvidia.x86_64 3:560.35.03-1.fc40 @rpmfusion-nonfree-updates
kmod-nvidia-6.10.9-200.fc40.x86_64.x86_64 3:560.35.03-1.fc40 @@commandline
nvidia-gpu-firmware.noarch 20240909-1.fc40 @updates
nvidia-modprobe.x86_64 3:560.35.03-1.fc40 @rpmfusion-nonfree-updates
nvidia-persistenced.x86_64 3:560.35.03-1.fc40 @rpmfusion-nonfree-updates
nvidia-settings.x86_64 3:560.35.03-1.fc40 @rpmfusion-nonfree-updates
xorg-x11-drv-nvidia.x86_64 3:560.35.03-3.fc40 @rpmfusion-nonfree-updates
xorg-x11-drv-nvidia-cuda.x86_64 3:560.35.03-3.fc40 @rpmfusion-nonfree-updates
xorg-x11-drv-nvidia-cuda-libs.i686 3:560.35.03-3.fc40 @rpmfusion-nonfree-updates
xorg-x11-drv-nvidia-cuda-libs.x86_64 3:560.35.03-3.fc40 @rpmfusion-nonfree-updates
xorg-x11-drv-nvidia-kmodsrc.x86_64 3:560.35.03-3.fc40 @rpmfusion-nonfree-updates
xorg-x11-drv-nvidia-libs.i686 3:560.35.03-3.fc40 @rpmfusion-nonfree-updates
xorg-x11-drv-nvidia-libs.x86_64 3:560.35.03-3.fc40 @rpmfusion-nonfree-updates
xorg-x11-drv-nvidia-power.x86_64 3:560.35.03-3.fc40 @rpmfusion-nonfree-updates
$ sudo dnf list installed "*mesa*"
Installierte Pakete
mesa-dri-drivers.i686 24.1.7-1.fc40 @updates
mesa-dri-drivers.x86_64 24.1.7-1.fc40 @updates
mesa-filesystem.i686 24.1.7-1.fc40 @updates
mesa-filesystem.x86_64 24.1.7-1.fc40 @updates
mesa-libEGL.i686 24.1.7-1.fc40 @updates
mesa-libEGL.x86_64 24.1.7-1.fc40 @updates
mesa-libGL.i686 24.1.7-1.fc40 @updates
mesa-libGL.x86_64 24.1.7-1.fc40 @updates
mesa-libGLU.x86_64 9.0.3-4.fc40 @fedora
mesa-libgbm.i686 24.1.7-1.fc40 @updates
mesa-libgbm.x86_64 24.1.7-1.fc40 @updates
mesa-libglapi.i686 24.1.7-1.fc40 @updates
mesa-libglapi.x86_64 24.1.7-1.fc40 @updates
mesa-libxatracker.x86_64 24.1.7-1.fc40 @updates
mesa-va-drivers.i686 24.1.7-1.fc40 @updates
mesa-va-drivers.x86_64 24.1.7-1.fc40 @updates
mesa-vulkan-drivers.i686 24.1.7-1.fc40 @updates
mesa-vulkan-drivers.x86_64 24.1.7-1.fc40 @updates
However this switch has not changed anything for the problem at hand.
The problem is not the driver or mesa.
Check which packages have been updated by using sudo dnf history and sudo dnf history info <transaction> .
Search for kde rpms and then maybe downgrade some of those packages.
kwin and friends would be my first candidates.
Unless someone else is seeing the same SIGILL crash without every installing the negativo RPMs then I an not sure you can assert this.
Do you have a new ABRT report that we can look at?
So you have a work around that fixed the problem for you 8n that ticket?
It is perhaps interesting to know that because of the problems described at https://discussion.fedoraproject.org/t/update-to-f40-renders-plasma-desktop-and-sddm-unusable-slow-sluggish-unresponsive I still use lightdm as display manager.
Hence I can assure you that plasma/kde is currently unusable even without using SDDM.
For me that looks like a really nasty bug…
I don’t use KDE. This was the result of a quick bodhi query.
It’s remarkable that this kwin upgrade was pushed to stable after only 2 positive feedbacks inside of only 2 days!
The kwin update works atleast for me on two systems.
A amdgpu system and a VM system.
There may have been an important fix in that update that is why it was pushed faster then usual. But I’m not sure I recall that correctly.
What I am wondering is if the install of the negativo RPMs changed your systems config.
Thoses changes may well not have been undone by removing there packages.
What do you have as kernel command line args? cat /proc/cmdline
$ cat /proc/cmdline
BOOT_IMAGE=(hd3,gpt2)/vmlinuz-6.10.9-200.fc40.x86_64 root=/dev/mapper/lvm2-root ro rd.luks.uuid=luks-<obmitted> rd.lvm.lv=lvm2/root rhgb quiet rd.neednet=1 ip=dhcp fbcon=font:TER16x32 rd.driver.blacklist=nouveau
Sorry, I don’t understand the kwin stuff. FEDORA-2024-b475969e23 — bugfix update for kwin — Fedora Updates System is stable and installed on my system. What to do to revert to an earlier version?
Apart from the “rd.neednet-1 ip=dhcp” there is nothing odd here that is good.
Just a thought, as you’ve mentioned XFCE. Do you start kde/plasma in Wayland or in a X11 session?
The gist of the ticket is that kwin has a dependency on libwayland-server >= 1.23.0 . But the system had a package amdgpu installed that also contained libwayland-server. This library was linked at runtime instead of the system library. So it’s probably unrelated to the issue you see.
In case you still want to downgrade to previous version.
You’ll find the dnf install command in the ticket , first message,
With lightdm (and with sddm) there is only the ‘plasma’ session option that will start plasma/kde on wayland.
In addition I’ve try to start plasma/kde from non graphical login with startplasma
and startplasma-wayland. For the problem at hand, this makes no difference.
problem is still there with plasma-workspace 6.1.5-1.fc40
For this version, I also reported the problem upstream: 493524 – Unable to start plasma/kde