plasmashell segmentation fault (signal 11) on KDE Neon Wayland session

Question

Goal: assess whether the experimental Wayland session in KDE Neon can serve as a stable base for a repeatable development environment when using NVIDIA proprietary drivers.

Constraint: the Wayland variant is marked experimental, and users with NVIDIA drivers frequently observe plasmashell crashes reporting a segmentation fault (signal 11), with no consensus on readiness.

Uncertainty: it remains unclear under which specific conditions the fault occurs, whether any configuration tweaks mitigate it without reverting to X11, and how the choice influences reproducibility of GUI‑based tests.

What specific driver or kernel versions trigger the plasmashell segmentation fault on Wayland?
Are there interim work‑arounds that allow Wayland to be used reliably for development without switching to the X11 ISO?
How does selecting Wayland versus X11 affect the consistency of automated GUI tests in a repeatable Neon‑based environment?

Tasadduq Burney · Answer

What triggers the crash?
In KDE Neon’s experimental Wayland session the most common culprits are:

Legacy NVIDIA driver families – 470.xx and 525.xx – which contain a known out‑of‑bounds write in the DMABUF feedback path that KWin’s compositor hits during a Wayland frame.
Kernel 5.15+ when the driver is compiled with DMABUF feedback enabled. The interaction with KWin’s triple‑buffering (KWIN_TRIPLE_BUFFER=1) can trigger the segfault.

Confirmed work‑arounds
These steps are not officially supported but have been observed to reduce crash frequency in a repeatable Neon build:

Upgrade the driver to the latest 535.xx series. The NVIDIA team back‑ported the DMABUF patch that fixes the out‑of‑bounds write.
sudo apt update
sudo apt install nvidia-driver-535
sudo reboot

Disable triple buffering for Wayland. Edit /etc/environment or add a session‑specific override:
export KWIN_TRIPLE_BUFFER=0

Then start a fresh Wayland session.
Drop DMABUF feedback for the driver. If you cannot upgrade, add a Xorg config snippet to disable DMABUF feedback:
Section "Device"
    Identifier "NVIDIA"
    Driver "nvidia"
    Option "AllowEmptyInitialConfiguration" "true"
    Option "RegistryDwords" "NvOptimusEnablement=0x00000000;PowerMizerEnablement=0x00000000"
EndSection

Enable Wayland‑specific debug logging. Launch plasmashell with:
QT_LOGGING_RULES=qt.qpa.wayland=true plasmashell

Then inspect journalctl -u plasmashell for warnings that precede a crash.
Collect a backtrace. After a segfault, run:
journalctl -u plasmashell | tail -n 200

or use the KDE crash handler to export the crash file for analysis.

Impact on automated GUI tests
Wayland’s non‑deterministic frame timing and the lack of a global X server mean tools that emulate mouse/keyboard events (xdotool, Sikuli) become unreliable. Tests that rely on xdotool will miss events or fail intermittently. Using wdm or kwin_wayland‑specific libraries, or running tests inside an Xwayland sandbox, restores determinism but adds overhead.
When to switch back to X11
If the above mitigations do not eliminate crashes or if your test matrix requires 100% deterministic event delivery, the safest path is to run the X11 ISO session. The X11 session is stable with NVIDIA drivers and fully compatible with existing automation tools.
Next diagnostic step
To fine‑tune the recommendation, could you confirm the exact NVIDIA driver version you are currently using? This will help determine if the 535.xx upgrade is already in place or if a driver downgrade might be necessary.

plasmashell segmentation fault (signal 11) on KDE Neon Wayland session

1 answer

What triggers the crash?

Confirmed work‑arounds

Impact on automated GUI tests

When to switch back to X11

Next diagnostic step

Your perspective belongs here.

0 question comments