https://bugs.kde.org/show_bug.cgi?id=525886
Bug ID: 525886
Summary: Panel frozen after resume from suspend (Wayland,
NVIDIA): plasmashell logs "QRhiGles2: Context is
lost", cleans up the scenegraph and never rebuilds it
Classification: Plasma
Product: plasmashell
Version First 6.7.5
Reported In:
Platform: CachyOS
OS: Linux
Status: REPORTED
Severity: major
Priority: NOR
Component: Panel
Assignee: [email protected]
Reporter: [email protected]
Target Milestone: 1.0
DESCRIPTION
After resuming from suspend the Plasma panel freezes visually, but plasmashell
itself stays alive:
* the clock stops updating;
* newly opened applications never appear in the Task Manager;
* the automatic day/night color scheme switch is never drawn (the panel keeps
the dark pre-suspend look).
Clicking pre-existing items still works: taskbar entries of windows opened
before the suspend still switch windows, and widget icons still react to
clicks. So the GUI thread and QML logic keep running, the panel just never gets
another frame and shows the last one from before the suspend.
plasmashell journal right at resume:
Sep 18 08:18:03 plasmashell[46016]: QRhiGles2: Context is lost.
Sep 18 08:18:03 plasmashell[46016]: Graphics device lost, cleaning up
scenegraph and releasing RHIs
Sep 18 08:18:03 plasmashell[46016]: QRhiGles2: Context is lost.
After these lines plasmashell keeps logging normally (PipeWire "No object for
name ..." lines continue with fresh timestamps, so the QML logic is running),
but the panel never renders again until the shell is restarted.
`systemctl --user restart plasma-plasmashell.service` fixes it right away: the
panel rebuilds and comes back with the correct (light) color scheme.
STEPS TO REPRODUCE
1. NVIDIA Wayland session (RTX 5060 Laptop GPU, NVIDIA open kernel module
615.71.09).
2. Suspend the machine.
3. Resume. That resume was messy on its own: kwin logged "atomic commit failed:
Permission denied" and nvidia_drm logged a sync-FD-semaphore registration error
in the same boot. kwin and regular applications recovered their contexts;
plasmashell did not.
4. Look at the panel.
OBSERVED RESULT
The panel keeps showing the stale pre-suspend frame indefinitely while
plasmashell stays healthy and interactive. Only restarting plasmashell brings
rendering back.
EXPECTED RESULT
After a GPU context loss the Qt Quick scenegraph should be rebuilt (or
plasmashell should recover on its own, for example by recreating the window).
The panel must start rendering again after resume.
SOFTWARE/OS VERSIONS
Operating System: CachyOS Linux (kernel 7.2.4-3-cachyos-bore-lto, systemd 261)
KDE Plasma Version: 6.7.5
KDE Frameworks Version: 6.30.0
Qt Version: 6.11.2
ADDITIONAL INFORMATION
Workaround: restart plasmashell automatically on every resume with a systemd
SYSTEM unit (user managers have no sleep targets as of systemd 261):
User=<user>, Environment=XDG_RUNTIME_DIR=/run/user/<uid>, ExecStart=systemctl
--user try-restart plasma-plasmashell.service, WantedBy=suspend.target.
Related reports: 496926 (X11 only, desktop black after resume until plasmashell
restart, the reporter confirms it doesn't happen on kwin_wayland), 496821
(calDAV-triggered freeze plus memory leak, different mechanism), 525537
(Wayland 6.7.5, the opposite symptom: endless repaint after resume, also fixed
by restarting plasmashell), 515416 (screen stays black and unresponsive, the
whole session is dead). Qt bug tracker: nothing found for scenegraph not being
rebuilt after device loss (searched 2026-09-18).
--
You are receiving this mail because:
You are watching all bug changes.
