https://bugs.kde.org/show_bug.cgi?id=460594
Sergio <sergio.calleg...@gmail.com> changed:
What |Removed |Added
----------------------------------------------------------------------------
CC| |sergio.calleg...@gmail.com
--- Comment #35 from Sergio <sergio.calleg...@gmail.com> ---
Experiencing the same issue on Manjaro (see
https://forum.manjaro.org/t/plasmashell-freezes/127677/2).
Good news is that it is easy to check whether the issue is a freeze in
plasmashell or another issue. Furthermore if the problem is a freeze in
plasmashell it can be worked around without any data loss.
1. switch to a virtual console (e.g. CTRL+ALT+F2), login there, kill
plasmashell (e.g. killall plasmashell);
2. go back to the graphics console (e.g. CTRL+ALT+F1), if the panel is now
gone, then the problem was plasmashell *freezing*.
3. now it should be possible to get the runner (ALT+F2), restart plasmashell
from there (e.g. plasmashell --replace).
The above procedure lets all your applications survive.
I am seeing the issue with X11 (on an Intel Haswell laptop with integrated
Intel graphics), so this is definitely not a Wayland-related or Wayland
specific problem.
If I understand correctly, plasmashell uses a *cooperative* multitasking model
for its widgets, so it is enough to have one widget freezing to get a complete
plasmashell freeze, which makes freezes like this somehow easy to see,
particularly if a non-official widget is used.
The model has the advantage of being very lightweight, but clearly has some
disadvantage. Is there any possibility to complement it with a watchdog, so
that when the shell freezes after some time it is automatically restarted and
one can get an indication of the last widget the execution thread was in at the
freeze time?
--
You are receiving this mail because:
You are watching all bug changes.
