https://bugs.kde.org/show_bug.cgi?id=523995
Bug ID: 523995
Summary: Panel struts are lost when the monitor comes back from
a long DPMS off: maximized windows extend under the
panel
Classification: Plasma
Product: plasmashell
Version First 6.7.4
Reported In:
Platform: CachyOS
OS: Linux
Status: REPORTED
Severity: normal
Priority: NOR
Component: Panel
Assignee: [email protected]
Reporter: [email protected]
CC: [email protected]
Target Milestone: 1.0
If the display stays off long enough, turning it back on makes every Qt client
lose
its outputs for a moment, and plasmashell logs:
plasmashell[...]: requesting unexisting screen available rect -1
After that the work area ignores the panel. Maximized windows go all the way
down and
the dock sits on top of their content. The panel is drawn fine, it just stops
reserving space. Restarting plasmashell fixes it until the next time.
I have two panels, a floating dock at the bottom and a normal one at the top.
Only the
dock loses its space.
STEPS TO REPRODUCE
1. Maximize a window, check that it stops at the dock.
2. Let the display turn off. Locking gets there in 5 minutes here, 15 unlocked.
3. Leave it off for about 20 minutes. A quick "kscreen-doctor -d off" and an
immediate
wakeup does not reproduce it, the outputs never go away.
4. Wake it up.
OBSERVED: the window now extends under the dock.
EXPECTED: the work area still accounts for the panel.
Happens on every long cycle here: six out of six in the last four days, and
none of
the ~35 short lock cycles. Log from one of them:
17:42:02 session locked
17:47:51 /sys/class/drm/card2-HDMI-A-2/enabled -> disabled
...23 minutes off...
18:10:59.934 unlocked (fingerprint)
18:11:03.637 every Qt client: "There are no outputs - creating placeholder
screen"
18:11:03.6xx two DRM change uevents on card2
18:11:03.639 plasmashell: requesting unexisting screen available rect -1
(x2)
18:11:51 connector enabled again
The unlock is not part of it, it finished 3.7s before the outputs went away.
Locking
only matters because it shortens the timeout that powers the display off.
RELATED
476037 (kwin, fixed) is the same wakeup. Comment 2 there: "If a display
temporarily
pretends it's disconnected after waking up, that's not a problem we can fix.
All we
can do is to restore the state when the display gets recognized again." In that
bug
the state that came back wrong was window placement, and it was fixed. Here it
is the
panel's reserved space.
485932 and 515786 are plasmashell dying on the same wakeup, with the same
"unexisting
screen" line. Mine survives, it just loses the strut.
The display side is https://gitlab.freedesktop.org/drm/amd/-/issues/2958, still
open.
That is why only a long off reproduces this.
SYSTEM
Plasma 6.7.4, KWin 6.7.4, Qt 6.11.1, Wayland, CachyOS, kernel 7.2.0-rc6. Single
monitor on HDMI-A-2 driven by the AMD Radeon iGPU (Granite Ridge, amdgpu).
There is
also a GeForce RTX 5090 (driver 610.57.04) in the machine, but it drives no
display.
--
You are receiving this mail because:
You are watching all bug changes.
