The PowerDevil service does not store screen brightness across restarts. It starts up with whatever the current screen brightness is, and other mechanisms will usually remember brightness upon shutdown (e.g. systemd for laptop screen backlight brightness, monitors with their own memory and firmware).

On Wayland, the direction (as of Plasma 6.2) is for KWin to store all brightness values and call PowerDevil to restore them for any monitors that are detected. Communication between KWin and PowerDevil happens via Wayland protocols.

On X11, there are no plans currently to introduce storage of brightness values across sessions. This is what you're hitting.

There is an ongoing effort - hopefully landing soon via https://invent.kde.org/plasma/powerdevil/-/merge_requests/466 - to make PowerDevil's screen brightness controller aware of the difference between nominal (configured) brightness and dimmed brightness.

Once we have this, we *could* try to detect when the service is shut down, delay shutdown and set the brightness back to un-dimmed brightness, and quit only afterwards. That seems like a whole bunch of work, however, and given the many other tasks that I've somehow ended up with, I'm not sure I can spare the time to implement this.

You write this:

> On boot, screen at 100% brightness, like it was before Plasma 6.2

That makes me assume you're using a laptop with built-in screen, plus an external monitor. Because the same issue was present long before Plasma 6.2, before 6.0 even, for people that used *only* a laptop screen or *only* an external monitor. The difference with 6.2 is that Plasma now controls more than one display, so it will *also* dim your external monitor in addition to the laptop screen (and then also leave that in a dimmed brightness level across restarts). That's our regression. I've been trying to come up with plans & some code in this issue: https://invent.kde.org/plasma/powerdevil/-/issues/38

But this particular issue won't be solved by my plans & code so far. One thing we could do on X11 across reboots, to make at least the regression go away, would be to avoid dimming external monitors for users with a built-in laptop display. On KWin, with brightness storage and hopefully soon a Wayland protocol for dimming, we can continue to dim. Does that sound reasonable?

For the remaining users, on X11 sessions with *only* a laptop screen or *only* external monitors attached to a desktop PC, this is a WONTFIX from my point of view unless someone else decides to step up and provide some patches to reset brightness on exit.

*** Bug 496954 has been marked as a duplicate of this bug. ***

Git commit ed7828d6c698b383bb16c71bc1b0e8cca16e2146 by Jakob Petsovits.
Committed on 16/12/2024 at 16:44.
Pushed by jpetso into branch 'master'.

daemon: Add virtuals to DisplayBrightness for optional "native" dimming

This is meant for display devices that provide a separate
dimming multiplier setting independently from the nominal
brightness value. By keeping nominal brightness untouched,
displays won't have incorrect brightness values in the edge case
when shutting down while the display was still dimmed.

KWin will implement this interface with a new Wayland protocol.

Direct-access display brightness controls (in X11) only provide
a single brightness value and are not expected to implement this.

M  +9    -0    daemon/controllers/displaybrightness.cpp
M  +16   -0    daemon/controllers/displaybrightness.h
M  +20   -11   daemon/controllers/screenbrightnesscontroller.cpp

https://invent.kde.org/plasma/powerdevil/-/commit/ed7828d6c698b383bb16c71bc1b0e8cca16e2146

Git commit 466555a78c34d1a5ce5543c1cfe701286565f4aa by Jakob Petsovits, on behalf of Xaver Hugl.
Committed on 16/12/2024 at 16:44.
Pushed by jpetso into branch 'master'.

daemon/controllers: implement dimming for kwin brightness

M  +13   -0    daemon/controllers/kwinbrightness.cpp
M  +3    -0    daemon/controllers/kwinbrightness.h

https://invent.kde.org/plasma/powerdevil/-/commit/466555a78c34d1a5ce5543c1cfe701286565f4aa