I was really hoping this would turn out to be the cause of our Firefox SHM bug, but having tested it doesn't seem like it :(

@ekurzinger Do you have commit access?

This does not take into account: A server should avoid signaling the frame callbacks if the surface is not visible in any way, e.g. the surface is off-screen, or completely obscured by other opaque surfaces.

Now all windows are signaled. Especially also when the screen is locked.

This does not take into account: A server should avoid signaling the frame callbacks if the surface is not visible in any way,

ToplevelList windows = Workspace::self()->xStackingOrder();

This should exclude unmapped or minimised.

or completely obscured by other opaque surfaces.

We don't do this now.

especially lockscreen

There's a line above that removes from the "windows" list. (composite.cpp:740)

foreach (Toplevel *t, windows) {
   if (waylandServer() && waylandServer()->isScreenLocked()) {
        if(!t->isLockScreen() && !t->isInputMethod()) {
            windows.removeAll(t);
        }
    }

If anything, that means it was broken before and this fixes it.

In D18307#394965, @zzag wrote:

@ekurzinger Do you have commit access?

I do not, I believe someone would need to commit on my behalf.
Also, the diff has been updated per David's suggestion (thanks)

I believe someone would need to commit on my behalf.

You should have access now.