DS and tearing sometimes fail to activate together

[fxzzi]

In some cases its not possible for DS and Tearing to activate at the same time for some reason.

It's possible for DS and Tearing to become active at the same time in certain cases described by Ujint34 in #9719. I could trigger it sometimes by switching between workspaces rapidly.

As explained by Ujint34:

DS activation-deactivation logic is not executed when pMonitor->tearingState.nextRenderTorn. Sometimes it is possible to activate DS with tearing enabled before pMonitor->tearingState.nextRenderTorn gets a correct value. In this case lastScanout is not reset and hyprctl monitors shows incorrect value for directScanoutTo.

Removing the !shouldTear doesn't seem to fix it weirdly enough. I'm unable to tell if it helps a little bit tho

Hyprland/src/render/Renderer.cpp

Lines 1228 to 1232 in e96b8ce

	if ((*PDIRECTSCANOUT == 1 ||
	(*PDIRECTSCANOUT == 2 && pMonitor->activeWorkspace && pMonitor->activeWorkspace->m_bHasFullscreenWindow &&
	pMonitor->activeWorkspace->m_efFullscreenMode == FSMODE_FULLSCREEN && pMonitor->activeWorkspace->getFullscreenWindow()->getContentType() == CONTENT_TYPE_GAME)) &&
	!shouldTear) {
	if (pMonitor->attemptDirectScanout()) {

[vaxerski]

DS and Tearing are not mutually exclusive. This is not an issue, it's just working as expected. Why not have direct scanout tear? It's just a rendering pipeline skip.

[vaxerski]

wdym? Like, DS doesn't activate or what?

[fxzzi]

as in the behaviour is, as far as I can tell, the same as before.

Not sure if maybe they both activate more often before, but I'm still very commonly (more than 50% of the time) seeing that only tearing is activated and not DS

[fxzzi]

after further testing i can also confirm that ds seems to work perfectly pretty much every time when I have general:allow_tearing 0 so the issue is still present weirdly enough after removing that check

[vaxerski]

DS is still off by default right? I think it's a bit of a meh atm

[fxzzi]

yeah its off by default.

i don't think that simple patch has have any negative side effects. So it could be pushed just to ensure at least slightly more correct logic (even if its not completely fixing it for me)

[fxzzi]

funny update to this issue - seems its easier to trigger this bug if animations are disabled.

Without animations activelyTearing becomes true almost instantly when i switch to a workspace with a fullscreen game (without ds activating)

With animations enabled, I move to the workspace, the animation completes first, then both activelyTearing and directScanoutTo are correct. If I move between the two workspaces fast however (maybe before the animation finishes) it still triggers the issue.

Just thought I'd share