https://bugs.kde.org/show_bug.cgi?id=509669

--- Comment #15 from fililip <[email protected]> ---
This issue manifests itself in the same manner regardless of whether or not the
hardware cursor is in use (I tried KWIN_FORCE_SW_CURSOR=1 as per my comment
above).

As such it can't be caused by DRM or amdgpu, especially when session content
presentation is otherwise stable (when the cursor position stutters there are
no desktop-wide freezes/mini-lockups), hardware cursor image changes work fine
prior to the regression commit on the same kernel version, and the problem
happens across multiple AMD Display Core Next ASIC versions.

Ideally I'd test this on an Intel machine as well, but I am unfortunately no
longer in possession of any. (NVIDIA has a hardware cursor position change +
presentation consistency issue so that is not a feasible option here)

It's not tied to overlay planes either, since I have both two active screens
and KWIN_USE_OVERLAYS set to 0 (and it's exactly the same even with overlays
enabled anyway).

My question is: is it possible to temporarily force KWin to perform these
cursor updates asynchronously again? How should I go about trying to fix it
myself (ie if it makes sense to somehow bring back connect() for
Cursor::positionChanged in compositor.cpp)?

I realize this issue is a bit nitpicky (in contrast to
https://bugs.kde.org/show_bug.cgi?id=504321 for example), and that there are
more pressing matters to attend to (and I do apologize for the noise), but
there's a chance that if this ends up in the release version, people will
notice it on their 1kHz/8kHz mice and highres + high refresh rate setups. I
caught this in around an hour after grabbing 6.4.90 packages.

-- 
You are receiving this mail because:
You are watching all bug changes.

Reply via email to