Created attachment 164805 [details] KScreen setup SUMMARY After update of kwin6 to 5.91.90git.20240111T012852~e427ad73-77.1 (openSUSE Tumbleweed KDE Unstable repository) mouse cursor disappeared on internal screen for any application or plasma shell. Even when cursor is near the edge of external screen and part of it should be visible on internal — that part became invisible. Also cursor has artifacts on the edge, btw I'm using older version of Bibata Amber cursors, not sure if this true for native Breeze, but these artifacts weren't present before on kwin6. I suspect this is a software cursors problem, as I have an Intel GPU on laptop. I think this commit: https://invent.kde.org/plasma/kwin/-/commit/b214251f81cbacacf3ef483081a7cef15343fa7f discovered this bug. If I add `export KWIN_FORCE_SW_CURSOR=0` to `~/.config/plasma-workspace/env/` folder to revert to hardware cursors, then both these problems go away — cursor is visible everywhere and no visual artifacts. Also maybe this is important, I'm using fractional scaling, 1.25 on external screen & 1.5 on internal. I think artifacts related to this, but whu they aren't present with HW cursor? Different resampling algorithm? Details in kscreen-doctor output in attachment. STEPS TO REPRODUCE 1. Use laptop with Intel GPU. 2. Setup internal & external screens. 3. Move mouse cursor between external & internal screens. OBSERVED RESULT Cursor is invisible on internal screen. EXPECTED RESULT Cursor should be visible on both screens. SOFTWARE/OS VERSIONS Linux/KDE Plasma: openSUSE 20240109 (available in About System) KDE Plasma Version: 6.0.80 KDE Frameworks Version: 5.249.0 Qt Version: 6.6.1 ADDITIONAL INFORMATION
Additonal information: 1. Cursor visible when external screen is not plugged or switched off in settings, but disappears again on plug in (switch on). 2. Cursor looks fine (no artifacts & visible) on screenshots. 3. I've tried to set 100% scaling and it seems doesn't affect, these bugs still present.
I've made a separate bug for cursor artifacts: https://bugs.kde.org/show_bug.cgi?id=479637
I have a similar issue. In my case the cursor is visible properly in 5.91.0. In 5.92.0 the cursor is only visible on the (primary) laptop screen that is driven by an Intel iGPU, but not on the external displays connected to the NVidia GPU. In my case I tried setting both WLR_NO_HARDWARE_CURSORS=1 and KWIN_FORCE_SW_CURSOR=1 in which the problem didn't go away (Reverting to 5.91.0 on an otherwise RC1 install "fixes" things, even with the environment variables still set. In dmesg I have the message: "[drm] [nvidia-drm] [GPU ID 0x00000100] Framebuffer memory not appropriate for scanout". It also seems related to https://forums.developer.nvidia.com/t/hardware-cursor-is-not-working-on-wayland-drm-sessions/261853
*** This bug has been marked as a duplicate of bug 479583 ***