Created attachment 164228 [details] The bug in action. Notice, how the cursor is almost correctly placed at the Show intro page button, but drastically misplaced at the X. SUMMARY The cursor is rendered at the wrong position on wayland. STEPS TO REPRODUCE 1. Start KDE Neon Unstable (at this moment running KDE Plasma 6 Beta 1), wayland session. 2. Open any window. 3. Try to click the X. OBSERVED RESULT The cursor clicks at a different position than it actually is. EXPECTED RESULT The cursor should click where it's rendered SOFTWARE/OS VERSIONS Windows: macOS: Linux/KDE Plasma: KDE Neon Unstable Edition (available in About System) KDE Plasma Version: 5.90.90 KDE Frameworks Version: 5.247.0 Qt Version: 6.6.1 ADDITIONAL INFORMATION Tested on QEMU, not a physical machine.
*** This bug has been marked as a duplicate of bug 427060 ***
See also Bug 478614, which was the same issue byt for VMWare and VIrtualBox VMs, and that one looks to have been fixed already.
Also encountered this... Seems OK just after an install as KVM guest (neon-unstable-20231220-0758.iso). It is wrong after I changed the Display/Monitor size (QXL graphics). I changed to 1440x900, although I had to log in with X to change it and then log back in with Wayland to check. There's a comment https://bugs.kde.org/show_bug.cgi?id=427060#c44 about the offset depending on where you are on the screen. Seems also to be the case here. It is better behaved with Virtio graphics.
If it depends on the mode, then I'm quite sure it's a bug in the VM driver. KWin just sets a cursor image with a hotspot (which is independent of the cursor position), and it's up to the VM driver and the VM to translate that to a proper cursor position on the host.
(In reply to Zamundaaa from comment #4) > If it depends on the mode, then I'm quite sure it's a bug in the VM driver. > KWin just sets a cursor image with a hotspot (which is independent of the > cursor position), and it's up to the VM driver and the VM to translate that > to a proper cursor position on the host. It does seem to be fragile... I know Neon Testing is in flux at the moment but Wayland with Video QXL gets the "offset" Wayland with Video Virtio is fine X11 with Video QXL is fine X11 with Video Virtio is fine I'm pretty sure it worked for a while and has slipped again. On Neon testing
I'm seeing offset today in Fedora 40 on Wayland with a libvirt VM with virtio graphics :( The usual "real position is slightly lower and to the right of the apparent position" case.