Blackscreen on opening Overview and Tiling-Manager with a Wayland session on aarch64

Hello Community,

I don’t want to file my issue as a bug yet although I have the suspicion it is in fact an bug.
As I couldn’t yet determine if it is a upstream bug or just a packaging issue at Debian and only have dug further into the issue as using journalctl, I deiced to first go to the community and have so team-troubleshooting.

Description

When opening the Overview on a Plasma Wayland Session using Meta + W instead of seeing the Overview I simply see a black screen. Once I close the overview again using ESC or Meta + W I see the Desktop again. The same happens with the tiling manager (Meta + T).
An important detail is that I use KDE Plasma on a Raspberry Pi 5, so the issue probably only affects the aarch64 architecture. I haven’t experienced it on x86 but I also don’t use Debian based distros as a desktop on x86.

As I have other issues with using an X11 Session for won’t I won’t bother filing a bug, as X11 is almost dead anyway.

System-Information

Variable Value
OS Debian 13.3 (Raspberry Pi OS 64bit)
KDE Plasma Version 6.3.6
KDE Frameworks Version 6.13.0
QT Version 6.8.2
Kernel 6.12.62+rpt-rpi-2712 (64-bit)
Instruction Set aarch64
Graphics Platform Wayland
Processors 4 × ARM Cortex-A76
Memory 8 GiB
Graphics Processor V3D 7.1.7.0
Product Raspberry Pi 5 Model B Rev 1.0

I installed the plasma desktop with apt install kde-full so there shouldn’t be much of missing dependencies.

Logs

I checked journalctl for what happens when I hit Meta + W in X11 and Wayland.

Jan 15 20:05:10 rasp-the-berry kwin_wayland_wrapper[7005]: kwin_scene_opengl: 0x1: GL_INVALID_ENUM in glEnable(GL_POINT_SPRITE)
Jan 15 20:05:10 rasp-the-berry kwin_wayland_wrapper[7005]: kwin_scene_opengl: 0x1: GL_INVALID_ENUM in glEnable(GL_POINT_SPRITE)
Jan 15 20:05:10 rasp-the-berry kwin_wayland_wrapper[7005]: kwin_scene_opengl: 0x1: GL_INVALID_VALUE in glTexImage2D(internalFormat=GL_ALPHA)
Jan 15 20:05:10 rasp-the-berry kwin_wayland_wrapper[7005]: kwin_scene_opengl: 0x1: GL_INVALID_VALUE in glTexImage2D(internalFormat=GL_ALPHA)
Jan 15 20:05:10 rasp-the-berry kwin_wayland_wrapper[7005]: kwin_scene_opengl: 0x1: GL_INVALID_VALUE in glTexImage2D(internalFormat=GL_ALPHA)
Jan 15 20:05:10 rasp-the-berry kwin_wayland_wrapper[7005]: kwin_scene_opengl: 0x1: GL_INVALID_VALUE in glTexImage2D(internalFormat=GL_ALPHA)
Jan 15 20:05:10 rasp-the-berry kwin_wayland_wrapper[7005]: kwin_scene_opengl: 0x1: GL_INVALID_VALUE in glTexImage2D(internalFormat=GL_ALPHA)
Jan 15 20:05:10 rasp-the-berry kwin_wayland_wrapper[7005]: kwin_scene_opengl: 0x1: GL_INVALID_VALUE in glTexImage2D(internalFormat=GL_ALPHA)
Jan 15 20:05:10 rasp-the-berry kwin_wayland_wrapper[7005]: kwin_scene_opengl: 0x1: GL_INVALID_OPERATION in glVertexAttribPointer(no array object bound)
Jan 15 20:05:10 rasp-the-berry kwin_wayland_wrapper[7005]: kwin_scene_opengl: 0x1: GL_INVALID_OPERATION in glVertexAttribPointer(no array object bound)
Jan 15 20:05:10 rasp-the-berry kwin_wayland_wrapper[7005]: kwin_scene_opengl: 0x1: GL_INVALID_OPERATION in glVertexAttribPointer(no array object bound)
Jan 15 20:05:10 rasp-the-berry kwin_wayland_wrapper[7005]: kwin_scene_opengl: 0x1: GL_INVALID_OPERATION in glDrawElements
# The last 4 lines key repeating unitl I hit Meta + W again

Sometimes a big block of this is in-between the 4 repeating lines

Jan 15 20:05:10 rasp-the-berry kwin_wayland_wrapper[7005]: kwin_scene_opengl: 0x1: GL_INVALID_OPERATION in glTexSubImage2D(invalid texture level 0)
Jan 15 20:05:10 rasp-the-berry kwin_wayland_wrapper[7005]: kwin_scene_opengl: 0x1: GL_INVALID_OPERATION in glTexSubImage2D(invalid texture level 0)
Jan 15 20:05:10 rasp-the-berry kwin_wayland_wrapper[7005]: kwin_scene_opengl: 0x1: GL_INVALID_OPERATION in glTexSubImage2D(invalid texture level 0)

I also once observed, that the kwin_scene_opengl: has the hexadecimal value 0x2 instead of 0x1

When I use Meta + W in X11 the journal does not report any kwin related messages.

Epilogue

If I miss any information I’m could provide.

What does the community thing is this a packaging issue from Debian or an issue in upstream KDE?

Regards

Pascal