SUMMARY KWin/Wayland fails to initialize correctly on NVIDIA dGPU at session start. When the display is connected to the NVIDIA GPU at boot, the BIOS splash appears, but the screen goes black when KWin/SDDM should appear. After logging in via the Intel iGPU output and then switching the monitor back to the NVIDIA GPU, KWin shows abnormally high GPU usage until `kwin_wayland --replace` is executed. Restarting KWin immediately restores normal functionality and GPU usage. STEPS TO REPRODUCE 1. Boot system with monitor connected to NVIDIA HDMI output. 2. BIOS splash displays, then screen goes black instead of showing SDDM. 3. Move monitor cable to Intel iGPU HDMI and log in. 4. Move monitor cable back to NVIDIA HDMI. 5. Observe high GPU usage until running `kwin_wayland --replace &`. ## OBSERVED RESULT - No display output from NVIDIA GPU during login (black screen). - KWin initializes in an unstable state when session begins. - GPU usage abnormally high. - Running `kwin_wayland --replace` resets GPU usage to expected idle (~1%) and restores proper compositing. - Console output contains EGL context errors, for example: - `QWaylandGLContext: Failed to create the decorations EGLContext. Decorations will not be drawn.` ## EXPECTED RESULT - KWin should start with hardware acceleration correctly active on the NVIDIA GPU from the start of the session. - No need to hot-switch display cables or restart KWin. - Normal GPU usage at idle without manual interaction. ## SOFTWARE/OS VERSIONS - Fedora Linux 43 (KDE Plasma Desktop Edition) - Kernel: 6.17.12-300.fc43.x86_64 - KDE Plasma: 6.5.4 - KWin: 6.5.4 - SDDM: 0.21.0-10.fc43 - Session type: Wayland - GPU: Intel UHD Graphics 770 (i915) + NVIDIA GeForce RTX 5060 Ti - NVIDIA driver: 580.119.02 (proprietary) - Monitor connection: HDMI (primary attempt on NVIDIA; iGPU used as fallback) - Secure Boot: enabled
Thanks for the bug report. Can you provide us some system information so we can investigate? 1. Copy the output of this into the report: kscreen-doctor -o 2. Please attach the output of drm_info after triggering the issue 3. Also attach kwin's logs when the high GPU usage occurs journalctl --boot --user-unit plasma-kwin_wayland > ~/log.txt Thanks!
.
Created attachment 188252 [details] drm_info output
Created attachment 188253 [details] journalctl output
In addition to the above attachments, here is the output of the first item requested: kscreen-doctor -o Output: 1 HDMI-A-3 f148c46d-810d-4a60-a8b6-bd5ed05782db enabled connected priority 1 HDMI replication source:0 Modes: 1:3840x2160@60.00*! 2:4096x2160@59.94 3:4096x2160@29.97 4:4096x2160@24.00 5:4096x2160@23.98 6:3840x2160@59.94 7:3840x2160@29.97 8:3840x2160@23.98 9:1920x1080@60.00 10:1920x1080@59.94 11:1920x1080@29.97 12:1920x1080@23.98 13:1680x1050@59.95 14:1280x1024@60.02 15:1440x900@59.89 16:1280x960@60.00 17:1280x800@74.93 18:1280x800@59.81 19:1280x720@60.00 20:1280x720@59.94 21:1280x720@29.97 22:1280x720@23.98 23:1024x768@60.00 24:800x600@60.32 25:800x600@56.25 26:720x480@59.94 27:640x480@59.94 28:640x480@59.93 Geometry: 0,0 1920x1080 Scale: 2 Rotation: 1 Overscan: 0 Vrr: incapable RgbRange: unknown HDR: disabled Wide Color Gamut: disabled ICC profile: none Color profile source: sRGB Color power preference: prefer efficiency and performance Brightness control: supported
(In reply to TraceyC from comment #2) > . All the above reports were done while the issue was persisting - before I fix it with kwin_wayland --replace