https://bugs.kde.org/show_bug.cgi?id=522564
--- Comment #5 from Michael <[email protected]> --- I believe I am seeing a related variant of this issue on Fedora KDE Wayland. **System** * Fedora Linux 44 * KDE Plasma 6.7.5 * KWin 6.7.5 * libkscreen 6.7.5 * PowerDevil 6.7.5 * Wayland session * AMDGPU * External monitor connected via HDMI-A-1 * Internal eDP-1 panel is connected but disabled * HDMI-A-1 is the active/primary display **Observed behaviour** After the system has been idle and is at the lock/login screen, the external HDMI monitor becomes black/standby. If I log in, the external display becomes available again. During the failure, the important symptom is that KDE temporarily reports that there are **no outputs at all**. At monotonic timestamp **6586.634**, many independent KDE processes simultaneously report: `There are no outputs - creating placeholder screen` This includes `kactivitymanagerd`, `kded6`, multiple Akonadi processes, `xdg-desktop-portal-kde`, `kdeconnectd`, `kscreenlocker_greet`, `kwalletd6`, `baloorunner`, `plasmashell`, and PowerDevil. `plasmashell` also reports: `requesting unexisting screen available rect -1` A second essentially identical wave occurs at approximately **6587.681**. **Notably, I cannot find a corresponding lower-level display event.** I checked the journal around the failure and found: * no DRM/AMDGPU kernel messages * no HDMI-A-1 hotplug/disconnect event * no KWin crash or restart * no PowerDevil restart * no suspend/resume event * no logged DPMS event * no KScreen configuration event The kernel journal specifically contains **no messages at all from monotonic 6500 through 6650**, covering the failure. KWin and PowerDevil also remain running; neither service has a state change in the relevant 6560–6620 interval. The PowerDevil DDC/I2C permission errors occur after the initial "no outputs" transition, so they appear to be downstream of the output loss rather than the initiating event. The relevant journal sequence therefore appears to be: 1. KDE's output information becomes empty. 2. Multiple independent KDE clients simultaneously create placeholder screens. 3. PowerDevil reacts to the same state. 4. There is no corresponding kernel/DRM/HDMI event in the journal. 5. Logging in subsequently restores the display. The current normal state, after recovery, is: * HDMI-A-1: connected, enabled, primary * eDP-1: connected, disabled Could this be related to the KWin/libkscreen output-device lifecycle/configuration issue in this bug, or is there another component in the Wayland output path that would be useful to instrument? I can provide the complete journal around the event, `kscreen-doctor -o` output, and any additional diagnostics that would help distinguish a KWin/KScreen userspace output-state problem from an underlying DRM connector problem. Note - journalctl -b 0 -k -o short-monotonic --no-pager | \ awk '{ t=$0 sub(/^\[[[:space:]]*/, "", t) sub(/\].*/, "", t) if (t+0 >= 6500 && t+0 <= 6650) print }' > kde-output-failure-kernel-6500-6650.log Produces no output ... Also note - this was all generated with the help of ChatGPT -- You are receiving this mail because: You are watching all bug changes.
