SUMMARY Please check [1]: when an incoming call is accepted in plasma-dialer on top of the lockscreen (i.e. through the kde_lockscreen_overlay_v1 protocol [2]), at the end of the call plasma-dialer exits but the "active call list" ghost window remains on top of the shell. As shown in the video in [1], it is still possible to interact with the shell underneath. A reboot is needed to get back to a normal system on Mobian, while this is reported not to be enough e.g. on pmOS, where the workaround is to disable "session restore" feature. Main difference is that Mobian features a login manager, while pmOS launches plasma-mobile service automatically at boot. [1] https://invent.kde.org/plasma-mobile/plasma-dialer/-/issues/89#note_1110009 [2] https://invent.kde.org/plasma/kwin/-/merge_requests/2385 STEPS TO REPRODUCE 1. press power button to get system locked (or wait for the lockscreen timeout to expire) 2. phone call is received on the plasma-mobile system, "incoming call" screen is shown on top of lockscreen 3. accept call, then close it OBSERVED RESULT "active call list" screen remains on top of the screen EXPECTED RESULT no dialer-related screen should be shown after the call is ended SOFTWARE/OS VERSIONS Linux/KDE Plasma: Mobian testing on op6-enchilada (Linux 6.12-qcom) KDE Plasma Version: 6.2.4 KDE Frameworks Version: 6.8.0 Qt Version: 6.7.2 ADDITIONAL INFORMATION The issue seems related to the Qt6 transition in plasma-dialer, then it could very well be that the dialer's invocation of the lockscreen_overlay protocol needs to be ported. Generally, it's not clear whether the issue is on the dialer's or on kwin's side, then any hint on getting additional logs is welcome.
Created attachment 178443 [details] journal log with KWin debug logs enabled I hope some log with KWin debug output can help: IMO those 2 lines feb 16 17:56:18 mobian kwin_wayland_wrapper[1250]: kwin_scene_opengl: 0x1: fd_acc_get_query_result: a busy "0xaaaad31f4570: target=buffer, format=r8_unorm, 4096x1x1, array_size=1, last_level=0, nr_samples=0, usage=0, bind=40000, flags=0" BO stalled (0.011 ms) feb 16 17:56:18 mobian kwin_wayland_wrapper[1250]: kwin_scene_opengl: 0x1: fd_acc_get_query_result: a busy "0xaaaad1d6f480: target=buffer, format=r8_unorm, 4096x1x1, array_size=1, last_level=0, nr_samples=0, usage=0, bind=40000, flags=0" BO stalled (0.036 ms) show something in the system being overloaded?
Created attachment 178444 [details] journal log with KWin debug logs enabled and triple buffering disabled Another log, now setting KWIN_DRM_DISABLE_TRIPLE_BUFFERING=1 before rebooting. Those feb 16 18:07:53 mobian kwin_wayland_wrapper[1221]: kwin_scene_opengl: 0x1: fd_acc_get_query_result: a busy "0xaaaaeac74b40: target=buffer, format=r8_unorm, 4096x1x1, array_size=1, last_level=0, nr_samples=0, usage=0, bind=40000, flags=0" BO stalled (0.027 ms) lines at the end refer to the "active call list" screen being stuck on the display: I tap on the screen at random places, sometime I get a vibration feedback (like if I'm interacting with an invisible Plasma shell under the ghost screen) and those lines are triggered in the logs.