SUMMARY I used the tool darkman (https://darkman.whynothugo.nl/) to automate dark theme switching at night. This has stopped working since updating to Plasma 5.26. I've found that lookandfeeltool, which was invoked by darkman via scripts, crashed every time darkman activates, but invoking it manually via command-line works. Using cron to invoke lookandfeeltool also appears to not work, but I have not found coredumps caused by that. Operating System: Arch Linux KDE Plasma Version: 5.26.0 KDE Frameworks Version: 5.99.0 Qt Version: 5.15.6 Kernel Version: 6.0.2-arch1-1 (64-bit) Graphics Platform: Wayland BACKTRACE #0 0x00007faa7ae81b1b in __GI_mincore () at ../sysdeps/unix/syscall-template.S:117 #1 0x00007faa740bfe68 in _eglPointerIsDereferencable () at ../mesa-22.2.1/src/egl/main/eglglobals.c:159 #2 0x00007faa740cf21b in get_wl_surface_proxy () at ../mesa-22.2.1/src/egl/drivers/dri2/platform_wayland.c:422 #3 dri2_wl_create_window_surface () at ../mesa-22.2.1/src/egl/drivers/dri2/platform_wayland.c:701 #4 0x00007faa740b6680 in _eglCreateWindowSurfaceCommon () at ../mesa-22.2.1/src/egl/main/eglapi.c:994 #5 0x00007faa740b7b19 in eglCreatePlatformWindowSurface () at ../mesa-22.2.1/src/egl/main/eglapi.c:1071 #6 0x00007faa75567ad4 in QtWaylandClient::QWaylandGLContext::updateGLFormat (this=0x555a86357980) at /usr/src/debug/qtwayland/src/hardwareintegration/client/wayland-egl/qwaylandglcontext.cpp:357 #7 QtWaylandClient::QWaylandGLContext::QWaylandGLContext (share=0x555a863579c0, format=..., display=<optimized out>, eglDisplay=<optimized out>, this=0x555a86357980) at /usr/src/debug/qtwayland/src/hardwareintegration/client/wayland-egl/qwaylandglcontext.cpp:340 #8 QtWaylandClient::QWaylandEglClientBufferIntegration::createPlatformOpenGLContext (this=this@entry=0x555a86275030, glFormat=..., share=share@entry=0x0) at /usr/src/debug/qtwayland/src/hardwareintegration/client/wayland-egl/qwaylandeglclientbufferintegration.cpp:155 #9 0x00007faa79f0b6e3 in QtWaylandClient::QWaylandIntegration::createPlatformOpenGLContext (this=<optimized out>, context=0x7ffc64a42e40) at /usr/src/debug/qtwayland/src/client/qwaylandintegration.cpp:178 #10 0x00007faa7ba710e3 in QOpenGLContext::create (this=this@entry=0x7ffc64a42e40) at kernel/qopenglcontext.cpp:612 #11 0x00007faa755a8191 in checkBackend (checkContext=...) at /usr/src/debug/plasma-integration-5.26.0/src/platformtheme/qtquickrenderersettings.cpp:43 #12 initializeRendererSessions () at /usr/src/debug/plasma-integration-5.26.0/src/platformtheme/qtquickrenderersettings.cpp:79 #13 initializeRendererSessions () at /usr/src/debug/plasma-integration-5.26.0/src/platformtheme/qtquickrenderersettings.cpp:52 #14 0x00007faa7b51902f in qt_call_pre_routines () at kernel/qcoreapplication.cpp:317 #15 QCoreApplicationPrivate::init (this=this@entry=0x555a86271b70) at kernel/qcoreapplication.cpp:849 #16 0x00007faa7ba2a1b9 in QGuiApplicationPrivate::init (this=this@entry=0x555a86271b70) at kernel/qguiapplication.cpp:1530 #17 0x00007faa7c0d0e0e in QApplicationPrivate::init (this=0x555a86271b70) at kernel/qapplication.cpp:513 #18 0x00007faa7c0d0f34 in QApplication::QApplication (this=<optimized out>, argc=<optimized out>, argv=<optimized out>, _internal=<optimized out>, this=<optimized out>, argc=<optimized out>, argv=<optimized out>, _internal=<optimized out>) at kernel/qapplication.cpp:501 #19 0x0000555a8536a09d in main (argc=<optimized out>, argv=<optimized out>) at /usr/src/debug/plasma-workspace-5.26.0/kcms/lookandfeel/lnftool.cpp:27
The crash makes some sense, we are trying to connect to a window system where there is none. What I don't understand is how this ever worked before. I can't see a relevant change between 5.25 and now
Using "lookandfeeltool -platform offscreen" should avoid the crash, but it could break some parts of the functionality
(In reply to Nicolas Fella from comment #2) > Using "lookandfeeltool -platform offscreen" should avoid the crash, but it > could break some parts of the functionality Yes it works and doesn't seem to break any functionality. Hooray. The darkman author should probably be notified so that he can update the example scripts. So should this bug be marked as resolved or is the crash without this flag unintended behavior?