https://bugs.kde.org/show_bug.cgi?id=357032
Thomas Lübking <thomas.luebk...@gmail.com> changed: What |Removed |Added ---------------------------------------------------------------------------- Status|NEEDSINFO |UNCONFIRMED Resolution|BACKTRACE |--- --- Comment #3 from Thomas Lübking <thomas.luebk...@gmail.com> --- Bad line is here: /usr/src/debug/kwin-5.5.90git~20151218T172621~ecce036/scene.cpp:302 The stacking_order will contain a dangeling entry what implies it gets deleted during the paint phase - what should not happen. Can you please a) attach the output of "qdbus org.kde.KWin /KWin supportInformation" b) apply this patch diff --git a/scene.cpp b/scene.cpp index a25dc8d..cebcbe9 100644 --- a/scene.cpp +++ b/scene.cpp @@ -421,6 +421,7 @@ void Scene::windowClosed(Toplevel *c, Deleted *deleted) } m_windows[ deleted ] = w; } else { + assert(stacking_order.isEmpty()); delete m_windows.take(c); c->effectWindow()->setSceneWindow(NULL); } @@ -428,6 +429,7 @@ void Scene::windowClosed(Toplevel *c, Deleted *deleted) void Scene::windowDeleted(Deleted *c) { + assert(stacking_order.isEmpty()); assert(m_windows.contains(c)); delete m_windows.take(c); c->effectWindow()->setSceneWindow(NULL); and see whether you hit the assert (and on what backtrace) The hanging instance is likely stopped by drkonqi, see bug #353428 - there's a *possible* patch to this in comment #33 -- You are receiving this mail because: You are watching all bug changes.