https://bugs.kde.org/show_bug.cgi?id=445561
--- Comment #16 from Dmitry Kazakov <dimul...@gmail.com> --- Hi, Ivan! Here is one more test for you. Please apply the following patch and try to load the failing image on M1: https://invent.kde.org/-/snippets/1983 The patch does two things: 1) It adds asserts to all the buffer-related calls. If the calls fails, then it will dump debugging info an crash. 2) It adds asserts to sanity check the current openGL context when creating buffers and when accessing them. That sounds as a quite possible cause of the problem. It might be that the buffers are allocated in a different openGL context, therefore they are inaccessible when we really need. If you happen to trip over any of these asserts, please also check what context is enabled instead of KisOpenGLCnavas2::context() and whether this context belongs to the same QOpenGLContext::shareGroup() as the canvas' context. PS: Btw, could you also check if openGL contexts do actually support sharing and add the following line into the end of KisOpenGLCanvas2::initializeGL(): ENTER_FUNCTION() << this << ppVar(context()->shareGroup()->shares()); There should be at least two contexts, the canvas one and the Qt's one. -- You are receiving this mail because: You are watching all bug changes.
