From: Denis O. K. <do...@di...> - 2003-04-24 18:32:07
|
Hi, today I updated my tree and made the necessary changes to use the mga driver with the current embedded-1-branch version. The segfault in the sample_server I got seems to be fb_dri.so related. Running server and clients is working now, demos are working. I would consider the driver as worth being added to the official branch. My next step is to integrate fbdri into DirectFB by using the required code from miniglx.c. DirectFBGL applications don't need X or GLX. I read about the source tree reorganization and had some thoughts about how DirectFB creates the context for rendering to any DirectFB surface (e.g. the ARGB surface of a window with an alpha channel). Besides the necessary generalization of MiniGLXDisplayRec there are a few other issues which require some discussions and cooperation. DRI and DirectFB have to use the same video memory manager. But for now one could split the framebuffer into two fixed size halfs (one for DRI, one for DirectFB). DRI on DirectFB should not allocate front and back buffers. DirectFB allocates the buffers and gives them to DRI. So the frontPitch & Co setup should be moved out of the one time intialization in the ScreenInit method into a method that can be called by each client or even multiple times. Maybe I should attend the next IRC meeting. -- Best regards, Denis Oliver Kropp .------------------------------------------. | DirectFB - Hardware accelerated graphics | | http://www.directfb.org/ | "------------------------------------------" Convergence GmbH |