> I also added the PCI > device ID (0x3118) of my graphics device to via_fbobj.c although > I'm not sure it really is supported). I've attached a patch to > linux-viafb showing my changes. > > Running the DirectFB examples, everything looks pretty good.
If you're able to run the examples, then the kernel framebuffer is doing everything it needs to do so adding your PCI ID seems reasonable. > But when I try to play back video, using e.g. > df_xine, it looks like the wrong colourspace gets used. The > image is stretched to fit the screen vertically, but horizontally > it only occupies the leftmost third of the screen. The rest of > the screen contains garbage. Also the colours are incorrect. OK. df_xine by default will use the video overlay. All the control of that is done by the DirectFB driver and not the kernel framebuffer. It's likely that there's a difference between the CLE266 and the CN400 that isn't taken into account. Out of interest, try df_xine -l0 (to force use of the primary layer) and also df_xine -l1 -p RGB32 (to use the overlay but with an RGB pixel format). > Any ideas? Is this because some features are simply missing > from the unichrome driver, or is it just that the driver doesn't > work with the newer Unichrome devices? It's the latter but it's the DirectFB driver that needs some work. Regards, Mark _______________________________________________ directfb-users mailing list [email protected] http://mail.directfb.org/cgi-bin/mailman/listinfo/directfb-users
