I just committed changes to the fbdev target which deserve some testing. Basically, they replace the hard-coded table that maps fbdev accel id's to accelerator sublibs with statements in the targets/fbdev.conf file. This will allow, say, matrox users to choose between using directfb or native GGI drivers using the config file. Works for me (tm), but I tested only 16bpp on my system. Especially 4bpp modes should be tested on the DFB accelerated chipsets, to see if they need a fb-generic-xx-linear-4 line in the config file like the matrox chipsets do. Also, since the autoconf issues are fixed I turned on fbdev by default, and for now all matrox use native drivers unless you change the config file, in which case you should be aware that the matrox SGRAM feature is enabled in that driver (will move that to use an option later). -- Brian
