Hello Hans, I've continued to try to achieve a working/useable DRI/X4 setup for my G400 and last night I got a combination working with my 2.4.0test5 kernel by doing the following:-
1. I downloaded the mga_drv.o driver from the Matrox web site and replaced the version that came with X4 - /usr/X11R6/lib/modules/drivers/mga_drv.o (I renamed the original mga_drv.o to mga_drv.o.old and installed the new mga_drv.o as mga_drv.o.new then created an mga_drv.o symlink to mga_drv.o.new. Also, I had to change permissions to include execute access on the new driver) 2. I renamed mga.o to mga.o.old in /lib/modules/2.4.0-test5/misc/ and replaced it with the mga.o that was created when I compiled the 2.4.0test11 kernel (from /lib/modules/2.4.0-test11/kernel/drivers/char/drm) Now, with my test5 kernel and with XF86Config-4 set to run in 16bpp, DRI initialises without errors and if I set XF86Config-4 to run in 24/32bpp I just get a DRI error about the bpp but X still starts and runs ok. Subjectively, X4/DRI in 16bpp still doesn't seem as good as the X3.3.6/utah_glx combination, performance-wise, but I don't have any s/w that I can use to reliably benchmark the two combos. One area where the older combo was definitely superior was with regard to the bpp setting - the X3.3.6/utah_glx combo would run ok in 32bpp but with X4 I have to specify 'DefaultDepth 24' and 'DefaultFbBpp 32' in XF86Config-4 to use 32bpp (according to something I read somewhere - I forget the reference now, but it's 'somewhere' on my hd:-) and this doesn't work with DRI. Yet... LeeE
