new radeon tiling patch

[Roland Scheidegger]

Alex Deucher wrote:
> Actually DRIAdjustframe() and friends in dri.c may still be the
> problem.  it does some wrapping of the adjustframe() functions and
> calls them, perhaps incorrectly for mergedfb.  I could be wrong though
> I haven't really traced all that is going on there.  Something is
> causing the frames to both end up at 0 when pageflipping.
Wouldn't you get errors with non-pageflip if the wrapping is wrong for 
mergedfb? The only difference with pageflip is that the drm will 
actually use the offset values, but DRIAdjustframe and RADEONAdjustFrame 
should be called exactly the same I think (unless pageflip causes some 
other effects I can't see, like for instance pDRIPriv->fullscreen to go 
true).

Roland