[Mesa-dev] Radeon DRI1 removal

[Michel Dänzer]

On Mit, 2011-10-19 at 17:50 -0700, Eric Anholt wrote: 
> For the MapRenderbuffer() code (See
> http://cgit.freedesktop.org/~anholt/mesa/log/?h=rbmap for the current
> work, which has glReadPixels() converted so far and tested on gen6 with
> only one regression in the tip commit), I need to be able to provide a
> linear mapping of arbitrary renderbuffers.  Radeon's legacy DRI1 support
> gets in the way of that -- I need to use the blitter to a temporary to
> do untiling (which means holding the hw lock across map/unmap cycles,
> and thus refcounting the lock), and I need to do clipping on read and
> write.
> 
> I think it would be much better to just ditch DRI1 at this point, since
> KMS and DRI2 are quite well established, and are relied on by shipping
> desktop environments.
> 
> Code to do so is at
> http://cgit.freedesktop.org/~anholt/mesa/log/?h=radeon-kill-dri1 which
> is tested on r300 only currently.  No piglit regressions there, though I
> do note that copyteximage is unreliable both before and after (rendering
> always looks good to my eyes, but readback apparently gets junk).
> 
> So, Radeon maintainers, what do you think?  And, does anyone else want
> to test it on the other drivers?  I caught two bugs in my r300 testing
> (one too many lines cut in one r300 commit, and I had also removed the
> texsubimage code in an unrelated-to-the-series commit, but it regressed
> s3tc-texsubimage so I backed it out).

I don't feel too strongly about it either way, but if you're going to do
this, you might as well remove the classic r300 and r600 drivers
altogether in favour of the Gallium based ones.


-- 
Earthling Michel Dänzer           |                   http://www.amd.com
Libre software enthusiast         |          Debian, X and DRI developer