On 04/11/2016 04:31 PM, Rob Clark wrote: > On Mon, Apr 11, 2016 at 10:09 AM, Emil Velikov <emil.l.veli...@gmail.com> > wrote: >> Hi Rob, >> >> On 31 March 2016 at 04:21, Rob Herring <r...@kernel.org> wrote: >>> This adds GBM map and unmap functions. Unlike prior attempts, this >>> version provides a region and usage flags for the mapping. The operation >>> follows gallium transfer_map() function. This is complete enough to work >>> on Android with gralloc using the GBM API. >>> >>> TODO: >>> - split up patches >>> - define and propagate GBM usage flags >>> - DRIImage extension version checking >>> - Need to flush context on unmap? >>> - Need fences for gralloc lockAsync >>> - integrate dumb buffer mapping >>> - bikeshed the name (map or lock) >>> >> With the risk of starting a war - wasn't there objections about having >> such interfaces in GBM a while back ? >> Personally I would CC the person so that he can clarify if current >> version is in better shape. Hopefully he'll bring forward what needs >> fixing (if any) to get this kind of interface in acceptable shape. >> >> Note: not sure who was against the previous patches so I haven't CC'd him. > iirc, it was Thomas (who I've added in CC).. and the issue was more > about how the previous implementation worked (ie. returning mmap > pointer to entire buffer). Rob's implementation fits on top of > pipe_transfer and should address Thomas's concerns. > >
As long as there is an "affected region" attached to the map I'm completely OK with it. Thanks, /Thomas _______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org https://lists.freedesktop.org/mailman/listinfo/mesa-dev
