[Mesa-dev] [PATCH 00/15] ARB_texture_view for i965

Chris Forbes chrisf at ijw.co.nz
Tue Jan 21 02:42:57 PST 2014


Oops, s/96/48/ and s/128/64/ in the description of the first patch.

On Tue, Jan 21, 2014 at 11:34 PM, Chris Forbes <chrisf at ijw.co.nz> wrote:
> This series adds support for ARB_texture_view on Ivybridge and Haswell.
>
> Since there's lots of little changes all over the place, here's a summary:
>
> [PATCH 01/15] i965: Pretend we don't support
>    - Fixes an issue where we could support one of the 96-bit formats directly,
>      but not its friends in the same view class. Since we need to be able to
>      alias the memory, just promote all the formats in this view class to 128 bits.
>
> [PATCH 02/15] i965: Include #slices in miptree debug
>    - Just a bit of additional debug.
>
> [PATCH 03/15] i965: Skip texture validation for immutable-format
> [PATCH 04/15] i965: Add driver hook for TextureView
> [PATCH 05/15] i965: Adjust surface_state emission to account for
>    - Basic support for views. This is enough to be able to texture from
>      a view whose data was provided through the underlying texture.
>
> [PATCH 06/15] i965: Don't try to use fast upload path for nontrivial
> [PATCH 07/15] i965: Adjust map/unmap code for MinLevel/MinLayer
> [PATCH 08/15] i965: Account for MinLevel in texture upload fast path
> [PATCH 09/15] i965: Enable texture upload fast path with MinLevel
>    - Support for uploading texture data through the view.
>
> [PATCH 10/15] i965: Adjust renderbuffer wrapper to account for
> [PATCH 11/15] i965: Add `layered` parameter to
> [PATCH 12/15] i965: Pull out layer_multiplier in
> [PATCH 13/15] i965: Add layer_count to intel_renderbuffer
> [PATCH 14/15] i965/blorp: Use irb->layer_count in clear
> [PATCH 15/15] i965/blorp: Account for nonzero MinLayer in layered
>    - Support for rendering & clears into a view.
>
>
> Things that I *know* don't work properly yet:
>    - fast depth clears with nonzero MinLevel or MinLayer on the depth buffer
>    - meta clears in the same cases
>
> There are probably more, but I think it's at a point where I'm ready for people to see it :)


More information about the mesa-dev mailing list