[Mesa-dev] [PROPOSAL] gallium: move state enable bits from clip_state to rasterizer_state

[Christoph Bumiller]

On 17.12.2011 15:45, Marek Olšák wrote:
> ---
>
> This was suggested by Keith Whitwell in this email addressed to me on 8/6/2010:
> http://lists.freedesktop.org/archives/mesa-dev/2010-August/001810.html
>
>
>  src/gallium/include/pipe/p_defines.h |    2 +-
>  src/gallium/include/pipe/p_state.h   |   14 ++++++++++++--
>  2 files changed, 13 insertions(+), 3 deletions(-)
>
> diff --git a/src/gallium/include/pipe/p_defines.h b/src/gallium/include/pipe/p_defines.h
> index 800a04c..c441a1f 100644
> --- a/src/gallium/include/pipe/p_defines.h
> +++ b/src/gallium/include/pipe/p_defines.h
> @@ -453,7 +453,7 @@ enum pipe_cap {
>     PIPE_CAP_TGSI_FS_COORD_ORIGIN_LOWER_LEFT = 38,
>     PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_HALF_INTEGER = 39,
>     PIPE_CAP_TGSI_FS_COORD_PIXEL_CENTER_INTEGER = 40,
> -   PIPE_CAP_DEPTH_CLAMP = 41,
> +   PIPE_CAP_DEPTH_CLIP_DISABLE = 41,
>     PIPE_CAP_SHADER_STENCIL_EXPORT = 42,
>     PIPE_CAP_TGSI_INSTANCEID = 43,
>     PIPE_CAP_VERTEX_ELEMENT_INSTANCE_DIVISOR = 44,
> diff --git a/src/gallium/include/pipe/p_state.h b/src/gallium/include/pipe/p_state.h
> index f943ca5..3aedada 100644
> --- a/src/gallium/include/pipe/p_state.h
> +++ b/src/gallium/include/pipe/p_state.h
> @@ -127,6 +127,18 @@ struct pipe_rasterizer_state
>      */
>     unsigned rasterizer_discard:1;
>  
> +   /**
> +    * When false, depth clipping is disabled and the depth value will be
> +    * clamped later at the per-pixel level before depth testing.
> +    * This depends on PIPE_CAP_DEPTH_CLIP_DISABLE.
> +    */
> +   unsigned depth_clip:1;
> +
> +   /**
> +    * Enable bits for user clip planes.
> +    */
> +   unsigned user_clip_plane_enable:PIPE_MAX_CLIP_PLANES;
> +

The first is fine, but I don't like having user clip plane enables here,
can't make them part of a hardware state buffer since whether a clip
plane is enabled or not also depends on the vertex (or domain or
geometry) shader, using both UCPs and gl_ClipDistance at the same time
doesn't work.

The rasterizer cso is already so large ... if we keep going like this at
some point you have re-validate everything when the rasterizer CSO
changes, because of interdependencies, and we'll get larger and larger
numbers of rasterizer CSOs with slow lookup and hashing.

>     unsigned line_stipple_factor:8;  /**< [1..256] actually */
>     unsigned line_stipple_pattern:16;
>  
> @@ -165,8 +177,6 @@ struct pipe_scissor_state
>  struct pipe_clip_state
>  {
>     float ucp[PIPE_MAX_CLIP_PLANES][4];
> -   unsigned nr;
> -   unsigned depth_clamp:1;
>  };
>  
>