[Mesa-dev] [PATCH 3/5] i965: Add support for gl_SkipComponents[1234].

[Ian Romanick]

On 10/26/2013 01:33 PM, Kenneth Graunke wrote:
> ARB_transform_feedback3 allows applications to insert blank space
> between interleaved varyings by adding fake 1, 2, 3, or 4-component
> varyings named gl_SkipComponents[1234].
> 
> Mesa's core data structures don't explicitly track these, instead simply
> tracking the buffer offset for each real varying.  If there is padding
> due to gl_SkipComponents, these will not be contiguous.
> 
> Our hardware takes the specification quite literally.  Instead of
> specifying offsets for each varying, it assumes they're all contiguous
> and requires you to program fake varyings for each "hole".
> 
> This patch adds support for emitting SO_DECL structures for these holes.
> Although we've lost the information about exactly how the application
> specified their padding (i.e. gl_SkipComponents2, gl_SkipComponents2
> vs. a single gl_SkipComponents4), it shouldn't matter.  We just need to
> emit the right amount of space.  This patch emits the minimal number of
> hole SO_DECL structures.
> 
> Signed-off-by: Kenneth Graunke <kenneth at whitecape.org>

Reviewed-by: Ian Romanick <ian.d.romanick at intel.com>

> ---
>  src/mesa/drivers/dri/i965/gen7_sol_state.c | 23 +++++++++++++++++++++--
>  1 file changed, 21 insertions(+), 2 deletions(-)
> 
> diff --git a/src/mesa/drivers/dri/i965/gen7_sol_state.c b/src/mesa/drivers/dri/i965/gen7_sol_state.c
> index b9debdb..6be658f 100644
> --- a/src/mesa/drivers/dri/i965/gen7_sol_state.c
> +++ b/src/mesa/drivers/dri/i965/gen7_sol_state.c
> @@ -140,9 +140,28 @@ gen7_upload_3dstate_so_decl_list(struct brw_context *brw,
>  	 SO_DECL_REGISTER_INDEX_SHIFT;
>        decl |= component_mask << SO_DECL_COMPONENT_MASK_SHIFT;
>  
> -      /* This assert should be true until GL_ARB_transform_feedback_instanced
> -       * is added and we start using the hole flag.
> +      /* Mesa doesn't store entries for gl_SkipComponents in the Outputs[]
> +       * array.  Instead, it simply increments DstOffset for the following
> +       * input by the number of components that should be skipped.
> +       *
> +       * Our hardware is unusual in that it requires us to program SO_DECLs
> +       * for fake "hole" components, rather than simply taking the offset
> +       * for each real varying.  Each hole can have size 1, 2, 3, or 4; we
> +       * program as many size = 4 holes as we can, then a final hole to
> +       * accomodate the final 1, 2, or 3 remaining.
>         */
> +      int skip_components =
> +         linked_xfb_info->Outputs[i].DstOffset - next_offset[buffer];
> +
> +      next_offset[buffer] += skip_components;
> +
> +      while (skip_components >= 4) {
> +         so_decl[decls++] = SO_DECL_HOLE_FLAG | 0xf;
> +         skip_components -= 4;
> +      }
> +      if (skip_components > 0)
> +         so_decl[decls++] = SO_DECL_HOLE_FLAG | ((1 << skip_components) - 1);
> +
>        assert(linked_xfb_info->Outputs[i].DstOffset == next_offset[buffer]);
>  
>        next_offset[buffer] += components;
>