[Mesa-dev] [RFC 18/24] nvc0/ir: fix 1D arrays images for Kepler

[Ilia Mirkin]

On Tue, Apr 12, 2016 at 7:57 PM, Samuel Pitoiset
<samuel.pitoiset at gmail.com> wrote:
> For 1D arrays, the array index is stored in the Z component.
>
> Signed-off-by: Samuel Pitoiset <samuel.pitoiset at gmail.com>
> ---
>  .../drivers/nouveau/codegen/nv50_ir_lowering_nvc0.cpp   | 17 +++++++++++++----
>  1 file changed, 13 insertions(+), 4 deletions(-)
>
> diff --git a/src/gallium/drivers/nouveau/codegen/nv50_ir_lowering_nvc0.cpp b/src/gallium/drivers/nouveau/codegen/nv50_ir_lowering_nvc0.cpp
> index 0ef0939..0a3daca 100644
> --- a/src/gallium/drivers/nouveau/codegen/nv50_ir_lowering_nvc0.cpp
> +++ b/src/gallium/drivers/nouveau/codegen/nv50_ir_lowering_nvc0.cpp
> @@ -1614,13 +1614,22 @@ NVC0LoweringPass::processSurfaceCoordsNVE4(TexInstruction *su)
>
>     // calculate clamped coordinates
>     for (c = 0; c < arg; ++c) {
> +      int subOp = getSuClampSubOp(su, c);
> +
>        src[c] = bld.getScratch();
> -      if (c == 0 && raw)
> +      if (c == 0 && raw) {
>           v = loadSuInfo32(NULL, base + NVE4_SU_INFO_RAW_X);
> -      else
> -         v = loadSuInfo32(NULL, base + NVE4_SU_INFO_DIM(c));
> +      } else {
> +         if (c == 1 && su->tex.target == TEX_TARGET_1D_ARRAY) {
> +            // The array index is stored in the Z component for 1D arrays.
> +            v = loadSuInfo32(NULL, base + NVE4_SU_INFO_DIM(2));
> +            subOp = getSuClampSubOp(su, 2);

This seems complicated... how about just doing

int dimc = c;
if (c == 1 && 1d array)
  dimc = 2;

And then you can avoid having the duplicate loadSuInfo logic and the
duplicate suclampsubop call.

> +         } else {
> +            v = loadSuInfo32(NULL, base + NVE4_SU_INFO_DIM(c));
> +         }
> +      }
>        bld.mkOp3(OP_SUCLAMP, TYPE_S32, src[c], su->getSrc(c), v, zero)
> -         ->subOp = getSuClampSubOp(su, c);
> +         ->subOp = subOp;
>     }
>     for (; c < 3; ++c)
>        src[c] = zero;
> --
> 2.8.0
>
> _______________________________________________
> mesa-dev mailing list
> mesa-dev at lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/mesa-dev