nir: Fix interaction of GL_CLAMP lowering with texture offsets.

We want the clamping of the coordinate to apply after the offset, so we
need to do math to lower the offset out of the instruction.  Fixes texwrap
offset cases for GL_CLAMP with GL_NEAREST on vc5.

Note: I moved the get_texture_size() verbatim, so that it was defined
before use.

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

diff --git a/src/compiler/nir/nir_lower_tex.c b/src/compiler/nir/nir_lower_tex.c
index a50cb52..bd603cc 100644 (file)
--- a/src/compiler/nir/nir_lower_tex.c
+++ b/src/compiler/nir/nir_lower_tex.c
@@ -101,6 +101,36 @@ project_src(nir_builder *b, nir_tex_instr *tex)
    nir_tex_instr_remove_src(tex, proj_index);
 }
 
+static nir_ssa_def *
+get_texture_size(nir_builder *b, nir_tex_instr *tex)
+{
+   b->cursor = nir_before_instr(&tex->instr);
+
+   nir_tex_instr *txs;
+
+   txs = nir_tex_instr_create(b->shader, 1);
+   txs->op = nir_texop_txs;
+   txs->sampler_dim = tex->sampler_dim;
+   txs->is_array = tex->is_array;
+   txs->is_shadow = tex->is_shadow;
+   txs->is_new_style_shadow = tex->is_new_style_shadow;
+   txs->texture_index = tex->texture_index;
+   txs->texture = nir_deref_var_clone(tex->texture, txs);
+   txs->sampler_index = tex->sampler_index;
+   txs->sampler = nir_deref_var_clone(tex->sampler, txs);
+   txs->dest_type = nir_type_int;
+
+   /* only single src, the lod: */
+   txs->src[0].src = nir_src_for_ssa(nir_imm_int(b, 0));
+   txs->src[0].src_type = nir_tex_src_lod;
+
+   nir_ssa_dest_init(&txs->instr, &txs->dest,
+                     nir_tex_instr_dest_size(txs), 32, NULL);
+   nir_builder_instr_insert(b, &txs->instr);
+
+   return nir_i2f32(b, &txs->dest.ssa);
+}
+
 static bool
 lower_offset(nir_builder *b, nir_tex_instr *tex)
 {
@@ -120,8 +150,17 @@ lower_offset(nir_builder *b, nir_tex_instr *tex)
 
    nir_ssa_def *offset_coord;
    if (nir_tex_instr_src_type(tex, coord_index) == nir_type_float) {
-      assert(tex->sampler_dim == GLSL_SAMPLER_DIM_RECT);
-      offset_coord = nir_fadd(b, coord, nir_i2f32(b, offset));
+      if (tex->sampler_dim == GLSL_SAMPLER_DIM_RECT) {
+         offset_coord = nir_fadd(b, coord, nir_i2f32(b, offset));
+      } else {
+         nir_ssa_def *txs = get_texture_size(b, tex);
+         nir_ssa_def *scale = nir_frcp(b, txs);
+
+         offset_coord = nir_fadd(b, coord,
+                                 nir_fmul(b,
+                                          nir_i2f32(b, offset),
+                                          scale));
+      }
    } else {
       offset_coord = nir_iadd(b, coord, offset);
    }
@@ -148,37 +187,6 @@ lower_offset(nir_builder *b, nir_tex_instr *tex)
    return true;
 }
 
-
-static nir_ssa_def *
-get_texture_size(nir_builder *b, nir_tex_instr *tex)
-{
-   b->cursor = nir_before_instr(&tex->instr);
-
-   nir_tex_instr *txs;
-
-   txs = nir_tex_instr_create(b->shader, 1);
-   txs->op = nir_texop_txs;
-   txs->sampler_dim = tex->sampler_dim;
-   txs->is_array = tex->is_array;
-   txs->is_shadow = tex->is_shadow;
-   txs->is_new_style_shadow = tex->is_new_style_shadow;
-   txs->texture_index = tex->texture_index;
-   txs->texture = nir_deref_var_clone(tex->texture, txs);
-   txs->sampler_index = tex->sampler_index;
-   txs->sampler = nir_deref_var_clone(tex->sampler, txs);
-   txs->dest_type = nir_type_int;
-
-   /* only single src, the lod: */
-   txs->src[0].src = nir_src_for_ssa(nir_imm_int(b, 0));
-   txs->src[0].src_type = nir_tex_src_lod;
-
-   nir_ssa_dest_init(&txs->instr, &txs->dest,
-                     nir_tex_instr_dest_size(txs), 32, NULL);
-   nir_builder_instr_insert(b, &txs->instr);
-
-   return nir_i2f32(b, &txs->dest.ssa);
-}
-
 static void
 lower_rect(nir_builder *b, nir_tex_instr *tex)
 {
@@ -749,6 +757,7 @@ nir_lower_tex_block(nir_block *block, nir_builder *b,
       }
 
       if ((tex->op == nir_texop_txf && options->lower_txf_offset) ||
+          (sat_mask && nir_tex_instr_src_index(tex, nir_tex_src_coord) >= 0) ||
           (tex->sampler_dim == GLSL_SAMPLER_DIM_RECT &&
            options->lower_rect_offset)) {
          progress = lower_offset(b, tex) || progress;