[PATCH 2/4] vkd3d-shader/hlsl: Handle complex types in add_cast().

[Francisco Casas]

From: Francisco Casas

This extends the support of this function, whether doing broadcasts or
component-wise casts, to struct and array types.

---

In the following patches, compatible_data_types() and
implicit_compatible_data_types() are improved so that:
* Casts that should be allowed actually reach add_cast().
* Casts that shouldn't be allowed because of special conditions (besides
  component count), e.g. matrix to matrix, don't reach add_cast().
---
 libs/vkd3d-shader/hlsl.y                      | 45 +++++++++----------
 tests/cast-broadcast.shader_test              |  4 +-
 .../cast-componentwise-compatible.shader_test | 20 ++++-----
 3 files changed, 33 insertions(+), 36 deletions(-)

diff --git a/libs/vkd3d-shader/hlsl.y b/libs/vkd3d-shader/hlsl.y
index cc0a40ff..f5f114c8 100644
--- a/libs/vkd3d-shader/hlsl.y
+++ b/libs/vkd3d-shader/hlsl.y
@@ -284,30 +284,30 @@ static struct hlsl_ir_node *add_cast(struct hlsl_ctx *ctx, struct list *instrs,
     if (hlsl_types_are_equal(src_type, dst_type))
         return node;

-    if ((src_type->type == HLSL_CLASS_MATRIX || dst_type->type == HLSL_CLASS_MATRIX)
-        && src_type->type <= HLSL_CLASS_LAST_NUMERIC && dst_type->type <= HLSL_CLASS_LAST_NUMERIC)
+    if (src_type->type > HLSL_CLASS_VECTOR || dst_type->type > HLSL_CLASS_VECTOR)
     {
+        unsigned int src_comp_count = hlsl_type_component_count(src_type);
+        unsigned int dst_comp_count = hlsl_type_component_count(dst_type);
         struct hlsl_deref var_deref;
+        bool broadcast, matrix_cast;
         struct hlsl_ir_load *load;
         struct hlsl_ir_var *var;
         unsigned int dst_idx;
-        bool broadcast;

-        broadcast = src_type->dimx == 1 && src_type->dimy == 1;
-        assert(dst_type->dimx * dst_type->dimy <= src_type->dimx * src_type->dimy || broadcast);
-        if (src_type->type == HLSL_CLASS_MATRIX && dst_type->type == HLSL_CLASS_MATRIX && !broadcast)
-        {
-            assert(dst_type->dimx <= src_type->dimx);
-            assert(dst_type->dimy <= src_type->dimy);
-        }
+        broadcast = src_type->type <= HLSL_CLASS_LAST_NUMERIC && src_type->dimx == 1 && src_type->dimy == 1;
+        matrix_cast = !broadcast && dst_comp_count != src_comp_count
+                && src_type->type == HLSL_CLASS_MATRIX && dst_type->type == HLSL_CLASS_MATRIX;
+        assert(src_comp_count >= dst_comp_count || broadcast);
+        assert(!matrix_cast || dst_type->dimx <= src_type->dimx);
+        assert(!matrix_cast || dst_type->dimy <= src_type->dimy);

if (!(var = hlsl_new_synthetic_var(ctx, "cast", dst_type, loc)))
             return NULL;
         hlsl_init_simple_deref_from_var(&var_deref, var);

-        for (dst_idx = 0; dst_idx < dst_type->dimx * dst_type->dimy; ++dst_idx)
+        for (dst_idx = 0; dst_idx < dst_comp_count; ++dst_idx)
         {
-            struct hlsl_type *dst_scalar_type;
+            struct hlsl_type *dst_comp_type;
             struct hlsl_ir_store *store;
             struct hlsl_block block;
             unsigned int src_idx;
@@ -316,26 +316,23 @@ static struct hlsl_ir_node *add_cast(struct hlsl_ctx *ctx, struct list *instrs,
             {
                 src_idx = 0;
             }
-            else
+            else if (matrix_cast)
             {
-                if (src_type->type == HLSL_CLASS_MATRIX && dst_type->type == HLSL_CLASS_MATRIX)
-                {
-                    unsigned int x = dst_idx % dst_type->dimx, y = dst_idx / dst_type->dimx;
+                unsigned int x = dst_idx % dst_type->dimx, y = dst_idx / dst_type->dimx;

-                    src_idx = y * src_type->dimx + x;
-                }
-                else
-                {
-                    src_idx = dst_idx;
-                }
+                src_idx = y * src_type->dimx + x;
+            }
+            else
+            {
+                src_idx = dst_idx;
             }

-            dst_scalar_type = hlsl_type_get_component_type(ctx, dst_type, dst_idx);
+            dst_comp_type = hlsl_type_get_component_type(ctx, dst_type, dst_idx);

if (!(load = add_load_component(ctx, instrs, node, src_idx, loc)))
                 return NULL;

-            if (!(cast = hlsl_new_cast(ctx, &load->node, dst_scalar_type, loc)))
+            if (!(cast = hlsl_new_cast(ctx, &load->node, dst_comp_type, loc)))
                 return NULL;
             list_add_tail(instrs, &cast->node.entry);

diff --git a/tests/cast-broadcast.shader_test b/tests/cast-broadcast.shader_test
index e21e65b0..d8571fc8 100644
--- a/tests/cast-broadcast.shader_test
+++ b/tests/cast-broadcast.shader_test
@@ -19,8 +19,8 @@ float4 main() : SV_TARGET
 }

[test]
-todo draw quad
-todo probe all rgba (84.0, 84.0, 84.0, 84.0)
+draw quad
+probe all rgba (84.0, 84.0, 84.0, 84.0)

[pixel shader fail]
diff --git a/tests/cast-componentwise-compatible.shader_test b/tests/cast-componentwise-compatible.shader_test
index c6402cf5..286c180b 100644
--- a/tests/cast-componentwise-compatible.shader_test
+++ b/tests/cast-componentwise-compatible.shader_test
@@ -16,8 +16,8 @@ float4 main() : sv_target

[test]
-todo draw quad
-todo probe all rgba (1.0, 2.0, 3.0, 1.0)
+draw quad
+probe all rgba (1.0, 2.0, 3.0, 1.0)

[pixel shader]
@@ -39,8 +39,8 @@ float4 main() : sv_target

[test]
-todo draw quad
-todo probe all rgba (5.0, 6.0, 7.0, 8.0)
+draw quad
+probe all rgba (5.0, 6.0, 7.0, 8.0)

[pixel shader]
@@ -55,8 +55,8 @@ float4 main() : sv_target

[test]
-todo draw quad
-todo probe all rgba (1.0, 2.0, 3.0, 4.0)
+draw quad
+probe all rgba (1.0, 2.0, 3.0, 4.0)

[pixel shader]
@@ -86,8 +86,8 @@ float4 main() : sv_target
 }

[test]
-todo draw quad
-todo probe all rgba (7.0, 7.0, 7.0, 7.0)
+draw quad
+probe all rgba (7.0, 7.0, 7.0, 7.0)

[pixel shader]
@@ -119,8 +119,8 @@ float4 main() : sv_target
 }

[test]
-todo draw quad
-todo probe all rgba (3.0, 3.0, 3.0, 3.0)
+draw quad
+probe all rgba (3.0, 3.0, 3.0, 3.0)

[pixel shader fail]