d3d12: ensure all compoents of clip-distances are written

This fixes a regression that happened after rebasing on master, where we
end up not writing all components of the clip-distance array, which the
DXIL validation code in the D3D12 runtime treats as an error.

To ensure we don't end up overwriting a previous wrire, enable
nir_shader_compiler_options::lower_all_io_to_temps as well.

Reviewed-By: Mike Blumenkrantz <michael.blumenkrantz@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/7477>

src/gallium/drivers/d3d12/d3d12_compiler.cpp

@@ -153,6 +153,7 @@ compile_nir(struct d3d12_context *ctx, struct d3d12_shader_selector *sel,
    NIR_PASS_V(nir, d3d12_lower_load_first_vertex);
    NIR_PASS_V(nir, d3d12_lower_state_vars, shader);
    NIR_PASS_V(nir, d3d12_lower_bool_input);
+   NIR_PASS_V(nir, d3d12_fixup_clipdist_writes);
 
    struct nir_to_dxil_options opts = {};
    opts.interpolate_at_vertex = screen->have_load_at_vertex;

src/gallium/drivers/d3d12/d3d12_nir_passes.c

@@ -996,3 +996,51 @@ d3d12_lower_triangle_strip(nir_shader *shader)
    nir_metadata_preserve(impl, 0);
    NIR_PASS_V(shader, nir_lower_var_copies);
 }
+
+void
+d3d12_fixup_clipdist_writes(nir_shader *shader)
+{
+   nir_builder b;
+   nir_function_impl *impl = nir_shader_get_entrypoint(shader);
+   nir_ssa_def *primitive_id;
+   nir_builder_init(&b, impl);
+
+   nir_foreach_block(block, impl) {
+      nir_foreach_instr_safe(instr, block) {
+         if (instr->type != nir_instr_type_intrinsic)
+            continue;
+
+         nir_intrinsic_instr *intr = nir_instr_as_intrinsic(instr);
+         if (intr->intrinsic != nir_intrinsic_store_deref)
+            continue;
+
+         nir_deref_instr *deref = nir_src_as_deref(intr->src[0]);
+         nir_variable *var = nir_deref_instr_get_variable(deref);
+         if (var->data.mode != nir_var_shader_out ||
+             (var->data.location != VARYING_SLOT_CLIP_DIST0 &&
+              var->data.location != VARYING_SLOT_CLIP_DIST1))
+            continue;
+
+         uint32_t writemask = nir_intrinsic_write_mask(intr);
+         if (writemask == 0xf)
+            continue;
+
+         b.cursor = nir_before_instr(instr);
+
+         nir_ssa_def *defs[4];
+         nir_ssa_def *prev = nir_ssa_for_src(&b, intr->src[1],
+                                             nir_src_num_components(intr->src[1]));
+         for (int i = 0; i < 4; ++i) {
+            if (writemask & (1 << i))
+               defs[i] = nir_channel(&b, prev, i);
+            else
+               defs[i] = nir_imm_zero(&b, 1, nir_src_bit_size(intr->src[1]));
+         }
+         nir_ssa_def *def = nir_vec4(&b, defs[0], defs[1], defs[2], defs[3]);
+         nir_instr_rewrite_src(&intr->instr, intr->src + 1, nir_src_for_ssa(def));
+         nir_intrinsic_set_write_mask(intr, 0xf);
+      }
+   }
+
+   nir_metadata_preserve(impl, nir_metadata_all);
+}

src/gallium/drivers/d3d12/d3d12_nir_passes.h

@@ -91,6 +91,9 @@ d3d12_lower_primitive_id(nir_shader *shader);
 void
 d3d12_lower_triangle_strip(nir_shader *shader);
 
+void
+d3d12_fixup_clipdist_writes(nir_shader *shader);
+
 #ifdef __cplusplus
 }
 #endif

src/microsoft/compiler/nir_to_dxil.c

@@ -87,6 +87,7 @@ nir_options = {
    .lower_extract_word = true,
    .lower_extract_byte = true,
    .lower_all_io_to_elements = true,
+   .lower_all_io_to_temps = true,
    .lower_hadd = true,
    .lower_add_sat = true,
    .lower_uadd_carry = true,