vc4: Handle nir_intrinsic_load_user_clip_plane as a vec4.

I liked having all my NIR be scalar, but nir_validate() complains that the
intrinsic writes 4 components but the destination we set up was only 1
component.  I could generate a new scalar variant, but it's a lot easier
to just leave it as a vec4.  This doesn't hurt codegen since we GC unused
uniforms, and UCP dot products use all the components anyway.

2 files changed, 12 insertions, 20 deletions

diff --git a/src/gallium/drivers/vc4/vc4_nir_lower_io.c b/src/gallium/drivers/vc4/vc4_nir_lower_io.c
index d08ad588e5b..d0391c0262e 100644
--- a/src/gallium/drivers/vc4/vc4_nir_lower_io.c
+++ b/src/gallium/drivers/vc4/vc4_nir_lower_io.c
@@ -380,24 +380,14 @@ vc4_nir_lower_uniform(struct vc4_compile *c, nir_builder *b,
                 intr_comp->num_components = 1;
                 nir_ssa_dest_init(&intr_comp->instr, &intr_comp->dest, 1, 32, NULL);
 
-                /* Convert the uniform (not user_clip_plane) offset to bytes.
-                 * If it happens to be a constant, constant-folding will clean
-                 * up the shift for us.
+                /* Convert the uniform offset to bytes.  If it happens to be a
+                 * constant, constant-folding will clean up the shift for us.
                  */
-                if (intr->intrinsic == nir_intrinsic_load_uniform) {
-                        /* Convert the base offset to bytes and add the
-                         * component
-                         */
-                        intr_comp->const_index[0] = (intr->const_index[0] * 16 + i * 4);
-
-                        intr_comp->src[0] =
-                                nir_src_for_ssa(nir_ishl(b, intr->src[0].ssa,
-                                                         nir_imm_int(b, 4)));
-                } else {
-                        assert(intr->intrinsic ==
-                               nir_intrinsic_load_user_clip_plane);
-                        intr_comp->const_index[0] = intr->const_index[0] * 4 + i;
-                }
+                intr_comp->const_index[0] = (intr->const_index[0] * 16 + i * 4);
+
+                intr_comp->src[0] =
+                        nir_src_for_ssa(nir_ishl(b, intr->src[0].ssa,
+                                                 nir_imm_int(b, 4)));
 
                 dests[i] = &intr_comp->dest.ssa;
 
@@ -428,10 +418,10 @@ vc4_nir_lower_io_instr(struct vc4_compile *c, nir_builder *b,
                 break;
 
         case nir_intrinsic_load_uniform:
-        case nir_intrinsic_load_user_clip_plane:
                 vc4_nir_lower_uniform(c, b, intr);
                 break;
 
+        case nir_intrinsic_load_user_clip_plane:
         default:
                 break;
         }
diff --git a/src/gallium/drivers/vc4/vc4_program.c b/src/gallium/drivers/vc4/vc4_program.c
index 6a8fad939ca..fbefd217ce0 100644
--- a/src/gallium/drivers/vc4/vc4_program.c
+++ b/src/gallium/drivers/vc4/vc4_program.c
@@ -1572,8 +1572,10 @@ ntq_emit_intrinsic(struct vc4_compile *c, nir_intrinsic_instr *instr)
                 break;
 
         case nir_intrinsic_load_user_clip_plane:
-                *dest = qir_uniform(c, QUNIFORM_USER_CLIP_PLANE,
-                                    instr->const_index[0]);
+                for (int i = 0; i < instr->num_components; i++) {
+                        dest[i] = qir_uniform(c, QUNIFORM_USER_CLIP_PLANE,
+                                              instr->const_index[0] * 4 + i);
+                }
                 break;
 
         case nir_intrinsic_load_sample_mask_in: