vc4: do not report alpha-test as supported

[mesa.git] / src / gallium / drivers / vc4 / vc4_program.c

diff --git a/src/gallium/drivers/vc4/vc4_program.c b/src/gallium/drivers/vc4/vc4_program.c
index 1d55b87e1efb821b27b1e2fa6397ccd45c11fa69..fc1488619892bf6bd66996ac57f6bc0be86dbea0 100644 (file)
--- a/src/gallium/drivers/vc4/vc4_program.c
+++ b/src/gallium/drivers/vc4/vc4_program.c
@@ -1530,7 +1530,7 @@ vc4_optimize_nir(struct nir_shader *s)
                 progress = false;
 
                 NIR_PASS_V(s, nir_lower_vars_to_ssa);
-                NIR_PASS(progress, s, nir_lower_alu_to_scalar, NULL);
+                NIR_PASS(progress, s, nir_lower_alu_to_scalar, NULL, NULL);
                 NIR_PASS(progress, s, nir_lower_phis_to_scalar);
                 NIR_PASS(progress, s, nir_copy_prop);
                 NIR_PASS(progress, s, nir_opt_remove_phis);
@@ -1806,11 +1806,6 @@ ntq_emit_intrinsic(struct vc4_compile *c, nir_intrinsic_instr *instr)
                                            0));
                 break;
 
-        case nir_intrinsic_load_alpha_ref_float:
-                ntq_store_dest(c, &instr->dest, 0,
-                               qir_uniform(c, QUNIFORM_ALPHA_REF, 0));
-                break;
-
         case nir_intrinsic_load_sample_mask_in:
                 ntq_store_dest(c, &instr->dest, 0,
                                qir_uniform(c, QUNIFORM_SAMPLE_MASK, 0));
@@ -2191,6 +2186,7 @@ static const nir_shader_compiler_options nir_options = {
         .lower_ldexp = true,
         .lower_negate = true,
         .lower_rotate = true,
+        .lower_to_scalar = true,
         .max_unroll_iterations = 32,
 };
 
@@ -2252,15 +2248,8 @@ vc4_shader_ntq(struct vc4_context *vc4, enum qstage stage,
 
         c->s = nir_shader_clone(c, key->shader_state->base.ir.nir);
 
-        if (stage == QSTAGE_FRAG) {
-                if (c->fs_key->alpha_test_func != COMPARE_FUNC_ALWAYS) {
-                        NIR_PASS_V(c->s, nir_lower_alpha_test,
-                                   c->fs_key->alpha_test_func,
-                                   c->fs_key->sample_alpha_to_one &&
-                                   c->fs_key->msaa);
-                }
+        if (stage == QSTAGE_FRAG)
                 NIR_PASS_V(c->s, vc4_nir_lower_blend, c);
-        }
 
         struct nir_lower_tex_options tex_options = {
                 /* We would need to implement txs, but we don't want the
@@ -2311,10 +2300,11 @@ vc4_shader_ntq(struct vc4_context *vc4, enum qstage stage,
 
         if (c->key->ucp_enables) {
                 if (stage == QSTAGE_FRAG) {
-                        NIR_PASS_V(c->s, nir_lower_clip_fs, c->key->ucp_enables);
+                        NIR_PASS_V(c->s, nir_lower_clip_fs,
+                                   c->key->ucp_enables, false);
                 } else {
                         NIR_PASS_V(c->s, nir_lower_clip_vs,
-                                  c->key->ucp_enables, false);
+                                   c->key->ucp_enables, false, false, NULL);
                         NIR_PASS_V(c->s, nir_lower_io_to_scalar,
                                    nir_var_shader_out);
                 }
@@ -2335,6 +2325,21 @@ vc4_shader_ntq(struct vc4_context *vc4, enum qstage stage,
 
         vc4_optimize_nir(c->s);
 
+        /* Do late algebraic optimization to turn add(a, neg(b)) back into
+         * subs, then the mandatory cleanup after algebraic.  Note that it may
+         * produce fnegs, and if so then we need to keep running to squash
+         * fneg(fneg(a)).
+         */
+        bool more_late_algebraic = true;
+        while (more_late_algebraic) {
+                more_late_algebraic = false;
+                NIR_PASS(more_late_algebraic, c->s, nir_opt_algebraic_late);
+                NIR_PASS_V(c->s, nir_opt_constant_folding);
+                NIR_PASS_V(c->s, nir_copy_prop);
+                NIR_PASS_V(c->s, nir_opt_dce);
+                NIR_PASS_V(c->s, nir_opt_cse);
+        }
+
         NIR_PASS_V(c->s, nir_lower_bool_to_int32);
 
         NIR_PASS_V(c->s, nir_convert_from_ssa, true);