panfrost/midgard: Extend copy propagation pass

This extends copy propagation to respect output modifiers for ALU
instructions, as well as potentially fixing some bugs related to looping
(all dEQP loop tests pass).

Signed-off-by: Alyssa Rosenzweig <alyssa@rosenzweig.io>

diff --git a/src/gallium/drivers/panfrost/midgard/midgard_compile.c b/src/gallium/drivers/panfrost/midgard/midgard_compile.c
index 5f62225de17d7a6035ba42587a5904e06f3f8e29..5b5a44013a2c78ee504e1b52313e8090514d24ea 100644 (file)
--- a/src/gallium/drivers/panfrost/midgard/midgard_compile.c
+++ b/src/gallium/drivers/panfrost/midgard/midgard_compile.c
@@ -3123,6 +3123,36 @@ midgard_opt_dead_code_eliminate(compiler_context *ctx, midgard_block *block)
         return progress;
 }
 
+/* Combines the two outmods if possible. Returns whether the combination was
+ * successful */
+
+static bool
+midgard_combine_outmod(midgard_outmod *main, midgard_outmod overlay)
+{
+        if (overlay == midgard_outmod_none)
+                return true;
+
+        if (*main == overlay)
+                return true;
+
+        if (*main == midgard_outmod_none) {
+                *main = overlay;
+                return true;
+        }
+
+        if (*main == midgard_outmod_pos && overlay == midgard_outmod_sat) {
+                *main = midgard_outmod_sat;
+                return true;
+        }
+
+        if (overlay == midgard_outmod_pos && *main == midgard_outmod_sat) {
+                *main = midgard_outmod_sat;
+                return true;
+        }
+
+        return false;
+}
+
 static bool
 midgard_opt_copy_prop(compiler_context *ctx, midgard_block *block)
 {
@@ -3139,8 +3169,12 @@ midgard_opt_copy_prop(compiler_context *ctx, midgard_block *block)
 
                 if (to >= SSA_FIXED_MINIMUM) continue;
                 if (from >= SSA_FIXED_MINIMUM) continue;
+                if (to >= ctx->func->impl->ssa_alloc) continue;
+                if (from >= ctx->func->impl->ssa_alloc) continue;
 
-                /* Also, if the move has side effects, we're helpless */
+                /* Also, if the move has source side effects, we're not sure
+                 * what to do. Destination side effects we can handle, though.
+                 */
 
                 midgard_vector_alu_src src =
                         vector_alu_from_unsigned(ins->alu.src2);
@@ -3149,17 +3183,23 @@ midgard_opt_copy_prop(compiler_context *ctx, midgard_block *block)
 
                 if (mir_nontrivial_mod(src, is_int, mask)) continue;
 
-                mir_foreach_instr_in_block_from(block, v, mir_next_op(ins)) {
-                        if (v->ssa_args.src0 == to) {
-                                v->ssa_args.src0 = from;
-                                progress = true;
-                        }
+                mir_foreach_instr_in_block_from_rev(block, v, mir_prev_op(ins)) {
+                        if (v->ssa_args.dest == from) {
+                                if (v->type == TAG_ALU_4) {
+                                        midgard_outmod final = v->alu.outmod;
 
-                        if (v->ssa_args.src1 == to && !v->ssa_args.inline_constant) {
-                                v->ssa_args.src1 = from;
+                                        if (!midgard_combine_outmod(&final, ins->alu.outmod))
+                                                continue;
+
+                                        v->alu.outmod = final;
+                                }
+
+                                v->ssa_args.dest = to;
                                 progress = true;
                         }
                 }
+
+                mir_remove_instruction(ins);
         }
 
         return progress;