From b70ecfa58872abfa17b56d0b2486ed3df982fb7f Mon Sep 17 00:00:00 2001
From: Rhys Perry <pendingchaos02@gmail.com>
Date: Fri, 2 May 2025 16:01:05 +0100
Subject: [PATCH] aco: only join barrier_imm/barrier_events for logical edges

fossil-db (gfx1201):
Totals from 3 (0.00% of 79653) affected shaders:
Instrs: 2904 -> 2893 (-0.38%)
CodeSize: 14944 -> 14900 (-0.29%)
Latency: 14703 -> 14248 (-3.09%)
InvThroughput: 1237 -> 1210 (-2.18%)

fossil-db (navi31):
Totals from 3 (0.00% of 79653) affected shaders:
Instrs: 2742 -> 2731 (-0.40%)
CodeSize: 14136 -> 14092 (-0.31%)
Latency: 14744 -> 14287 (-3.10%)
InvThroughput: 1241 -> 1213 (-2.26%)

fossil-db (navi21):
Totals from 3 (0.00% of 79653) affected shaders:
Instrs: 2326 -> 2315 (-0.47%)
CodeSize: 12472 -> 12428 (-0.35%)
Latency: 14921 -> 14465 (-3.06%)

Signed-off-by: Rhys Perry <pendingchaos02@gmail.com>
Reviewed-by: Georg Lehmann <dadschoorse@gmail.com>
Part-of: <https://gitlab.freedesktop.org/mesa/mesa/-/merge_requests/34978>
---
 src/amd/compiler/aco_insert_waitcnt.cpp | 10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)

diff --git a/src/amd/compiler/aco_insert_waitcnt.cpp b/src/amd/compiler/aco_insert_waitcnt.cpp
index 2ebc80fb20f..025e8f14290 100644
--- a/src/amd/compiler/aco_insert_waitcnt.cpp
+++ b/src/amd/compiler/aco_insert_waitcnt.cpp
@@ -208,10 +208,12 @@ struct wait_ctx {
          }
       }
 
-      for (unsigned i = 0; i < storage_count; i++) {
-         changed |= barrier_imm[i].combine(other->barrier_imm[i]);
-         changed |= (other->barrier_events[i] & ~barrier_events[i]) != 0;
-         barrier_events[i] |= other->barrier_events[i];
+      if (logical) {
+         for (unsigned i = 0; i < storage_count; i++) {
+            changed |= barrier_imm[i].combine(other->barrier_imm[i]);
+            changed |= (other->barrier_events[i] & ~barrier_events[i]) != 0;
+            barrier_events[i] |= other->barrier_events[i];
+         }
       }
 
       return changed;