[clang] [llvm] [AMDGPU] Add an async barrier arrival workaround (PR #238539)
Pierre van Houtryve via cfe-commits
cfe-commits at lists.llvm.org
Thu Nov 5 00:07:27 PST 2026
================
@@ -0,0 +1,104 @@
+//===- SIInsertAsyncBarrierArriveWorkaround.cpp ---------------------------===//
+//
+// Part of the LLVM Project, under the Apache License v2.0 with LLVM Exceptions.
+// See https://llvm.org/LICENSE.txt for license information.
+// SPDX-License-Identifier: Apache-2.0 WITH LLVM-exception
+//
+//===----------------------------------------------------------------------===//
+//
+/// \file
+/// Insert the async barrier arrival replay workaround after scheduling and
+/// before wait-count analysis so the dummy stays immediately after its arrival
+/// and its async counter effects are included in wait generation.
----------------
Pierre-vh wrote:
Elaborate a bit more, add an example if possible.
https://github.com/llvm/llvm-project/pull/238539
More information about the cfe-commits
mailing list