[Mlir-commits] [mlir] [MLIR][AMDGPU]Add refactoring for shared-mem optimization (PR #81791)

[Oleksandr Alex Zinenko]

================
@@ -218,7 +218,8 @@ mlir::amdgpu::optimizeSharedMemoryReadsAndWrites(Operation *parentOp,
   return success();
 }
 
-void amdgpu::optimizeSharedMemoryReadsAndWritesOp(func::FuncOp funcOp) {
+std::optional<mlir::LogicalResult>
----------------
ftynse wrote:

I don't see why the result is optional, nothing returns `nullopt` here.

https://github.com/llvm/llvm-project/pull/81791