[Mlir-commits] [mlir] [mlir][bufferization] Allow returning allocs in empty tensor elimination (PR #66198)

[llvmlistbot at llvm.org]

llvmbot wrote:


<!--LLVM PR SUMMARY COMMENT-->

@llvm/pr-subscribers-mlir-bufferization
            
<details>
<summary>Changes</summary>
This flag was renamed in 6a91dfedeb956dfa092a6a3f411e8b02f0d5d289 and accidentally removed from the pass.
--
Full diff: https://github.com/llvm/llvm-project/pull/66198.diff

1 Files Affected:

- (modified) mlir/lib/Dialect/Bufferization/Transforms/EmptyTensorElimination.cpp (+1) 


<pre>
diff --git a/mlir/lib/Dialect/Bufferization/Transforms/EmptyTensorElimination.cpp b/mlir/lib/Dialect/Bufferization/Transforms/EmptyTensorElimination.cpp
index 4dbf22593b911ae..d2e3f512ff30c66 100644
--- a/mlir/lib/Dialect/Bufferization/Transforms/EmptyTensorElimination.cpp
+++ b/mlir/lib/Dialect/Bufferization/Transforms/EmptyTensorElimination.cpp
@@ -268,6 +268,7 @@ struct EmptyTensorElimination
 void EmptyTensorElimination::runOnOperation() {
   Operation *op = getOperation();
   OneShotBufferizationOptions options;
+  options.allowReturnAllocsFromLoops = true;
   OneShotAnalysisState state(op, options);
   if (failed(analyzeOp(op, state))) {
     signalPassFailure();
</pre>
</details>


https://github.com/llvm/llvm-project/pull/66198