[Mlir-commits] [mlir] [MLIR][Tensor] Fix out-of-bounds FoldEmptyTensorWithDimOp crash #111270 (PR #112196)

[Mehdi Amini]

================
@@ -979,7 +979,9 @@ struct FoldEmptyTensorWithDimOp : public OpRewritePattern<DimOp> {
     auto emptyTensorOp = dimOp.getSource().getDefiningOp<EmptyOp>();
     if (!emptyTensorOp || !maybeConstantIndex)
       return failure();
-    if (!emptyTensorOp.getType().isDynamicDim(*maybeConstantIndex))
+    auto emptyTensorType = emptyTensorOp.getType();
----------------
joker-eph wrote:

Of course, but the current consensus and principle exists for now and so this is what we live with at the moment.. Feel free to challenge it but you have to run through things through first principles, and present strong argument and examples to reverse such policy.


https://github.com/llvm/llvm-project/pull/112196