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

[llvmlistbot at llvm.org]

================
@@ -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();
----------------
brod4910 wrote:

Thank you both for the fruitful and insightful discussion, I've learned a ton from your conversation.

Judging from @MaheshRavishankar comments, there might be other places where non-local properties are being used to verify valid IR?

If that is the case, is this something that should be changed to uphold the non-localness principle? Do you have any examples of when this happens that should be changed? I'm happy to take on the task.

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