[Mlir-commits] [mlir] Fix unsupported transpose ops for scalable vectors in LowerVectorTransfer (PR #86163)

[Benjamin Maxwell]

================
@@ -201,12 +205,19 @@ struct TransferWritePermutationLowering
     // Generate new transfer_write operation.
     Value newVec = rewriter.create<vector::TransposeOp>(
         op.getLoc(), op.getVector(), indices);
+
+    auto vectorType = cast<VectorType>(newVec.getType());
+
+    if (vectorType.isScalable() && !*vectorType.getScalableDims().end()) {
----------------
MacDue wrote:

I believe `*vectorType.getScalableDims().end()` is could be an out-of-bounds access. The `.end()` method returns an iterator passed the end of the array, I think what you probably want is: 

```suggestion
    if (vectorType.isScalable() && !vectorType.getScalableDims().back()) {
```

https://github.com/llvm/llvm-project/pull/86163