[Mlir-commits] [mlir] [mlir][vector] Fix parser of vector.transfer_read (PR #133721)

[Andrzej Warzyński]

================
@@ -152,18 +152,22 @@ static bool isSupportedCombiningKind(CombiningKind combiningKind,
 
 AffineMap mlir::vector::getTransferMinorIdentityMap(ShapedType shapedType,
                                                     VectorType vectorType) {
-  int64_t elementVectorRank = 0;
-  VectorType elementVectorType =
-      llvm::dyn_cast<VectorType>(shapedType.getElementType());
-  if (elementVectorType)
-    elementVectorRank += elementVectorType.getRank();
   // 0-d transfers are to/from tensor<t>/memref<t> and vector<1xt>.
   // TODO: replace once we have 0-d vectors.
   if (shapedType.getRank() == 0 &&
       vectorType.getShape() == ArrayRef<int64_t>{1})
     return AffineMap::get(
         /*numDims=*/0, /*numSymbols=*/0,
         getAffineConstantExpr(0, shapedType.getContext()));
+  int64_t elementVectorRank = 0;
+  VectorType elementVectorType =
+      llvm::dyn_cast<VectorType>(shapedType.getElementType());
+  if (elementVectorType)
+    elementVectorRank += elementVectorType.getRank();
+  if (shapedType.getRank() < vectorType.getRank() - elementVectorRank) {
+    // Not enough dimensions in the shaped type to form a minor identity map.
+    return AffineMap();
+  }
----------------
banach-space wrote:

I think that you can restore it back to the top and also add an `assert` instead of `return AffineMap()`:
```suggestion
  int64_t elementVectorRank = 0;
  VectorType elementVectorType =
      llvm::dyn_cast<VectorType>(shapedType.getElementType());
  if (elementVectorType)
    elementVectorRank += elementVectorType.getRank();
  assert(shapedType.getRank() < vectorType.getRank() - elementVectorRank && "Not enough dimensions in the shaped type to form a minor identity map.")
  }
```

https://github.com/llvm/llvm-project/pull/133721