[Mlir-commits] [mlir] [mlir][vector] Handle corner cases in DropUnitDimsFromTransposeOp. (PR #102518)

[llvmlistbot at llvm.org]

llvmbot wrote:


<!--LLVM PR SUMMARY COMMENT-->

@llvm/pr-subscribers-mlir

Author: Han-Chung Wang (hanhanW)

<details>
<summary>Changes</summary>

https://github.com/llvm/llvm-project/commit/da8778e499d8049ac68c2e152941a38ff2bc9fb2 breaks the lowering of vector.transpose that all the dimensions are unit dimensions. The revision fixes the issue and add a test.

---
Full diff: https://github.com/llvm/llvm-project/pull/102518.diff


2 Files Affected:

- (modified) mlir/lib/Dialect/Vector/Transforms/VectorTransforms.cpp (+8) 
- (modified) mlir/test/Dialect/Vector/vector-transfer-flatten.mlir (+12) 


``````````diff
diff --git a/mlir/lib/Dialect/Vector/Transforms/VectorTransforms.cpp b/mlir/lib/Dialect/Vector/Transforms/VectorTransforms.cpp
index bc0c96b32a80f..b4ae9b319343a 100644
--- a/mlir/lib/Dialect/Vector/Transforms/VectorTransforms.cpp
+++ b/mlir/lib/Dialect/Vector/Transforms/VectorTransforms.cpp
@@ -1771,6 +1771,14 @@ struct DropUnitDimsFromTransposeOp final
       newPerm.push_back(idx - droppedDimsBefore[idx]);
     }
 
+    // Fixup for `newPerm`. The `sourceTypeWithoutUnitDims` could be vector<1xT>
+    // type when the dimensions are unit dimensions. In this case, the newPerm
+    // should be [0].
+    if (sourceTypeWithoutUnitDims.getRank() == 1 &&
+        sourceTypeWithoutUnitDims.getShape()[0] == 1 && newPerm.empty()) {
+      newPerm.push_back(0);
+    }
+
     Location loc = op.getLoc();
     // Drop the unit dims via shape_cast.
     auto dropDimsShapeCast = rewriter.create<vector::ShapeCastOp>(
diff --git a/mlir/test/Dialect/Vector/vector-transfer-flatten.mlir b/mlir/test/Dialect/Vector/vector-transfer-flatten.mlir
index 937dbf22bb713..0d34d692393fd 100644
--- a/mlir/test/Dialect/Vector/vector-transfer-flatten.mlir
+++ b/mlir/test/Dialect/Vector/vector-transfer-flatten.mlir
@@ -737,6 +737,18 @@ func.func @transpose_with_scalable_unit_dims(%vec: vector<[1]x1x2x4x1xf32>) -> v
 
 // -----
 
+func.func @transpose_with_all_unit_dims(%vec: vector<1x1x1xf32>) -> vector<1x1x1xf32> {
+  %res = vector.transpose %vec, [0, 2, 1] : vector<1x1x1xf32> to vector<1x1x1xf32>
+  return %res : vector<1x1x1xf32>
+}
+// The `vec` is returned because there are other flattening patterns fold
+// vector.shape_cast ops away.
+// CHECK-LABEL: func.func @transpose_with_all_unit_dims
+// CHECK-SAME:      %[[VEC:.[a-zA-Z0-9]+]]
+// CHECK-NEXT:    return %[[VEC]]
+
+// -----
+
 func.func @negative_transpose_with_no_unit_dims(%vec: vector<4x2x3xf32>) -> vector<4x3x2xf32> {
   %res = vector.transpose %vec, [0, 2, 1] : vector<4x2x3xf32> to vector<4x3x2xf32>
   return %res : vector<4x3x2xf32>

``````````

</details>


https://github.com/llvm/llvm-project/pull/102518