[Mlir-commits] [mlir] 56ada0f - [mlir][vector] Use dyn_cast instead of cast in patterns

[Lei Zhang]

Author: Lei Zhang
Date: 2021-11-08T12:34:14-05:00
New Revision: 56ada0f80df26a09a0e84a67720efd916633b1f3

URL: https://github.com/llvm/llvm-project/commit/56ada0f80df26a09a0e84a67720efd916633b1f3
DIFF: https://github.com/llvm/llvm-project/commit/56ada0f80df26a09a0e84a67720efd916633b1f3.diff

LOG: [mlir][vector] Use dyn_cast instead of cast in patterns

This avoids crashes when the pattern is applied to ops with
tensor semantics.

Reviewed By: ThomasRaoux

Differential Revision: https://reviews.llvm.org/D113415

Added: 
    

Modified: 
    mlir/lib/Dialect/Vector/VectorTransforms.cpp

Removed: 
    


################################################################################
diff  --git a/mlir/lib/Dialect/Vector/VectorTransforms.cpp b/mlir/lib/Dialect/Vector/VectorTransforms.cpp
index 912031035ed7e..6528789810bfa 100644
--- a/mlir/lib/Dialect/Vector/VectorTransforms.cpp
+++ b/mlir/lib/Dialect/Vector/VectorTransforms.cpp
@@ -3579,7 +3579,7 @@ class DropInnerMostUnitDims : public OpRewritePattern<vector::TransferReadOp> {
 
   LogicalResult matchAndRewrite(vector::TransferReadOp readOp,
                                 PatternRewriter &rewriter) const override {
-    auto srcType = readOp.source().getType().cast<MemRefType>();
+    auto srcType = readOp.source().getType().dyn_cast<MemRefType>();
     if (!srcType || !srcType.hasStaticShape())
       return failure();