[Mlir-commits] [mlir] [mlir][linalg] Move transpose_matmul to targeted transform op (PR #89717)

[Oleksandr Alex Zinenko]

================
@@ -3422,6 +3416,34 @@ DiagnosedSilenceableFailure transform::TransposeConv2DOp::applyToOne(
   return DiagnosedSilenceableFailure::success();
 }
 
+//===----------------------------------------------------------------------===//
+// TransposeMatmulOp
+//===----------------------------------------------------------------------===//
+
+DiagnosedSilenceableFailure transform::TransposeMatmulOp::applyToOne(
+    transform::TransformRewriter &rewriter, linalg::LinalgOp target,
+    transform::ApplyToEachResultList &results,
+    transform::TransformState &state) {
+  rewriter.setInsertionPoint(target);
+  bool transposeLHS = getInputToTranspose() == TransposeMatmulInput::lhs;
+  auto maybeTransformed =
+      TypeSwitch<Operation *, FailureOr<Operation *>>(target)
+          .Case([&](linalg::MatmulOp op) {
+            return transposeMatmul(rewriter, op, transposeLHS);
+          })
+          .Case([&](linalg::BatchMatmulOp op) {
+            return transposeBatchMatmul(rewriter, op, transposeLHS);
+          })
+          .Default([&](Operation *op) {
+            return rewriter.notifyMatchFailure(op, "not supported");
+          });
+  if (failed(maybeTransformed))
+    return emitDefaultSilenceableFailure(target);
----------------
ftynse wrote:

Nit: could we rather say "not supported" here than in the `notifyMatchFailure` that is ignored by the infra?

https://github.com/llvm/llvm-project/pull/89717