[Mlir-commits] [mlir] [mlir][python] meta region_op (PR #75673)

[llvmlistbot at llvm.org]

================
@@ -93,3 +118,73 @@ def test_match_ops_mixed(op: OpHandle):
     # CHECK-NEXT: %[[VAL_1:.*]] = transform.structured.match
     # CHECK-SAME:   ops{["scf.for", "linalg.matmul", "scf.forall"]} in %[[VAL_0]]
     # CHECK-SAME:     -> !transform.any_op
+
+
+# CHECK-LABEL: TEST: test_sequence_region
+ at construct_and_print_in_module
+def test_sequence_region():
+    # CHECK-LABEL:   func.func @loop_unroll_op() {
+    # CHECK:           %[[VAL_0:.*]] = arith.constant 0 : index
+    # CHECK:           %[[VAL_1:.*]] = arith.constant 42 : index
+    # CHECK:           %[[VAL_2:.*]] = arith.constant 5 : index
+    # CHECK:           scf.for %[[VAL_3:.*]] = %[[VAL_0]] to %[[VAL_1]] step %[[VAL_2]] {
+    # CHECK:             %[[VAL_4:.*]] = arith.addi %[[VAL_3]], %[[VAL_3]] : index
+    # CHECK:           }
+    # CHECK:           return
+    # CHECK:         }
+    @func.func()
+    def loop_unroll_op():
+        for i in scf.for_(0, 42, 5):
+            v = arith.addi(i, i)
+            scf.yield_([])
+
+    # CHECK:   transform.sequence  failures(propagate) {
+    # CHECK:   ^bb0(%[[VAL_0:.*]]: !transform.any_op):
----------------
martin-luecke wrote:

nit: don't match block labels here and below

https://github.com/llvm/llvm-project/pull/75673