func.func @partitioned_call(%arg0: tensor<1x2x2x3xf32>) -> (tensor<1x2x2x3xf32>) {
    %0 = "tf.StatefulPartitionedCall"(%arg0) <{
      config = "", config_proto = "", executor_type = "", f = @some_func
    }> {
      _collective_manager_ids = [], device = ""
    } : (tensor<1x2x2x3xf32>) -> tensor<1x2x2x3xf32>
    // CHECK-SKIP: tf.StatefulPartitionedCall
    // CHECK-NOSKIP: call @some_func
    // CHECK-NOSKIP-NOT: tf.StatefulPartitionedCall

    %0 = tf_executor.graph {

    %0 = "tf.StatefulPartitionedCall"(%arg0) <{
      config = "", config_proto = "", executor_type = "", f = @some_func
    }> {
      _collective_manager_ids = [], device = ""
    } : (tensor<1x2x2x3xf32>) -> tensor<1x2x2x3xf32>
    func.return %0: tensor<1x2x2x3xf32>
  }

      config = "", config_proto = "", executor_type = "", f = @some_func
    }> {
      _collective_manager_ids = [], device = ""
    } : (tensor<3x3xf32>, tensor<3x3xf32>) -> tensor<3x3xf32>
    %3 = "tf.PartitionedCall"(%2, %1) <{
      config = "", config_proto = "", executor_type = "", f = @some_other_func
    }> {
      _collective_manager_ids = [], device = ""
    } : (tensor<3x3xf32>, tensor<3x3xf32>) -> tensor<3x3xf32>

    ```

    `_XlaMustCompile=true` in the following code

    ```mlir

}

// -----

module {
func.func @dense_image_warp(%arg0: tensor<2x4x4x1xf32>, %arg1: tensor<2x4x4x2xf32>) -> tensor<2x4x4x1xf32> {

```

`_XlaMustCompile=true` in the following code

```mlir

        }
      }
      attr {
        key: "Tout"
        value {
          list {
            type: DT_DOUBLE
          }
        }
      }
      attr {
        key: "_collective_manager_ids"
        value {
          list {
          }
        }
      }
      attr {
        key: "_output_shapes"
        value {
          list {
            shape {
              dim {