// CHECK: [[handle:%.*]] = "tf.VarHandleOp"()
  // CHECK: "tf.ReadVariableOp"([[handle]])
  %0 = "tf.ReadVariableOp"(%arg1) {device = "/device:CPU:0"} : (tensor<*x!tf_type.resource>) -> tensor<1x3xf32>
  %1 = "tf.AddV2"(%arg0, %0) {device = "/device:CPU:0"} : (tensor<1x3xf32>, tensor<1x3xf32>) -> tensor<1x3xf32>
  %2 = "tf.Identity"(%1) {device = "/device:CPU:0"} : (tensor<1x3xf32>) -> tensor<1x3xf32>
  func.return %2 : tensor<1x3xf32>
}

		map1, map2 = fake, fake
	}
	return
}

func TestConvertCPUProfile(t *testing.T) {
	addr1, addr2, map1, map2 := testPCs(t)

	b := []uint64{
		3, 0, 500, // hz = 500
		5, 0, 10, uint64(addr1 + 1), uint64(addr1 + 2), // 10 samples in addr1
		5, 0, 40, uint64(addr2 + 1), uint64(addr2 + 2), // 40 samples in addr2
		5, 0, 10, uint64(addr1 + 1), uint64(addr1 + 2), // 10 samples in addr1
	}

  // CHECK-NEXT: "tf.AddV2"({{.*}}, [[v]]) {device = "/CPU:0"} : (tensor<i32>, tensor<i32>) -> tensor<i32>
  // CHECK-NEXT: return
  %const = "tf.Const"() {device = "/CPU:0", value = dense<0> : tensor<i32>} : () -> tensor<i32>
  %x = "tf.AddV2"(%const, %const) {device = "/CPU:0"} : (tensor<i32>, tensor<i32>) -> tensor<i32>
  %r = "tf.AddV2"(%arg, %x) {device = "/CPU:0"} : (tensor<i32>, tensor<i32>) -> tensor<i32>

// CHECK: %[[mul:.*]] = "tf.Mul"(%arg0, %[[CONST_0]]) : (tensor<*xf32>, tensor<2xf32>) -> tensor<*xf32>
// CHECK: %[[add:.*]] = "tf.AddV2"(%[[mul]], %[[CONST]]) : (tensor<*xf32>, tensor<2xf32>) -> tensor<*xf32>
// CHECK-NEXT: return %[[add]] : tensor<*xf32>

// -----

func.func @not_decompose_batch_norm(%arg0: tensor<*xf32>) -> (tensor<*xf32>) {

// CHECK-LABEL: func @otherVariantWhileBody
// CHECK:       [[CST:%.*]] = "tf.Const"()
// CHECK-NEXT:  [[ADD:%.*]] = "tf.AddV2"(%arg2, [[CST]])
// CHECK-NEXT:  [[TENSOR_MAP_INSERT_RESULT:%.*]] = "tf.TensorMapInsert"(%arg3, %arg2, %arg2)
// CHECK-NEXT:  [[ADD_2:%.*]] = "tf.AddV2"(%arg0, [[CST]])
// CHECK-NEXT:  return [[ADD_2]], %arg1, [[ADD]], [[TENSOR_MAP_INSERT_RESULT]]

    %add2, %add2_control = tf_executor.island(%add1_control) wraps "tf.Add"(%add1, %arg1) : (tensor<*xi32>, tensor<i32>) -> tensor<*xi32>
    %print, %print_control = tf_executor.island wraps "tf.Print"(%add2) {message = "add2 result"} : (tensor<*xi32>) -> tensor<*xi32>
    tf_executor.fetch %add1, %add2, %add2_control : tensor<*xi32>, tensor<*xi32>, !tf_executor.control
  }

  %0 = "tf.Identity"(%cst) {device = ""} : (tensor<1xf32>) -> tensor<1xf32>
  %1 = "tf.AddV2"(%0, %cst_1) {device = ""} : (tensor<1xf32>, tensor<f32>) -> tensor<1xf32>
  %2 = "tf.Conv2DBackpropInput"(%arg0, %arg1, %arg2) {strides = [1, 2, 2, 1], padding="SAME", dilations=[1, 1, 1, 1]}: (tensor<4xi32>, tensor<3x3x1x32xf32>, tensor<15x14x14x32xf32>) -> tensor<15x28x28x1xf32>
  %3 = "tf.AddV2"(%2, %1): (tensor<15x28x28x1xf32>, tensor<1xf32>) -> tensor<15x28x28x1xf32>

      *literal2, xla::LiteralUtil::CreateR2<int32_t>({{3, 4}})));
}

TEST_F(PjRtExecutionUtilTest, PopulateCtxOutputs) {
  XlaOpRegistry::RegisterCompilationKernels();
  TF_EXPECT_OK(NodeDefBuilder("AddV2", "AddV2")
                   .Input(FakeInput(DT_INT32))
                   .Input(FakeInput(DT_INT32))
                   .Attr("T", DT_INT32)
                   .Device("/job:localhost/replica:0/task:0/device:XLA_CPU:0")

    %3 = "tf.LookupTableSizeV2"(%1) {device = ""} : (tensor<!tf_type.resource>) -> tensor<i64>
    %4 = "tf.AddV2"(%2, %3) {device = ""} : (tensor<i64>, tensor<i64>) -> tensor<i64>
    %5 = "tf.LookupTableFindV2"(%0, %arg0, %cst) {device = ""} : (tensor<!tf_type.resource>, tensor<?x!tf_type.string>, tensor<i64>) -> tensor<*xi64>
    %6 = "tf.AddV2"(%5, %4) {device = ""} : (tensor<*xi64>, tensor<i64>) -> tensor<*xi64>
    return %6 : tensor<*xi64>
  }

    %island1:3 = tf_executor.island {
      %add1 = "tf.Add"(%arg0, %arg1) : (tensor<*xi32>, tensor<i32>) -> tensor<*xi32>
      %add2 = "tf.Add"(%add1, %arg1) : (tensor<*xi32>, tensor<i32>) -> tensor<*xi32>
      %res = "tf.Print"(%add2) { message = "add result" } : (tensor<*xi32>) -> (tensor<*xi32>)
      tf_executor.yield %add1, %add2 : tensor<*xi32>, tensor<*xi32>
    }
    tf_executor.fetch %island1#0, %island1#1 : tensor<*xi32>, tensor<*xi32>
  }