%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>
  }

  %0 = "tf.AddV2"(%arg0, %arg1) : (tensor<i32>, tensor<i32>) -> tensor<i32>
  return %0: tensor<i32>
}

// CHECK-LABEL: func.func private @"random/while_body/TfMlrtAsyncWhileBody"(%arg0: tensor<i32>, %arg1: tensor<i32>) -> tensor<i32> {
func.func private @"random/while_body/TfMlrtAsyncWhileBody"(%arg0: tensor<i32>, %arg1: tensor<i32>) -> tensor<i32> {

  TF_Operation* feed3 = Placeholder(func_graph_, s_, "feed3");
  TF_Operation* add1 = Add(feed1, feed2, func_graph_, s_, "add1");
  TF_Operation* add2 = Add(add1, feed3, func_graph_, s_, "add2");
  Define(1, {add2}, {add1, feed3}, {add2}, {});

  // Use, run, and verify
  TF_Operation* two = ScalarConst(2, host_graph_, s_, "two");
  TF_Operation* func_feed = Placeholder(host_graph_, s_);

  %res:2 = "tf.IfRegion"(%arg0) ({
    %call = func.call @test_if_then1(%arg1) : (tensor<4xf32>) -> tensor<4xf32>
    %add = "tf.AddV2"(%call, %call) : (tensor<4xf32>, tensor<4xf32>) -> tensor<4xf32>
    "tf.Yield"(%call, %add) : (tensor<4xf32>, tensor<4xf32>) -> ()
  },  {
    %call_0 = func.call @test_if_else1(%arg1) : (tensor<4xf32>) -> tensor<4xf32>

${generate} \
  --category=array \
  Identity \
  IdentityN \
  ZerosLike \
  Shape \
  ExpandDims \
  OnesLike

${generate} \
  --category=math \
  Mul \
  Conj \
  AddV2 \
  MatMul \
  Neg \
  Sum \
  Sub \
  Div \
  DivNoNan \
  Exp \
  Sqrt \
  SqrtGrad \
  Log1p

${generate} \
  --category=nn \
  SparseSoftmaxCrossEntropyWithLogits \
  ReluGrad \

    %0 = "tf.MatMul"(%arg0, %cst_0) {attr_map = "0:transpose_a,1:transpose_a", device = "", transpose_a = false, transpose_b = false} : (tensor<1x2x2x2xf32>, tensor<2x1024xf32>) -> tensor<*xf32>
    // AddV2 not in quantizable op list.
    %1 = "tf.AddV2"(%arg1, %cst_0) {device = ""} : (tensor<2x1024xf32>, tensor<2x1024xf32>) -> tensor<2x1024xf32>
    func.return %0, %1 : tensor<*xf32>, tensor<2x1024xf32>
  }

// CHECK-LABEL: func @matmul_multiuses

			Valu("sb2", OpSB, c.config.Types.Uintptr, 0, nil),
			Valu("addr1", OpAddr, c.config.Types.Int64.PtrTo(), 0, nil, "sb1"),
			Valu("addr2", OpAddr, c.config.Types.Int64.PtrTo(), 0, nil, "sb2"),
			Valu("a1ld", OpLoad, c.config.Types.Int64, 0, nil, "addr1", "start"),
			Valu("a2ld", OpLoad, c.config.Types.Int64, 0, nil, "addr2", "start"),
			Valu("c1", OpConst64, c.config.Types.Int64, 1, nil),
			Valu("r1", OpAdd64, c.config.Types.Int64, 0, nil, "a1ld", "c1"),

	c1.SetReadBuffer(2048)
	c1.SetWriteBuffer(2048)

	a2, err := ResolveUnixAddr("unixgram", addr2)
	if err != nil {
		t.Fatal(err)
	}
	c2, err := DialUnix("unixgram", a2, nil)
	if err != nil {
		t.Fatal(err)
	}
	defer c2.Close()
	defer os.Remove(addr2)
	c2.LocalAddr()
	c2.RemoteAddr()
	c2.SetDeadline(time.Now().Add(someTimeout))

  // CHECK-NEXT: [[t3:%.*]] = tf_mlrt.await [[f2]]
  // CHECK-NEXT: return [[t2]], [[t3]]
  %t0 = tf_mlrt.await %f0
  %t1 = tf_mlrt.await %f1
  %t2 = tf_mlrt.executeop(%t0, %t1) {node_def = "AddV2", op_key = 0 : i32} : (!tf_mlrt.tensor, !tf_mlrt.tensor) -> (!tf_mlrt.tensor)
  %t3 = tf_mlrt.await %f2
  func.return %t2, %t3 : !tf_mlrt.tensor, !tf_mlrt.tensor
}

// -----

// CHECK-LABEL: @main

    %0 = "tf.AddV2"(%arg0, %cst) {device = ""} : (tensor<i32>, tensor<i32>) -> tensor<i32>
    %1 = "tf.Identity"(%0) {device = ""} : (tensor<i32>) -> tensor<i32>
    %2 = "tf.Identity"(%arg1) {device = ""} : (tensor<i32>) -> tensor<i32>
    %4 = "tf.AddV2"(%arg2, %cst_0) {device = ""} : (tensor<1x5x5x1024xf32>, tensor<1x5x5x1024xf32>) -> tensor<1x5x5x1024xf32>