func.func @func_with_sharding_inside_while_body(%arg0: tensor<i32>, %arg1: tensor<128x1024xf32>) -> (tensor<128x1024xf32>) {
  %cst = "tf.Const"() <{value = dense<0> : tensor<i32>}> {device = ""} : () -> tensor<i32>
  %0:2 = "tf.WhileRegion"(%cst, %arg1) <{is_stateless = false, parallel_iterations = 1 : i64}> ({
    ^bb0(%arg2: tensor<i32>, %arg3: tensor<128x1024xf32>):
      %1 = "tf.Less"(%arg2, %arg0) : (tensor<i32>, tensor<i32>) -> tensor<i1>

func.func @unknown_resource_op_while_body(%arg0: !tf_res, %arg1: !tf_res, %arg2: tensor<f32>, %arg3: tensor<f32>) -> (!tf_res, !tf_res, tensor<f32>, tensor<f32>) {
  %graph:4 = tf_executor.graph {
    %assign_control_0 = tf_executor.island wraps "tf.AssignVariableOp"(%arg0, %arg2) : (tensor<!tf_type.resource<tensor<f32>>>, tensor<f32>) -> ()

		{name: "FNMSUBS", argLength: 3, reg: fp31, asm: "FNMSUBS"}, // -arg0 + (arg1 * arg2)
		{name: "FNMSUBD", argLength: 3, reg: fp31, asm: "FNMSUBD"}, // -arg0 + (arg1 * arg2)
		{name: "MADD", argLength: 3, reg: gp31, asm: "MADD"},       // +arg0 + (arg1 * arg2)
		{name: "MADDW", argLength: 3, reg: gp31, asm: "MADDW"},     // +arg0 + (arg1 * arg2), 32-bit
		{name: "MSUB", argLength: 3, reg: gp31, asm: "MSUB"},       // +arg0 - (arg1 * arg2)

  // CHECK: [[PI_0:%.*]] = "tf.TPUPartitionedInputV2"([[ARG0]], [[ARG1]])
  %pi_0 = "tf.TPUPartitionedInputV2"(%arg0, %arg1) {device = "", partition_dims = []} : (tensor<!tf_type.resource<tensor<10x3xf32>>>, tensor<!tf_type.resource<tensor<10x3xf32>>>) -> tensor<!tf_type.resource<tensor<10x3xf32>>>
  // CHECK: [[PI_1:%.*]] = "tf.TPUPartitionedInputV2"([[ARG2]], [[ARG3]])

  %0 = "tfl.transpose"(%arg0, %cst_1) : (tensor<4x4x3xf32>, tensor<3xi32>) -> tensor<4x4x3xf32>
  %1:6 = "tfl.while"(%cst_7, %cst_7, %cst_2, %cst, %cst, %0) ({
  ^bb0(%arg1: tensor<i32>, %arg2: tensor<i32>, %arg3: tensor<*xf32>, %arg4: tensor<4x2xf32>, %arg5: tensor<4x2xf32>, %arg6: tensor<*xf32>):
    %5 = "tfl.less"(%arg2, %cst_3) : (tensor<i32>, tensor<i32>) -> tensor<i1>

		{name: "SUBLloadidx4", argLength: 4, reg: gp21loadidx, asm: "SUBL", aux: "SymOff", resultInArg0: true, clobberFlags: true, symEffect: "Read"},  // arg0 - tmp, tmp loaded from  arg1+arg2*4+auxint+aux, arg3 = mem

// CHECK:           return %[[VAL_4]] : tensor<2x3xf16>
// CHECK:         }
func.func @floordiv_f16_broadcast(%arg0: tensor<2x3xf16>, %arg1: tensor<3xf16>) -> tensor<2x3xf16> {
  %0 = "chlo.broadcast_divide"(%arg0, %arg1) {broadcast_dimensions = array<i64: 1>} : (tensor<2x3xf16>, tensor<3xf16>) -> tensor<2x3xf16>
  %1 = "chlo.broadcast_divide"(%arg0, %arg1) {broadcast_dimensions = array<i64: 1>} : (tensor<2x3xf16>, tensor<3xf16>) -> tensor<2x3xf16>

  %0 = func.call @f(%arg0, %arg1, %arg0, %arg1) : (f32, f32, f32, f32) -> f32
  return %0 : f32
}

// -----

// CHECK-LABEL: leaves_while_loops_alone
func.func @leaves_while_loops_alone(%arg0: tensor<i32>, %arg1: tensor<f32>) {
  // CHECK: While{{.*}}(%arg0, %arg1)
  %0, %1 = "tf.While"(%arg0, %arg1) {body = @body, cond = @cond, is_stateless = false} :
      (tensor<i32>, tensor<f32>) -> (tensor<i32>, tensor<f32>)

		{name: "ADDF", argLength: 2, reg: fp21, asm: "ADDF", commutative: true}, // arg0 + arg1
		{name: "ADDD", argLength: 2, reg: fp21, asm: "ADDD", commutative: true}, // arg0 + arg1
		{name: "SUBF", argLength: 2, reg: fp21, asm: "SUBF"},                    // arg0 - arg1
		{name: "SUBD", argLength: 2, reg: fp21, asm: "SUBD"},                    // arg0 - arg1

    func.return
}

func.func private @__reduce_func_2(%arg0: tensor<i64> {tf._user_specified_name = "args_0"}, %arg1: tensor<32xf32> {tf._user_specified_name = "args_1"}, %arg2: tensor<!tf_type.resource<tensor<64xf32>>>, %arg3: tensor<!tf_type.resource<tensor<128xf32>>>)
  -> (tensor<i64>) attributes {tf._tf_data_function = true, tf.signature.is_stateful} {