TFE_TensorHandle* packed_handle =
      TFE_CreatePackedTensorHandle(ctx, handles.data(), &num_replicas, status);
  ASSERT_EQ(TF_GetCode(status), TF_OK) << TF_Message(status);
  EXPECT_EQ(TFE_TensorHandleDataType(packed_handle), TF_RESOURCE);
  EXPECT_EQ(TFE_TensorHandleNumDims(packed_handle, status), 0);
  EXPECT_EQ(TFE_TensorHandleNumElements(packed_handle, status), 1);

  if (*requested_placement == '\0') {
    TF_SetStatus(
        status, TF_INTERNAL,
        "Ops must be placed on the parallel device explicitly, or their inputs "
        "first un-packed. Got an un-placed op with an input placed on the "
        "parallel device.");
    return;
  }
  TFE_Context* context = TFE_OpGetContext(original_op, status);
  if (TF_GetCode(status) != TF_OK) return;

    ASSERT_EQ(TF_OK, TF_GetCode(status.get())) << TF_Message(status.get());
    ASSERT_TRUE(absl::StrContains(device_name, "GPU:0")) << device_name;

    // .backing_device of shape is CPU since the tensor is backed by CPU
    backing_device_name =
        TFE_TensorHandleBackingDeviceName(retvals[0], status.get());
    ASSERT_EQ(TF_OK, TF_GetCode(status.get())) << TF_Message(status.get());

}

void NoOpDeallocator(void* data, size_t, void*) {}

TEST(CAPI, MalformedTensor) {
  // See https://github.com/tensorflow/tensorflow/issues/7394
  // num_dims = 0 implies a scalar, so should be backed by at least 4 bytes of
  // data.
  TF_Tensor* t =
      TF_NewTensor(TF_FLOAT, nullptr, 0, nullptr, 0, &NoOpDeallocator, nullptr);
  ASSERT_TRUE(t == nullptr);
}

TEST(CAPI, AllocateTensor) {