# CUDA init.createctx,ipcgetmemhandle act normal, but get CUDA\_EEROR\_INVALID\_CONTEXT when use Ipcopenmemhandle

**URL:** <https://forums.developer.nvidia.com/t/cuda-init-createctx-ipcgetmemhandle-act-normal-but-get-cuda-eeror-invalid-context-when-use-ipcopenmemhandle/221754>\
**Category:** General Topics & Other SDKs\
**Tags:** cuda\
**Created:** [July 25, 2022, 4:03am UTC](https://forums.developer.nvidia.com/t/cuda-init-createctx-ipcgetmemhandle-act-normal-but-get-cuda-eeror-invalid-context-when-use-ipcopenmemhandle/221754 "2022-07-25T04:03:45Z")\
**Posts on this page:** 1\
**Page:** 1

<div class="post-metadata">

**Author:** ![Tumb1eweed](https://developer.download.nvidia.com/images/forums/profile-default-devtalk-84.png) [@Tumb1eweed](https://forums.developer.nvidia.com/u/Tumb1eweed)\
**Post date:** [July 25, 2022, 4:03am UTC](https://forums.developer.nvidia.com/t/cuda-init-createctx-ipcgetmemhandle-act-normal-but-get-cuda-eeror-invalid-context-when-use-ipcopenmemhandle/221754/1 "2022-07-25T04:03:45Z")

</div>

```auto
class ImagePublisher : public rclcpp::Node
{
public:
  ImagePublisher()
  : Node("image_publisher")
  {
    Mat img(1920, 1080, CV_8UC3);
    randu(img, Scalar(0, 0, 0), Scalar(255, 255, 255));
    imwrite("publish.jpg", img);
    
    int devCount;
    void *ptr = NULL;
    CUresult rc;
    CUdevice cuda_device;
    CUcontext cuda_context;
    CUdeviceptr cuda_devptr;
    CUipcMemHandle ipc_memhandle;

    rc = cuInit(0);
    if (rc != CUDA_SUCCESS)	Elog("failed on cuInit: %s", cudaErrorName(rc));
    rc = cuDeviceGet(&cuda_device, 0);
    if (rc != CUDA_SUCCESS)	Elog("failed on cuDeviceGet: %s", cudaErrorName(rc));
    rc = cuCtxCreate(&cuda_context, 0, cuda_device);
    if (rc != CUDA_SUCCESS)	Elog("failed on cuCxtCreate: %s", cudaErrorName(rc));

    size_t count = img.total() * img.elemSize();
    rc = cuMemAlloc(&cuda_devptr, count);
    if (rc != CUDA_SUCCESS)	Elog("failed on cuMemAlloc: %s", cudaErrorName(rc));
    rc= cuIpcGetMemHandle(&ipc_memhandle, cuda_devptr);
    if (rc != CUDA_SUCCESS)	Elog("failed on cuMemGetMemHandle: %s", cudaErrorName(rc));
    RCLCPP_INFO(this->get_logger(), "%d", count);
    uint8_t* byte_array = reinterpret_cast<uint8_t*>(&ipc_memhandle);

    rc = cuIpcOpenMemHandle(&cuda_devptr, ipc_memhandle, CU_IPC_MEM_LAZY_ENABLE_PEER_ACCESS);
    if (rc != CUDA_SUCCESS)	Elog("failed on cuIpcOpenMemHandle: %s", cudaErrorName(rc));

    publisher_ = this->create_publisher<std_msgs::msg::UInt8MultiArray>("cuda_transport", 10);

    auto callback = [&]() -> void {
      std_msgs::msg::UInt8MultiArray msg;
      msg.layout.dim.push_back(std_msgs::msg::MultiArrayDimension());
      msg.layout.dim[0].size = sizeof(ipc_memhandle);
      msg.data.resize(sizeof(ipc_memhandle));
      memcpy(msg.data.data(), byte_array, sizeof(ipc_memhandle));
      rosidl_generator_traits::to_yaml(msg, std::cout);
      publisher_->publish(msg);
      RCLCPP_INFO(this->get_logger(), "publish end!");
    };
    timer_ = this->create_wall_timer(1s, callback);
  }

private:
  rclcpp::TimerBase::SharedPtr timer_;
  rclcpp::Publisher<std_msgs::msg::UInt8MultiArray>::SharedPtr publisher_;
};

```

![image](https://global.discourse-cdn.com/nvidia/original/3X/c/e/ce7fc3f958050d72e45586d2297ea2c37a2d715f.png)