Main Content

Use Handwritten Code to Integrate C++ Messages with POSIX

This example shows how to integrate generated C++ code with POSIX to facilitate message communication. The example application is a multisensor positioning system designed to estimate the position of a vehicle. The multisensor positioning system is composed of two components: sensors (multiple models) and an estimation model. The outlined workflow shows how to prepare each component, generate code, and integrate the generated code to pass messages between system components. This example assumes you are familiar with using messages in the Simulink environment. For more information, see Generate C++ Messages to Communicate Between Simulink and an Operating System or Middleware.

Sensor Models

The first component in the multisensor positioning system is the sensors. The positioning system uses an accelometer and a GPS sensor to estimate the vehicle position. Each sensor model is composed of a Function block, Send block, and Output port. Each sensor model sends messages to the estimation model by using POSIX messages.

Prepare Models

Open the sensor models rtwdemo_accel_send and rtwdemo_gps_send.

For each model, perform these steps:

  1. In the Apps Gallery, click Embedded Coder.

  2. On the toolstrip, set the Stop time to 100 seconds. The stop time defines the duration of the data collection when estimating the vehicle position.

  3. In the model, click to open the Function block. In the Block Parameters, set the File name to the MAT-file for the model. Set the Sample time to 0.01.

Generate Code

1. For each sensor model, in the Configuration Parameters dialog box, set these parameters:

  • On the Code Generation pane, set Language to C++. Under Advanced parameters, set Build configuration to Specify and add -lrt to the C++ Linker.

  • On the Interface pane , set Code interface packaging to C++ class and Multi-instance code error diagnostic to none. Under Advanced parameters, select MAT-file logging.

2. Generate code. On the C++ Code tab, click Build.

Integrate Code

To integrate the generated C++ code from each sensor model with POSIX, hand write the implementation of the send class and generated main program (or create your own). This example provides a customized main program with an implemented send class. To integrate the code from each sensor model, replace the generated main file with the customized main file.

Review the customized main file for each sensor:

1. The customized main file for each sensor is implemented in the same way. As an example, open the customized main file rtwdemo_accel_send_ert_main_customized.cpp.

2. Review the implemented send class:

  • The send class function, SendData, calls into the POSIX API (mq_send) to send messages.

  • To provide access to a message queue, a POSIX message queue is added as a class variable and a set method is implemented to set the queue.

class accelSendModelClassSendData_real_T : public SendData_real_T{
     mqd_t msgQueue;
     public:
     // Use POSIX to send messages
     void SendData(const real_T* data, int32_T length, int32_T* status)
     {
         unsigned int priority = 1;
         *status = mq_send(msgQueue, (char*)data, length, priority);
     }
     // Set method for class POSIX queue
     void SetMQ(mqd_t mq)
     {
         msgQueue = mq;
     }
 };

After the send class implementation, an instance of the send class (a send object) is created. The send object is then attached to the sensor model to enable the sensor to send messages to the estimate model.

// Create send object
static accelSendModelClassSendData_real_T Out1SendData_arg;
// Attach send object to sensor model to send messages
static accelSendModelClass rtObj( Out1SendData_arg);

3. Review the added POSIX queue behavior functions. The main function uses these functions to open and close the POSIX queue.

mqd_t openMQ(const char *name, int flags, int maxmsg, int msgsize)
 {
     struct mq_attr attr;
     attr.mq_flags = 0;
     attr.mq_maxmsg = maxmsg;
     attr.mq_msgsize = msgsize;
     mqd_t msgQueue = mq_open(name, flags, 0664, &attr);
     if (msgQueue < 0) {
         printf("mq_open failed\n");
         exit(1);
     }
     return msgQueue;
 }
void closeMQ(mqd_t msgQueue)
 {
     mq_close(msgQueue);
 }

4. Review the main function. The implemented main function uses the previously defined class and queue functions to send messages and to manage the POSIX message queue. The main function performs the following:

  • Opens a POSIX message queue.

  • Sends messages.

  • Closes the POSIX queue.

int_T main(int_T argc, const char *argv[])
 {
     // Unused arguments
     (void)(argc);
     (void)(argv);
     // Initialize model
     rtObj.initialize();
     // Open POSIX message queue
     mqd_t msgQueue = openMQ("/PosixMQ_accel", O_CREAT | O_WRONLY, 2, 16);
     Out1SendData_arg.SetMQ(msgQueue);
     // Send messages
     while (rtmGetErrorStatus(rtObj.getRTM()) == (NULL)) {
     rt_OneStep();
     }
     // Matfile logging
     rt_StopDataLogging(MATFILE, rtObj.getRTM()->rtwLogInfo);
     // Disable rt_OneStep() here
     // Close POSIX message queue
     closeMQ(msgQueue);
     return 0;
 }

Estimate Model

The second component is the estimate model. The estimate model receives messages that have input data from the sensor models and calculates the estimated position of the vehicle.

Prepare Model

Open the model rtwdemo_pos_estimate.

  1. In the Apps Gallery, click Embedded Coder.

  2. On the toolstrip, set the Stop time to 100 seconds.

  3. In the model, click to open each inport. In the Block Parameters, set Data type to double, Port dimensions to 3, and Sample time to 0.01.

  4. Click to open each Message Receive block. Clear the Use internal queue parameter to enable the model to use a POSIX message queue.

  5. Click to open the Function block. This block implements the estimation algorithm.

Generate Code

1. In the Configuration Parameters dialog box, set these parameters:

  • On the Code Generation pane, set Language to C++. Under Advanced parameters, set Build configuration to Specify and add -lrt to the C++ Linker.

  • On the Interface pane, set Code interface packaging to C++ class and set Multi-instance code error diagnostic to none. Under Advanced parameters, select MAT-file logging.

2. Generate code. On the C++ Code tab, click Build.

Integrate Code

To integrate the generated C++ code from the estimate model with POSIX, hand write the implementation of the receive class and generated main program (or create your own). The estimate model receives messages from the sensor models and calculates an estimated position of the vehicle. This examples provides a customized main program with an implemented receive class. To integrate the code from the estimate model, replace the generated main file with the customized main file.

Review the customized main file:

1. Open the customized main file rtwdemo_pos_estimate_ert_main_customized.cpp.

2. Review the implemented receive class:

  • The receive class function, RecvData, calls into the POSIX API (mq_receive) to receive messages.

  • To provide access to a message queue, a POSIX message queue is added as a class variable and a set method is implemented to set the queue.

class positionEstimateModelClassRecvData_real_T : public RecvData_real_T{
     mqd_t msgQueue;
     public:
     // Uses POSIX API mq_receive to receive messages
     void RecvData(real_T* data, int32_T length, int32_T* status)
     {
     // Add receive data logic here
     unsigned int priority = 1;
     *status = mq_receive(msgQueue, (char *)data, length, &priority);
     }
     // Set method for class POSIX queue
     void SetMQ(mqd_t mq)
     {
         msgQueue = mq;
     }
  };

After the receive class implementation, an instance of the receive class (a receive object) is created for each sensor model. The receive objects are then attached to the estimate model to enable the model to receive messages from each sensor.

// Create receive objects
static positionEstimateModelClassRecvData_real_T In1RecvData_arg;
static positionEstimateModelClassRecvData_real_T In2RecvData_arg;
// Attach receive objects to estimate model to receive messages
static positionEstimateModelClass rtObj( In1RecvData_arg, In2RecvData_arg);

3. Review the added POSIX queue behavior functions. The main function uses these functions to open, close, and unlink the POSIX queue.

mqd_t openMQ(const char *name, int flags, int maxmsg, int msgsize)
 {
     struct mq_attr attr;
     attr.mq_flags = 0;
     attr.mq_maxmsg = maxmsg;
     attr.mq_msgsize = msgsize;
     mqd_t msgQueue = mq_open(name, flags, 0664, &attr);
     if (msgQueue < 0) {
         printf("mq_open failed\n");
         exit(1);
     }
     return msgQueue;
 }
void closeMQ(mqd_t msgQueue)
 {
     mq_close(msgQueue);
 }
void unlinkMQ(const char *name)
 {
     if (mq_unlink(name) < 0) {
         printf("mq_unlink failed\n");
         exit(1);
     }
 }

4. Review the main function. The implemented main function uses the previously defined class and queue functions to receive messages and to manage the POSIX queue. The main function performs the following:

  • Opens a POSIX message queue.

  • Receives messages.

  • Closes the POSIX queue.

int_T main(int_T argc, const char *argv[])
 {
     // Unused arguments
     (void)(argc);
     (void)(argv);
     // Initialize model
     rtObj.initialize();
     // Open POSIX message queues
     mqd_t msgQueueAccel = openMQ("/PosixMQ_accel", O_CREAT | O_RDONLY, 2, 16);
     mqd_t msgQueueGPS = openMQ("/PosixMQ_gpspos", O_CREAT | O_RDONLY, 2, 16);
     In1RecvData_arg.SetMQ(msgQueueAccel);
     In2RecvData_arg.SetMQ(msgQueueGPS);
     // Receive messages
     while (rtmGetErrorStatus(rtObj.getRTM()) == (NULL)) {
         rt_OneStep();
     }
     // Matfile logging
     rt_StopDataLogging(MATFILE, rtObj.getRTM()->rtwLogInfo);
     // Disable rt_OneStep() here
     // Close POSIX message queues
      closeMQ(msgQueueAccel);
      closeMQ(msgQueueGPS);
      unlinkMQ("/PosixMQ_accel");
      unlinkMQ("/PosixMQ_gpspos");
     return 0;
 }

Execute Multisensor Positioning System

To execute the multisensor positioning system, rebuild and run the executables for all models. To view results, run the provided build script rtwdemo_positioning_system_script. An example of results is shown in this figure.