Azure RTOS prevents CAN Interrupts

Rob Ashworth · ‎2025-10-23

I am running an application with Azure RTOS, and CAN bus. All was well until I needed to increase the tick time to 2000/second (from 1000/second). When I did this the CAN Interrupts stop working. To make things even worse, when I change the tick time back to 1000/sec within CubeIDE, it still doesn't work and I have to pull and older version off Github to get it going again. I'm not overly familiar with Azure RTOS, usually I just do my own OS. CAN interrupts are priority 0, RTOS 15. TIM6 is used for tick for RTOS.

CAN Initilization:

  /* Configure Rx filter */
   sFilterConfig.IdType = FDCAN_STANDARD_ID;
   sFilterConfig.FilterIndex = 0;
 //  sFilterConfig.FilterType = FDCAN_FILTER_DUAL;
   sFilterConfig.FilterType = FDCAN_FILTER_MASK;//Classic filter
   sFilterConfig.FilterConfig = FDCAN_FILTER_TO_RXFIFO0;
   sFilterConfig.FilterID1 = 0x7F1;
   sFilterConfig.FilterID2 = 0x7F1;

   if (HAL_FDCAN_ConfigFilter(&hfdcan1, &sFilterConfig) != HAL_OK)
   {
     Error_Handler();
   }

   sFilterConfig.IdType = FDCAN_STANDARD_ID;
   sFilterConfig.FilterIndex = 1;
 //  sFilterConfig.FilterType = FDCAN_FILTER_DUAL;
   sFilterConfig.FilterType = FDCAN_FILTER_MASK;//classic
   sFilterConfig.FilterConfig = FDCAN_FILTER_TO_RXFIFO1;
   sFilterConfig.FilterID1 = 0x7F2;
   sFilterConfig.FilterID2 = 0x7F2;

   if (HAL_FDCAN_ConfigFilter(&hfdcan1, &sFilterConfig) != HAL_OK)
   {
    /* Filter configuration Error */
    Error_Handler();
   }
   //GLobal Filter setting
   /* Configure global filter to reject all non-matching frames */
   HAL_FDCAN_ConfigGlobalFilter(&hfdcan1, FDCAN_REJECT, FDCAN_REJECT, FDCAN_FILTER_REMOTE, FDCAN_FILTER_REMOTE);

   //Activate the Configuration CAN reception Interrupt
   if (HAL_FDCAN_ActivateNotification(&hfdcan1, FDCAN_IT_RX_FIFO0_NEW_MESSAGE, 0) != HAL_OK)
   {
    /* Notification Error */
    Error_Handler();
   }

   //Activate the Configuration CAN reception Interrupt
   if (HAL_FDCAN_ActivateNotification(&hfdcan1, FDCAN_IT_RX_FIFO1_NEW_MESSAGE, 0) != HAL_OK)
   {
    /* Notification Error */
    Error_Handler();
   }
   HAL_FDCAN_Start(&hfdcan1);

CAN NVIC configuration:

   /**FDCAN1 GPIO Configuration
    PA11     ------> FDCAN1_RX
    PA12     ------> FDCAN1_TX
    */
    GPIO_InitStruct.Pin = GPIO_PIN_11|GPIO_PIN_12;
    GPIO_InitStruct.Mode = GPIO_MODE_AF_PP;
    GPIO_InitStruct.Pull = GPIO_NOPULL;
    GPIO_InitStruct.Speed = GPIO_SPEED_FREQ_LOW;
    GPIO_InitStruct.Alternate = GPIO_AF9_FDCAN1;
    HAL_GPIO_Init(GPIOA, &GPIO_InitStruct);

    /* FDCAN1 interrupt Init */
    HAL_NVIC_SetPriority(FDCAN1_IT0_IRQn, 0, 0);
    HAL_NVIC_EnableIRQ(FDCAN1_IT0_IRQn);
    HAL_NVIC_SetPriority(FDCAN1_IT1_IRQn, 0, 0);
    HAL_NVIC_EnableIRQ(FDCAN1_IT1_IRQn);

And Azure RTOS:

UINT App_ThreadX_Init(VOID *memory_ptr)
{
  UINT ret = TX_SUCCESS;
  /* USER CODE BEGIN App_ThreadX_MEM_POOL */
  tx_thread_create(&thread_ptr,"my_thread",my_thread_entry,0x1234,thread_stack,THREAD_STACK_SIZE,15,15,1,TX_AUTO_START);
  tx_thread_create(&thread_ptr_HK,"SlowRowThread",SlowRowThread_Entry,0x1234,thread_stack_HK,THREAD_STACK_SIZE_HK,15,14,1,TX_AUTO_START);
  /* USER CODE END App_ThreadX_MEM_POOL */
  /* USER CODE BEGIN App_ThreadX_Init */
  /* USER CODE END App_ThreadX_Init */

  return ret;
}

mbarg.1 · ‎2025-10-24

Interrupts have few to share with ThreadX even if sometime ThreadX suspends interrupts but it is for few clock cycles ...

One question: what do you expect to happen when CAN triggers IT ? Can you share code of CAN IT Callback?

Can you share .ioc before and after can changes (both first and second) ? It seems you are cumulating some bad side effect from IDE ..

To double ckeck IDE, change something very useless like a GPIO pin and revert - is your design back to original? can use compare files - should be identical.

Rob Ashworth · ‎2025-10-24

Here is the callback for both CAN RX functions. I will try and compare the 2 .ioc files, but not sure the best way of doing this, as not sure their the easiest of things to look at in a text editor...also the .ioc obviously modifies a lot of files in the background!

//FIFO0 is used by customer - Period, CAN ID, System Zero, CAN-FD or Classic CAN
void HAL_FDCAN_RxFifo0Callback(FDCAN_HandleTypeDef *hfdcan, uint32_t RxFifo0ITs)
{
   u16 SensorNumberCheck;
   u8 MUX;

   //debug message
   TxData[0]=0x05;
//   TxMsg();//debug

   if((RxFifo0ITs & FDCAN_IT_RX_FIFO0_NEW_MESSAGE) != RESET)
   {
      /* Retreive Rx messages from RX FIFO0 */
      if (HAL_FDCAN_GetRxMessage(hfdcan, FDCAN_RX_FIFO0, &RxHeader, CANRx) != HAL_OK)
      {
      /* Reception Error */
      Error_Handler();
      }

      //Format of the message is:
      //BYTE 0..1 = SENSOR NUMBER U16
      //BYTE 2    = CODE CHECK1
      //BYTY 3    = CODE CHECK2
      //BYTE 4    = MUX - TYPE OF MESSAGE U8
      //BYTE 5..6 = U16 Value
      //BYTE 7..10= Floating point value - N.B. CAN-FD Only

      //Check the sensor number agrees
      U16Value.b[1]=CANRx[0];
      U16Value.b[0]=CANRx[1];
      SensorNumberCheck=U16Value.d;
      MUX=CANRx[4];
      //Only allow this sensor number to be updated  - otherwise any other sensor on the bus will!
      if(SensorNumberCheck==Config_Msg.SENSORNO)//check whole sensor number
      {

and so on...FIFO1 callback:

//CANRx Interrupt processing
void HAL_FDCAN_RxFifo1Callback(FDCAN_HandleTypeDef *hfdcan, uint32_t RxFifo1ITs)
{
    u16 SensorNumberCheck;
    u8 MUX;

// HAL_FDCAN_AddMessageToTxFifoQ(&hfdcan1, &TxHeaderDebug, TxDataSlowRow);
    //debug message
//    TxData[0]=0x10;
    TxMsg();

  if((RxFifo1ITs & FDCAN_IT_RX_FIFO1_NEW_MESSAGE) != RESET)
  {
    // Retreive Rx messages from RX FIFO0 //
    if (HAL_FDCAN_GetRxMessage(hfdcan, FDCAN_RX_FIFO1, &RxHeader, CANRx) != HAL_OK)
    {
    // Reception Error //
    Error_Handler();
    }

    //Format of the message is:
    //BYTE 0..1 = SENSOR NUMBER U16
    //BYTE 2    = CODE CHECK1
    //BYTY 3    = CODE CHECK2
    //BYTE 4    = MUX - TYPE OF MESSAGE U8
    //BYTE 5..6 = U16 Value
    //BYTE 7..10= Floating point value

I wil

Haithem Rahmani · ‎2025-10-24

Hi @Rob Ashworth

Try to configure ThreadX to use the BASEPRI register, as shown below:

set the 'TX_BASEPRI_VALUE' to a value different from 0, then the CAN interrupts won't be masked.

regards
Haithem.