STM32cubemx-HAL库串口断线问题

STM32cubemx：version5.1

Chip: STM32F446RE

IDE:Keil5

Q：小项目上写了个简单的通信包，波特率230400，数据量较大1600Byte/s,DMA的方式实现接收，量产后发现跑久了部分机器会有只能发送不能接收的问题。

查了很久没查到。中间加了断线检测，重新初始化串口，却无法解决。最后面发现，HAL的锅。

部分代码：

void HAL_UART_RxCpltCallback(UART_HandleTypeDef *huart)
{        
    if(huart->Instance == UART4) {
           uint8_t ret=HAL_OK;
        if(frame_in_sync) {
            if(UART_Rx_Buffer.head == 0x55 && UART_Rx_Buffer.end == 0xAA) { // TODO: check check-sum
                UART_Rx_Data = UART_Rx_Buffer;
                ret=HAL_UART_Receive_IT(&huart4, (uint8_t *)&UART_Rx_Buffer, sizeof(UART_Rx_Buffer));
                uart_updata_time= HAL_GetTick();
                            
            } else { // frame not in sync, drop coming bytes until read an end(0xAA)
                frame_in_sync = 0;
                ret=HAL_UART_Receive_IT(&huart4, (uint8_t *)&UART_Rx_Buffer, 1);
            }
        } else {
             　if( *(uint8_t *)&UART_Rx_Buffer == 0xAA) {
                frame_in_sync = 1;
                ret=HAL_UART_Receive_IT(&huart4, (uint8_t *)&UART_Rx_Buffer, sizeof(UART_Rx_Buffer));
            } else {
                ret=HAL_UART_Receive_IT(&huart4, (uint8_t *)&UART_Rx_Buffer, 1);
            }
        }
    }
}

上述的代码绝大多数时间都能够正常运行。但是，总会出现莫名奇妙的掉线问题。并且难以复线。查了好久，看到有类似的问题，HAL库会不处理异常的行为，要用户处理。

在某些情况下，HAL_UART_Receive_IT 并没有成功执行，返回了busy 或者erro，这时候整个流程就背打断了。并且串口还会被莫名其妙锁上。

void HAL_UART_RxCpltCallback(UART_HandleTypeDef *huart)
{        
    if(huart->Instance == UART4) {
                uint8_t ret=HAL_OK;
        if(frame_in_sync) {
            if(UART_Rx_Buffer.head == 0x55 && UART_Rx_Buffer.end == 0xAA) { // TODO: check check-sum
                UART_Rx_Data = UART_Rx_Buffer;
                ret=HAL_UART_Receive_IT(&huart4, (uint8_t *)&UART_Rx_Buffer, sizeof(UART_Rx_Buffer));
                uart_updata_time= HAL_GetTick();
                            
            } else { // frame not in sync, drop coming bytes until read an end(0xAA)
                frame_in_sync = 0;
                ret=HAL_UART_Receive_IT(&huart4, (uint8_t *)&UART_Rx_Buffer, 1);
            }
        } else {
            if( *(uint8_t *)&UART_Rx_Buffer == 0xAA) {
                frame_in_sync = 1;
                ret=HAL_UART_Receive_IT(&huart4, (uint8_t *)&UART_Rx_Buffer, sizeof(UART_Rx_Buffer));
            } else {
                ret=HAL_UART_Receive_IT(&huart4, (uint8_t *)&UART_Rx_Buffer, 1);
            }
        }
                while(ret!=HAL_OK){
                     /*try to fix connection bug  20191225*/
                      printf("UART4:RECEIVE_IT:%d
",ret);　　//for log 
                      huart->RxState = HAL_UART_STATE_READY;
                      __HAL_UNLOCK(huart);
                      ret=HAL_UART_Receive_IT(&huart4, (uint8_t *)&UART_Rx_Buffer, 1); 
                }
    }
}


void HAL_UART_ErrorCallback(UART_HandleTypeDef *huart)
{
  /* Prevent unused argument(s) compilation warning */
    if(huart->ErrorCode&HAL_UART_ERROR_ORE)
    {
        __HAL_UART_CLEAR_OREFLAG(huart);
    }
  /* NOTE : This function should not be modified, when the callback is needed,
            the HAL_UART_ErrorCallback can be implemented in the user file.
   */
}

修改后，出现busy或者erro时强制解锁，重新接收。根据log，大约半小时会出现一次。三台经常性出现的机子更新后运行一天，没有出现这个问题。在log中查看到了对应的修正过程。