uart_ERROR callback triggered but errorCode is zero

[DavidWahlberg]

I am programming a STM32 nucleo-f746ZG development board where I am using USART2 with RX interrupt.

I have setup USART using CubeMx (async, baud:250Khz, data: 8, stopBit: 2, noParity). I am reading a single byte from a DMX stream by calling HAL_UART_Receive_IT(&huart2, rx_buff, 1);

Furthermore, I have declared the two callbacks: HAL_UART_RxCpltCallback() and HAL_UART_ErrorCallback().

I am using the error_callback to detect whenever there is a BREAK in the DMX-stream (afterwards 512 bytes will be recieved) which will trigger an FRAME_ERROR.

The problem is that when the HAL_UART_Error_callback is called, the errorCode from the uart (error = huart->ErrorCode) is returning 0. I am assuming that if the error_callback is being triggered, there must be a value in ErrorCode != 0.

I will check if the error is a frameError and not an overrun error for instance.

 

I hope someone has an answer to this :)

[Sarra.S]

Hello @DavidWahlberg 

 >>I am assuming that if the error_callback is being triggered, there must be a value in ErrorCode != 0.

There are different values of ErrorCode indicating different error types. 

You can do something like this to determine what error occurred to handle it appropriately.

 

void HAL_UART_ErrorCallback(UART_HandleTypeDef *huart) {
 if (huart->ErrorCode & HAL_UART_ERROR_PE) {
 // parity error
 }
 if (huart->ErrorCode & HAL_UART_ERROR_FE) {
 // frame error
 }
 // etc...
}

 Refer to UART Error Code  to check each error code  

[DavidWahlberg]

Thank you for fast reply!

I am aware that i can check each type of error. The problem is that reading the value of huart->ErrorCode is 0, and because of that, none of the if statements will be true.

my error callback is:

void HAL_UART_ErrorCallback(UART_HandleTypeDef *huart)
{

	if(huart == &huart2)
	{
		uint32_t error = huart->ErrorCode;

		if(error & HAL_UART_ERROR_FE)
		{
			//frame error detected
			dmxCounter = 0; //reset counter
			__HAL_UART_CLEAR_FLAG(huart, UART_FLAG_FE); //clear frame error flag
			HAL_GPIO_TogglePin(LD2_GPIO_Port, LD2_Pin); // toggle led to indicate frame error

		}

		if(error & HAL_UART_ERROR_ORE)
		{
			//overrun error
			__HAL_UART_CLEAR_FLAG(huart, UART_FLAG_ORE); //clear overrun flag
			__HAL_UART_FLUSH_DRREGISTER(huart); //flush data
		}

		HAL_UART_Receive_IT(&huart2, byteRead_buff, 1); // restart uart RX
	}

}

[TDK]

Set a breakpoint and step through HAL_UART_IRQHandler to understand where ErrorCode gets set when an error is triggered.

Looks like it gets set correctly here to me:

https://github.com/STMicroelectronics/stm32f7xx_hal_driver/blob/87047e59aa3883da281a7de519723545e6be4513/Src/stm32f7xx_hal_uart.c#L2171

 

[DavidWahlberg]

Hi

I managed to find the issue by stepping.

The RxCompleteteCallback will, what I have experienced, always be called first and afterwards ErrorCallback only if

HAL_UART_Receive_IT(&huart2, byteRead_buff, 1) (1 byte).

I was at the very end of RxCompleteCallback calling HAL_UART_Receive_IT(&huart2, byteRead_buff, 1) to init a new interrupt, which reset the huart->errorCode variable. And because of that the value was always 0 when the ErrorCallback was called afterwards.

However, If calling HAL_UART_Receive_IT(&huart2, byteRead_buff, 2) (read: >1 byte) the errorCallback will be called first (if an error happened) and no RxCompleteCallback.

In my usecase I need to read every single byte one at a time to look for a frameError to know when a complete new DMX frame will come.

My solution is, since the RxCompleteCallback is always being call first, to read the huart->errorCode in the RxCompleteCallback, do the logic (frameError true or not) and at the end call HAL_UART_Receive_IT(&huart2, byteRead_buff, 1). This means no code written in uartErrorCallback.

I hope it makes sense :)