Using DMA with UART on STM32H753ZI and getting HAL_DMA_ERROR_TE (Transmission Error)

[KEria.1]

Hello,

I am trying to set the UART5 TX with the DMA. I am using Cube, and in the GUI I selected UART5 to asynchronous, and added DMA for UART5_TX. Then, in the generated code, I added this part as I didn't see initialization for the DMA (inside the MX_DMA_Init()):

  /* Peripheral DMA init*/
  hdma_uart5_tx.Instance = DMA1_Stream0;
  //hdma_uart5_tx.Init.Channel = DMA_CHANNEL_4;
  hdma_uart5_tx.Init.Direction = DMA_MEMORY_TO_PERIPH;
  hdma_uart5_tx.Init.PeriphInc = DMA_PINC_DISABLE;
  hdma_uart5_tx.Init.MemInc = DMA_MINC_ENABLE;
  hdma_uart5_tx.Init.PeriphDataAlignment = DMA_MDATAALIGN_BYTE;
  hdma_uart5_tx.Init.MemDataAlignment = DMA_MDATAALIGN_BYTE;
  hdma_uart5_tx.Init.Mode = DMA_NORMAL;
  hdma_uart5_tx.Init.Priority = DMA_PRIORITY_MEDIUM;
  hdma_uart5_tx.Init.FIFOMode = DMA_FIFOMODE_DISABLE;
    HAL_DMA_Init(&hdma_uart5_tx);
 
    __HAL_LINKDMA(&huart5,hdmatx,hdma_uart5_tx);

The Main function, I didn't do much except I added this code in the while(1):

while (1)
  {
    /* USER CODE END WHILE */
 
    /* USER CODE BEGIN 3 */
	  if (HAL_UART_Transmit_DMA(&huart5,au8ch, 1) == HAL_OK)
	  {
		  HAL_GPIO_WritePin(GPIOE, GPIO_PIN_1, GPIO_PIN_SET);
	  }
	  else
	  {
		  HAL_GPIO_WritePin(GPIOE, GPIO_PIN_1, GPIO_PIN_RESET);
	  }
	  HAL_Delay(1000);
  }
  /* USER CODE END 3 */

Which is a code to transmit the first byte of an array. I connected the Nucleo-H753ZI board pins to a terminal software, and I didn't get anything. (Note that I was able to get something without using the DMA). The LED turned ON, which means the function is returning HAL_OK, but still nothing on the screen.

I tried to debug the code, I put a breakpoint in the function "void HAL_DMA_IRQHandler(DMA_HandleTypeDef *hdma)" and it stops there, which means it got the interrupt to send something, but then it goes into this autogenerated code:

if(IS_DMA_STREAM_INSTANCE(hdma->Instance) != 0U)  /* DMA1 or DMA2 instance */
  {
    /* Transfer Error Interrupt management ***************************************/
    if ((tmpisr_dma & (DMA_FLAG_TEIF0_4 << (hdma->StreamIndex & 0x1FU))) != 0U)
    {
      if(__HAL_DMA_GET_IT_SOURCE(hdma, DMA_IT_TE) != 0U)
      {
        /* Disable the transfer error interrupt */
        ((DMA_Stream_TypeDef   *)hdma->Instance)->CR  &= ~(DMA_IT_TE);
 
        /* Clear the transfer error flag */
        regs_dma->IFCR = DMA_FLAG_TEIF0_4 << (hdma->StreamIndex & 0x1FU);
 
        /* Update error code */
        hdma->ErrorCode |= HAL_DMA_ERROR_TE;
      }
    }
...
...
...
/* manage error case */
    if(hdma->ErrorCode != HAL_DMA_ERROR_NONE)
    {
      if((hdma->ErrorCode & HAL_DMA_ERROR_TE) != 0U)
      {
        hdma->State = HAL_DMA_STATE_ABORT;
 
        /* Disable the stream */
        __HAL_DMA_DISABLE(hdma);
 
        do
        {
          if (++count > timeout)
          {
            break;
          }
        }
        while((((DMA_Stream_TypeDef   *)hdma->Instance)->CR & DMA_SxCR_EN) != 0U);
 
        /* Process Unlocked */
        __HAL_UNLOCK(hdma);
 
        if((((DMA_Stream_TypeDef   *)hdma->Instance)->CR & DMA_SxCR_EN) != 0U)
        {
          /* Change the DMA state to error if DMA disable fails */
          hdma->State = HAL_DMA_STATE_ERROR;
        }
        else
        {
          /* Change the DMA state to Ready if DMA disable success */
          hdma->State = HAL_DMA_STATE_READY;
        }
      }
 
      if(hdma->XferErrorCallback != NULL)
      {
        /* Transfer error callback */
        hdma->XferErrorCallback(hdma);
      }
    }
  }

Any idea what I may be missing?

[waclawek.jan]

https://community.st.com/s/article/FAQ-DMA-is-not-working-on-STM32H7-devices

[TDK]

Make sure the au8ch buffer is somewhere accessible by DMA. What memory address is it at?

[KEria.1]

0x2001fff4

I initialized it at the beginning of the main()

[TDK]

That address is in DTCM, which is not accessible to DMA.
https://community.st.com/s/article/FAQ-DMA-is-not-working-on-STM32H7-devices

[KEria.1]

I was reading this link, I tried placing it outside the main as a global var but it was 0x20000000 which is still the same domain.

How can it place it elsewhere?

[KEria.1]

I added this part to my code:

but still getting the same error, the array is now located at 0x20000010 which is still in the same range.

#if defined( __ICCARM__ )
  #define DMA_BUFFER \
      _Pragma("location=\".dma_buffer\"")
#else
  #define DMA_BUFFER \
      __attribute__((section(".dma_buffer")))
#endif
 
DMA_BUFFER unsigned char au8ch[4] = {70,54,35,4};

[waclawek.jan]

Have you read the article I gave link to? Read it. Reread it. For gcc (which is the default compiler for CubeIDE), you have to read the section about the linker script.

JW

[KEria.1]

Thanks, I read it, and reread it, and re-reread it.. Here is what I did:

in the file: STM32H753ZITX_RAM.ld I changed this code (I put what I changed in the comments below in the file) as recommended in solutions 1 and 2.

I only deployed solution 1 at first and it didn't fix it, so I added this section at the end.

  .data :
  {
    . = ALIGN(4);
    _sdata = .;        /* create a global symbol at data start */
    *(.data)           /* .data sections */
    *(.data*)          /* .data* sections */
 
    . = ALIGN(4);
    _edata = .;        /* define a global symbol at data end */
  } >RAM_D1 AT> FLASH /* was >RAM AT> RAM_EXEC*/
 
  /* Uninitialized data section */
  . = ALIGN(4);
  .bss :
  {
    /* This is used by the startup in order to initialize the .bss secion */
    _sbss = .;         /* define a global symbol at bss start */
    __bss_start__ = _sbss;
    *(.bss)
    *(.bss*)
    *(COMMON)
 
    . = ALIGN(4);
    _ebss = .;         /* define a global symbol at bss end */
    __bss_end__ = _ebss;
  } >RAM_D1 AT> FLASH /* was >RAM*/
 
  /* User_heap_stack section, used to check that there is enough RAM left */
  ._user_heap_stack :
  {
    . = ALIGN(8);
    PROVIDE ( end = . );
    PROVIDE ( _end = . );
    . = . + _Min_Heap_Size;
    . = . + _Min_Stack_Size;
    . = ALIGN(8);
  } >RAM_D1 AT> FLASH /* was >RAM*/
 
/* I also added this part: */
  .dma_buffer : 
  {
    *(.dma_buffer)
  } >RAM_D2

Then I added the code in the main.c in the previous comment.

#if defined( __ICCARM__ )
  #define DMA_BUFFER \
      _Pragma("location=\".dma_buffer\"")
#else
  #define DMA_BUFFER \
      __attribute__((section(".dma_buffer")))
#endif
 
DMA_BUFFER unsigned char au8ch[4] = {70,54,35,4};

But still, I got nothing. Any advice?

[waclawek.jan]

> But still, I got nothing.

That's an equivalent of "it does not work" and is not helpful.

So, how are your variables allocated now? What's in the mapfile? How is RAM_D2 set? Do you change the linker script you are actually using?

I don't use Cube/CubeIDE/Eclipse so can't really help in this detail.

JW