STM32F4XX: DMA HAL busy/ready conditionals are incorrect

[davedesro]

The stm32f4xx has additional states as compared to other stm32 parts.

/** 
  * @brief  HAL DMA State structures definition
  */
typedef enum
{
  HAL_DMA_STATE_RESET             = 0x00U,  /*!< DMA not yet initialized or disabled */
  HAL_DMA_STATE_READY             = 0x01U,  /*!< DMA initialized and ready for use   */
  HAL_DMA_STATE_READY_MEM0        = 0x11U,  /*!< DMA Mem0 process success            */ // FIX
  HAL_DMA_STATE_READY_MEM1        = 0x21U,  /*!< DMA Mem1 process success            */ // FIX
  HAL_DMA_STATE_READY_HALF_MEM0   = 0x31U,  /*!< DMA Mem0 Half process success       */ // FIX
  HAL_DMA_STATE_READY_HALF_MEM1   = 0x41U,  /*!< DMA Mem1 Half process success       */ // FIX
  HAL_DMA_STATE_BUSY              = 0x02U,  /*!< DMA process is ongoing              */
  HAL_DMA_STATE_BUSY_MEM0         = 0x12U,  /*!< DMA Mem0 process is ongoing         */ // FIX
  HAL_DMA_STATE_BUSY_MEM1         = 0x22U,  /*!< DMA Mem1 process is ongoing         */ // FIX
  HAL_DMA_STATE_TIMEOUT           = 0x03U,  /*!< DMA timeout state                   */
  HAL_DMA_STATE_ERROR             = 0x04U,  /*!< DMA error state                     */
  HAL_DMA_STATE_ABORT             = 0x05U,  /*!< DMA Abort state                     */
}HAL_DMA_StateTypeDef;

These state enumeration values for HAL_DMA_STATE_READY and HAL_DMA_STATE_BUSY are set up as maskable flags.

However, stm32f4xx_hal_dma.c assumes natural enumerations in the conditionals. For example:

if(hdma->State != HAL_DMA_STATE_BUSY)
  {
    hdma->ErrorCode = HAL_DMA_ERROR_NO_XFER;
    
    /* Process Unlocked */
    __HAL_UNLOCK(hdma);
    
    return HAL_ERROR;
  }

This introduces potential corner case issues relating to stopping, starting, init/deinit, and setting callbacks.

[davedesro]

Note: stm32f1xx_hal_dma.h also makes this maskable enumeration assumption for READY states, however, currently, the stm32f1xx_hal_dma.c code does not have any conditionals checking for this state, so the F1 part seems to be ok for now.

[LMESTM]

@pliny thanks for reporting the issue.
From what I've seen, your proposal makes sense.
I shared your report and proposed fix to our DMA team and will let you know when we get some feedback.

[betzw]

Pls. refer to this comment of mine!

[LMESTM]

@pliny @betzw As the previous patch from @betzw has been reverted, shall we close this ticket as well ?

[davedesro]

yes, should be closed