Skip to main content
TomC
Senior
Mar 15, 2024 3:29 AM UTC
Solved

Hardfault (FORCED, IMPRECISERR) when erasing flash page. [STM32WB55]

Hi there,

I am receiving a hardfault when erasing pages of flash, seemingly randomly. I can recreate the issue somewhat consistently if I rapidly erase flash pages. I have followed the example project from the STM32WB firmware package (erase flash example). 

This is my function for erasing flash.

 

flash_status_t user_flash_erase_page(uint8_t page_number, uint8_t number_of_pages)
{
// static FLASH_EraseInitTypeDef EraseInitStruct = {0};
 static FLASH_EraseInitTypeDef EraseInitStruct = {0};

 if(HAL_FLASH_Unlock()!= HAL_OK)
 {
 return FLASH_ERROR;
 }

 /* Clear OPTVERR bit set on virgin samples */
 __HAL_FLASH_CLEAR_FLAG(FLASH_FLAG_OPTVERR);

 EraseInitStruct.TypeErase = FLASH_TYPEERASE_PAGES;
 EraseInitStruct.Page = page_number;
 EraseInitStruct.NbPages = number_of_pages;

 uint32_t page_error = 0;

 if (HAL_FLASHEx_Erase(&EraseInitStruct, &page_error) != HAL_OK)
 {
 HAL_FLASH_Lock();
 return FLASH_ERROR;
 }
 else
 {
 HAL_FLASH_Lock();
 return FLASH_OK;
 }

 return FLASH_OK;
}

 

 

Specifically I will hardfault when trying to perform HAL_FLASH_Lock( ) when HAL_FLASHEx_Erase fails (line 23 in the code snippet above)

 

if (HAL_FLASHEx_Erase(&EraseInitStruct, &page_error) != HAL_OK)
{
 HAL_FLASH_Lock();
 return FLASH_ERROR;
}

 

 
From the reference manual (RM0434 - Section 3.10.5 FLASH_CR) for the STM32WB I see that 

"This register (FLASH_CR) must not be modified when CFGBSY in Flash memory status register
(FLASH_SR) is set. This would result in... a bus error, if CFGBSY is set due to a double-word programming sequence and the write to FLASH_CR is performed after the first word is sent for programming and before
the second."

The hardfault I receive is a bus error

TomC_0-1710470822320.png

Further, when I inspect the CFGBSY bit in the FLASH_SR register I find that it is set

TomC_1-1710470921502.png

TomC_2-1710471950862.png

 

So I figure the fault occurs because the program tries to modify the FLASH_CR register when the CFGBSY bit is set in the HAL_FLASH_Lock( ) function

 

/**
 * @brief Lock the FLASH control register access.
 * @retval HAL Status
 */
HAL_StatusTypeDef HAL_FLASH_Lock(void)
{
 HAL_StatusTypeDef status = HAL_OK;

 /* Set the LOCK Bit to lock the FLASH Registers access */
 /* @Note The lock and unlock procedure is done only using CR registers even from CPU2 */
 SET_BIT(FLASH->CR, FLASH_CR_LOCK);

 /* verify Flash is locked */
 if (READ_BIT(FLASH->CR, FLASH_CR_LOCK) == 0U)
 {
 status = HAL_ERROR;
 }

 return status;
}

 


The specific pages of flash that I am modifying (erasing) are 167 - 170. According to the memory map for the STM32WB55 this is okay

TomC_3-1710473337783.png

 

This topic has been closed for replies.
TDK
TDKBest answer
Mar 15, 2024 1:55 PM UTC

Why is HAL_FLASHEx_Erase failing? By that I mean what is in pFlash.ErrorCode and what address is in page_error when the call returns HAL_ERROR?

"If you feel a post has answered your question, please click ""Accept as Solution""."

4 replies

TDK
TDKBest answer
Mar 15, 2024 1:55 PM UTC

Why is HAL_FLASHEx_Erase failing? By that I mean what is in pFlash.ErrorCode and what address is in page_error when the call returns HAL_ERROR?

"If you feel a post has answered your question, please click ""Accept as Solution""."
TomC
TomCAuthor
Senior
Mar 17, 2024 8:40 PM UTC

Hi TDK,

Thank you for your response, I believe the problem was caused by me overwritting some memory with memcpy() somewhere else hence the IMPRECISERR. I will keep in mind to check the error code next time this happens.

Joe WILLIAMS
ST Employee
Mar 16, 2024 4:49 AM UTC

Hi TomC

Your case is being reviewed through the online support portal.

Thank you for your patience.

 

Kind Regards

Joe WILLIAMS

STMicro Support

TomC
TomCAuthor
Senior
Mar 17, 2024 8:40 PM UTC

Hi Joe,

I believe I caused this problem by overwritting memory with a memcpy () elsewhere in my program. Thank you for your response.