osdelay stucked after I called HAL_FLASHEx_Erase and HAL_FLASH_Prog in task - stm32

Question

I am porting LittleFS on STM32 G431Rb internal Flash. Every thing is OK when I read and write file on main function. But when I write some thing in Task, System will be stuck on osdlelay.
/* USER CODE END Header_StartDefaultTask */
void StartDefaultTask(void *argument)
{
/* USER CODE BEGIN 5 */
/* Infinite loop */
int i = 0;
for(;;)
{
ULOG_TRACE("Trace count = %d",i);
i++;
osDelay(5);
}
/* USER CODE END 5 */
}
In ULOG_TRACE function, I called lfs_fs_write function.
lfs_fs_write function call HAL_FLASHEx_Erase and HAL_FLASH_Prog.
static FLASH_EraseInitTypeDef EraseInitStruct;
int stm32_interl_flash_block_erase(const struct lfs_config *c, lfs_block_t block)
{
uint32_t PageError;
__disable_irq();
HAL_FLASH_Unlock();
__HAL_FLASH_CLEAR_FLAG(FLASH_FLAG_EOP |
FLASH_FLAG_OPERR |
FLASH_FLAG_PROGERR |
FLASH_FLAG_WRPERR |
FLASH_FLAG_PGAERR |
FLASH_FLAG_SIZERR |
FLASH_FLAG_PGSERR |
FLASH_FLAG_MISERR );
EraseInitStruct.TypeErase = FLASH_TYPEERASE_PAGES;
EraseInitStruct.Banks = FLASH_BANK_1;
EraseInitStruct.Page = FS_BASE_PAGE_START + block;
EraseInitStruct.NbPages = 1;
if (HAL_FLASHEx_Erase(&EraseInitStruct,&PageError)!= HAL_OK){
__enable_irq();
return HAL_FLASH_GetError();
}
HAL_FLASH_Lock();
__enable_irq();
return 0;
}
int stm32_interl_flash_block_prog(const struct lfs_config *c, lfs_block_t block,
lfs_off_t off, const void *buffer, lfs_size_t size)
{
__disable_irq();
HAL_FLASH_Unlock();
__HAL_FLASH_CLEAR_FLAG(FLASH_FLAG_EOP |
FLASH_FLAG_OPERR |
FLASH_FLAG_PROGERR |
FLASH_FLAG_WRPERR |
FLASH_FLAG_PGAERR |
FLASH_FLAG_SIZERR |
FLASH_FLAG_PGSERR |
FLASH_FLAG_MISERR );
uint32_t dest_addr = FS_BASE_ADDR + c->block_size*block +off;
uint64_t *pSrc = (uint64_t*)buffer;
uint32_t write_size = 0;
while(write_size < size){
if (HAL_FLASH_Program(FLASH_TYPEPROGRAM_DOUBLEWORD,dest_addr,*(pSrc)) != HAL_OK){
HAL_FLASH_Lock();
__enable_irq();
return HAL_FLASH_GetError();
}
pSrc++;
dest_addr += 8;
write_size += 8;
}
HAL_FLASH_Lock();
__enable_irq();
return 0;
}
I googled these problem, some guys said the problem is that the task schedule interrupt occur when I erase or prog internal flash.But I add disable_irq, it also have the problem.

Answer 1

Do not disable all interrupts. At least the HAL tick should run further. The HAL_FLASH_* methods depend on them.
Use osKernelSuspend() and osKernelResume() to stop and continue thread scheduling.
Further more check all return codes, also from HAL_FLASH_Unlock.
The flash peripheral is a little nasty on the STM32 controller family.
If something going wrong, either you access addresses unaligned, out of bounds or simple not in the correct order the peripheral set its error bits and the HAL functions do not operate anymore. You already clear the error flags.
I've experienced that checking and clearing the error flags also after HAL_FLASH_Program is always a good idea.
PS: If your code really stuck in osDelay then your cmsis os scheduler interrupt is not running or the scheduler is disabled/stoped.

Answer 2

You haven't specified how long the code gets stuck for or why you think it is stuck in osDelay, but here is one guess:
If you are writing or erasing in the same flash bank as you are executing code from, your code will stop until the operation has completed. This will block all tasks and all interrupts whether they are turned on or off. This is because no instructions can be fetched from flash while it is writing or erasing. A few instructions might still execute if they have already been prefetched.
If you are able to use dual-bank mode you can write one bank while executing from the other, but I think the 128kB parts might only support single bank flash, and others may have dual-bank disabled by default.