Hard fault when exiting stop mode with RTC alarm on STM32F051

[storzinc]

There is already one nop after wfi, if I add another one - nothing changes, the program still crash and still in the middle of the bl, but if I add 3 or more nops, or remove all of them - program stops crashing. The function call seems to be important in the reproduction of this crash, I firstly tried to move __disable_irq() directly in the body of the waiting loop, but with that the crash is also disappears, no matter how many nops I add around it, so I moved it in its own noinline function.

[storzinc]

Tried this on the STM32F318K8U6 board and there everything works fine, without hard faults. I can not be 100% sure though, the problem is tricky to reproduce, with all these nops in the right places, maybe I just did not made the timing right.

[storzinc]

I have reproduced this problem without chibios, using STM32 HAL and STM32CubeIDE 1.0.2, it also turned out that the function call is unnecessary for reproduction and I further simplified RTC_Wait:

Code: Select all

void RTC_Wait()
{
nop();
//nop();

  lexit=0;

  PWR->CR|=PWR_CR_LPDS;              // enter stop mode
  SCB->SCR|=SCB_SCR_SLEEPDEEP_Msk;   //
  while(1)
  {
nop();
    __disable_irq();
    if(lexit)
    {
      __enable_irq();
      break;
    }
    __WFI();
nop();
//nop();
    __enable_irq();
  }
  SCB->SCR&=~SCB_SCR_SLEEPDEEP_Msk;
}

Now the hard fault happens at the address 8000e24, and there is actually no instructions at all!

Code: Select all

 8000e16:   bf30         wfi
 8000e18:   46c0         nop         ; (mov r8, r8)
 8000e1a:   b662         cpsie   i
 8000e1c:   e7f0         b.n   8000e00 <RTC_Wait+0x1c>
 8000e1e:   46c0         nop         ; (mov r8, r8)
 8000e20:   20000020    .word   0x20000020
 8000e24:   40007000    .word   0x40007000
 8000e28:   e000ed00    .word   0xe000ed00

Now it really looks like a hardware bug, it seems as if the execution continues from the wrong address.

If someone is interested in reproducing this my full STM32CubeIDE project is in the attachment.

[Giovanni]

Hi,

I would try to post this on the ST forum, it is possible they already handled this problem. Of course we would be interested in hearing the solution of this mystery

Giovanni