Enabling usage_fault exception with division_by_zero fault

[amigneault]

Hi, I am developing on a FRDM-K66F and I am trying to trap floating point exception by generating an interrupt.

I understand that it should be a usage fault interrupt that will be triggered.

The sample code below show that I am enabling the usage_fault interrupt, I set a priority then I enable all possible floating point exceptions in the Interrupt Status Register (MCM_ISCR) .

If I run this part of code, after the div/0 I see that the MCM detect it because if I read MCM_ISCR back I see 0x9F120200, FDZC is now set as well as the FPSCR[DZC].

Question: What am I missing to generate the interrupt?  Code goes through, never calling the ISR (which is a while(1) for now.

float toto=1.0;

SCB->SHCSR|=SCB_SHCSR_USGFAULTENA_Msk |SCB_SHCSR_BUSFAULTENA_Msk |SCB_SHCSR_MEMFAULTENA_Msk;

NVIC_SetPriority (UsageFault_IRQn,0x08);

*(uint8_t *)0xe0080010=(uint8_t)0x9F;

toto=toto/0.0F;

Thanks

[Omar_Anguiano · Accepted solution]

Hello

Please add these lines to enable the registers:

NVIC_EnableIRQ(UsageFault_IRQn);
SCB->CCR |= (SCB_CCR_DIV_0_TRP_Msk);
SCB->SHCSR|= SCB_SHCSR_USGFAULTENA_Msk |SCB_SHCSR_BUSFAULTENA_Msk |SCB_SHCSR_MEMFAULTENA_Msk;

The reason why the interruption is not detected is the line "toto=toto/0.0F;"
If you add this function to your code we can see that the Usage fault is called:

static int _DivideByZero(void) {
  int r;
  volatile unsigned int a;
  volatile unsigned int b;

  a = 1.0;
  //          movs r3, #1       <- Load dividend
  b = 0.0;
  //          movs r3, #0       <- Load divisor
  r = a / b;
  //      udiv r3, r2, r3   <- divide by 0 raises fault exception
  return r;
}

If you have more questions do not hesitate to ask me.
Best regards,
Omar