What is the behavior of BASEPRI==1 and PRIGROUP == 0?

 Hi ARM expert,

When I am porting FreeRTOS to my Cortex M33 CPU, I met a secure hard fault exception because of running SVC instruction with BASEPRI_NS.BASEPRI[7:0] setting to 1.

I am curious why there is hard fault, because I think the value in BASEPRI is not equal or lower than the SVCall Priority Number which is configured in SHPR2.PRI_11. Let me explain my understanding here:

 

In my environment:

  • The SVCall priority is not configured, which is using default value 0. That means SHPR2.PRI_11 have defaulted all 0 value.
  • The __NVIC_PRIO_BITS is 3, that is only 3 bits in BASEPRI_NS.BASEPRI, bits BASEPRI_NS.BASEPRI[7:5] are valid, and the lower BASEPRI_NS.BASEPRI[4:0] are all zero.
  • The AIRCR.PRIGROUP is configured as 0b000, that is, Group priority [7:1], subpriority [0]. As my understanding, because __NVIC_PRIO_BITS==3, and PRIGROUP==0, in this case, all interrupts’ subpriority are 0, that means:

Configured interrupt priority == group priority,

Right?

If so, when BASEPRI_NS.BASEPRI[7:5] is set to 1, then the interrupts with priority <= 1 (on the contrary, with interrupt priority value >= 1) will be masked, and interrupts with priority value 0 (which has higher priority than BASEPRI) will not be masked.

But from my experimental phenomenon, it seems that when BASEPRI_NS.BASEPRI[7:5] is set to 1, the SVCall interrupt with priority value 0 is also treated as equal or lower priority than BASEPRI==1, which cause the hard fault.

Could you please help to explain it? And what is exact comparison formula in my case? Thank you very much!

Parents
  • Hi, your understand is correct. But as your setting is BASEPRI_NS, it only effects the non-secure interrupt. While there is a secure hard fault happening, it doesn't seem due to BASEPRI_NS. Could you please provide more information about that, like the code section about triggering exceptions and tarmac file? Thanks

Reply
  • Hi, your understand is correct. But as your setting is BASEPRI_NS, it only effects the non-secure interrupt. While there is a secure hard fault happening, it doesn't seem due to BASEPRI_NS. Could you please provide more information about that, like the code section about triggering exceptions and tarmac file? Thanks

Children
No data