I ran into a problem concerning the configMAX_SYSCALL_INTERRUPT_PRIORITY value.
I’m using the EFM32 Cortex-M3 port of FreeRTOS V6.1.0. Since I didn’t want to care about the interrupt priorities, I thought it would be a good idea to set configMAX_SYSCALL_INTERRUPT_PRIORITY to 0 (the highest interrupt priority). But this leads to a sporadically crashing OS. Further examination of the problem showed that the configMAX_SYSCALL_INTERRUPT_PRIORITY is used in the function vPortSetInterruptMask to set the BASEPRI register. But I think this has not the intended effect since setting BASEPRI to zero actually enables all interrupts (disables the BASEPRI functionality) instead of disabling them.
Can anyone confirm this concern?
Yes I can confirm it. Don’t set it to 0 because zero has a special meaning in basepri, as you found out yourself.
I just got bitten by not configuring the priorities of my interrupts, so I did some research to understand ARM Cortex interrupt priorities and how they affect FreeRTOS.
One thing I came across while researching was this tidbit int he ARM v7-M architecture reference manual: “When (BASEPRI) is cleared to 0, it has no effect on the priority. A non-zero value will act as a priority mask, the execution priority when the priority defined by the BASEPRI is higher than the current executing priority”.
So basically, you’re right, setting configMAX_SYSCALL_INTERRUPT_PRIORITY to zero actually prevents any interrupts from being “masked”, so none of your interrupt handlers can safely call into any FreeRTOS APIs.
Here are some relevant links on the FreeRTOS web site:
(see very last sentence).