I did a port on plasma mcu (MIPS I instruction set from opencores.com).
I made it by merging demos from PIC32 and ARM CM3. I implemented both configKERNEL_INTERRUPT_PRIORITY and configMAX_SYSCALL_INTERRUPT_PRIORITY. But a difference about the two port is than PIC32 uses portCRITICAL_NESTING_IN_TCB=1 whereas ARM-CM3 uses a global static uxInterruptNesting variable, so my question is why to use portCRITICAL_NESTING_IN_TCB=1? Indeed it uses more RAM to store uxInterruptNesting for each task. What is its benefit?
Thank in advance
Santo
The Cortex M3 is a bit unique in that a context switch will only occur when the critical nesting count is zero. Most ports (all other ports?) store a critical nesting count per task. Whether this is stored in the TCB or task stack does not make a difference to the amount of RAM required. Storing it in the TCB is much simpler (less context switch code) but can also be less flexible and less efficient - so it really depends on the architecture.