The code for the PIC32 in portmacro.h does a read-modify-write on CP0 to control stuff like interrupts, which does not seem safe. For example this code enables interrupts.
ulStatus = _CP0_GET_STATUS()
ulStatus &= ~portALL_IPL_BITS;
_CP0_SET_STATUS( ulStatus );
why do that rather than this
_CP0_BIC_STATUS(portALL)IPL_BITS);
PIC32 CP0 operations
PIC32 CP0 operations
The code generated by your suggestion is:
9D000020 40026000 mfc0 v0,Status << read
9D000024 00401821 addu v1,v0,zero
9D000028 3C02FFFF lui v0,0xffff
9D00002C 344203FF ori v0,v0,0x3ff
9D000030 00621024 and v0,v1,v0
9D000034 40826000 mtc0 v0,Status << write so it is still performing a read/modify/write sequence so I’m not sure what is to be gained. That said, the original code is probably less than ideal, however I don’t think it is problematic because each task has its own Status value, and its own stack onto which ulStatus is stored. Regards.
9D000024 00401821 addu v1,v0,zero
9D000028 3C02FFFF lui v0,0xffff
9D00002C 344203FF ori v0,v0,0x3ff
9D000030 00621024 and v0,v1,v0
9D000034 40826000 mtc0 v0,Status << write so it is still performing a read/modify/write sequence so I’m not sure what is to be gained. That said, the original code is probably less than ideal, however I don’t think it is problematic because each task has its own Status value, and its own stack onto which ulStatus is stored. Regards.
PIC32 CP0 operations
Thanks for the reply, I should have looked at the assembly first.
I’m still left wondering how atomic the operation is, but the is independent of FreeRTOS. Eric
I’m still left wondering how atomic the operation is, but the is independent of FreeRTOS. Eric