why is xPortStartScheduler proto not portable

Kernel
1

imajeff wrote on Tuesday, April 18, 2006:

Strange, I don’t see why the prototype for xPortStartScheduler() is not in portable code (i.e. “Source/port/XXX/portmacro.h”).

It seems to me the prototype for xPortStartScheduler() obviously needs to be just as portable as is the prototype for vPortTickInterrupt() which IS in "portmacro.h".

So we shouldn’t have to design this sort of nonsense:

==========
portBASE_TYPE xPortStartScheduler( void )
{
    /* xPortStartScheduler() does not start the scheduler directly because
    the header file containing the xPortStartScheduler() prototype is part
    of the common kernel code, and therefore cannot use the CODE_SEG pragma.
    Instead it simply calls the locally defined xNearStartScheduler() -
    which does use the CODE_SEG pragma. */

    return xBankedStartScheduler();
}

2

nobody wrote on Tuesday, April 18, 2006:

xPortStartScheduler prototype is in portable.h?  Don’t understand.  ARM port it is started directly and prototype for vPortTickInterrupt is not in portmacro.h.

3

imajeff wrote on Thursday, May 18, 2006:

I almost forgot to respond since I wasn’t clear enough the first time…

It seems to me like the solution to the problem would be to put xPortStartScheduler() prototype in "portable.h". See, I want to tell GCC to make it __attribute__((near)), but I cannot because xPortStartScheduler() is NOT in "portable.h".

Here is the workaround they had to do for CodeWarrior compiler:

portBASE_TYPE xPortStartScheduler( void )
{
    /* xPortStartScheduler() does not start the scheduler directly because
    the header file containing the xPortStartScheduler() prototype is part
    of the common kernel code, and therefore cannot use the CODE_SEG pragma.
    Instead it simply calls the locally defined xBankedStartScheduler() -
    which does use the CODE_SEG pragma. */

    return xBankedStartScheduler();
}