RE: [avr-gcc-list] Question about gcc preprocessing and port/pin assignments

[Dave Hylands]

Hi James,

Thanks for pointing out (off list) that this doesn't actually compile
using gcc - Doh - That'll teach me to post without testing :)

OK. So here's a way that does work under gcc and a non-gcc compiler that
failed the original way. There are a few sublte changes, like requiring
parentheses around the data pin definitions and removal of parenthesis
around the PORT and PIN_MASK usage in say SETPIN.

Adding the extra set of parenthesis

        SETPIN(DATA1) becomes SETPIN((D,7)) which causes only a single
argument of (D,7) to be passed in.
        PORT pindef   become  PORT (D,7) - which is where the
parenthesis come back into play.

//Example defines
#define DATA1         (D, 7)
#define DATA2         (C, 3)
#define DATA3         (B, 3)

//Code to manipulate PORTx, PINx, and DDRx using definitions like 
//those above
#define PORT(port,pin)     (PORT##port)
#define DDR(port,pin)      (DDR##port)
#define         PIN(port,pin)  (PIN##port)
#define PIN_MASK(port,pin) (1<<(pin))

//Set a pin
#define SETPIN(pindef)  PORT pindef |= PIN_MASK pindef 

//Clear a pin
#define CLRPIN(pindef)  PORT pindef &= ~PIN_MASK pindef

//Set a IO to output
#define SET_DDR_OUT(pindef)     DDR pindef |= PIN_MASK pindef

//Set a IO to input
#define SET_DDR_IN(pindef)      DDR pindef &= ~PIN_MASK pindef

//Read the value of a pin
#define GETPIN(pindef)  PIN pindef & PIN_MASK pindef

SETPIN( DATA1 )

So, this should now be completely portable. If you really object to the
parenthesis, then the original 

#define DATA1   D,7
#define SETPIN(pindef)  PORT(pindef) |= PIN_MASK(pindef)

Will work in gcc, but may fail in other compilers. You can make your own
trade off.

--
Dave Hylands
Vancouver, BC, Canada
http://www.DaveHylands.com/