Jason Thorpe wrote:
> Now, if we think about the case where you have a FooBar ZapStation,
> which comes in the Whuzzle and Zoobie models, each of which have
> an NCR53c94 on very different busses, you'd have to do dispatching
> for each macro call.  I.e., take the case of ESP_READ_REG():
> 
> static __inline u_char ESP_READ_REG __P((struct esp_softc *, int));
> static __inline u_char
> ESP_READ_REG(sc, reg)
> 	struct esp_softc *sc;
> 	int reg;
> {
> 	u_char v;
> 
> 	if (sc->sc_bustype == WHUZZLE) {
> 		/* do Whuzzle stuff */
> 	} else {
> 		/* so Zoobie stuff */
> 	}
> 
> 	return (v);
> }
> 
> That has the possiblity of unnecessarily increasing code size, and
> paying the cost of extra comparisons, as more ZapStation models
> are introduced.  Also, it may be necessary to call bus-specific
> functions inside of this macro or inline, which inappropriately
> exposes that bus to the outside world.
> 
> This just makes me feel better about setting up a set of function
> pointers once at attach time.

Although the machine-independent driver would support a number of different
busses, this overhead could be avoided if I know that I have the chip at
a specific bus, by compiling with specific bus support, e.g.:

gcc -DESP_SUPPORT_WHUZZLE=1 -DESP_SUPPORT_ZOOBIE=0 ..... 

and having the source look like:

static __inline u_char
ESP_READ_REG(sc, reg)
	struct esp_softc *sc;
	int reg;
{
	u_char v;

#if ESP_SUPPORT_WHUZZLE + ESP_SUPPORT_ZOOBIE > 1
	switch (sc->sc_bustype) {
#if ESP_SUPPORT_WHUZZLE > 0
	case WHUZZLE:	
#endif
#if ESP_SUPPORT_WHUZZLE > 0
		/* do Whuzzle stuff */
#endif
#if ESP_SUPPORT_WHUZZLE + ESP_SUPPORT_ZOOBIE > 1
		break;
#if ESP_SUPPORT_ZOOBIE > 0
	case ZOOBIE:
#endif
#endif
#if ESP_SUPPORT_ZOOBIE > 0
		/* so Zoobie stuff */
#endif
#if ESP_SUPPORT_WHUZZLE + ESP_SUPPORT_ZOOBIE > 1
		break;
	}
#endif

	return (v);
}

Looks a bit ugly, but avoids increasing code size and paying the cost of
extra comparisons for customized kernels.

Joachim