> That works when you know the intended _bus_ byte order, and bus
> byteorder is always the same as the driver hardware byte-order. 
> 
> I don't see how it would work, in general, in chipset-level drivers,
> for chips that appear on busses with different endian-ness.

Uh, these chips have to deal with differing access methods, etc.,
etc., etc., already.  Certainly, the #ifdef as-is wouldn't work, but
solving the problem _is_ pretty much trivial.



cgd