| [complete rewrite] Yes, that's probably the right thing to do.

  We thought so. BTW, one of the card variants uses a CD1400,
as the Cyclades do -- wouldn't make it sense to abstract the
"access register" function, pass it in as function pointer (or
whatever), have a higher level driver that did CD1400 handling
and a lower level driver that translates stuff into however the
board designed laid out their register map?

| And, write 3 different ones (for the 3 very different cards :-)

  Well, I only have one model of those cards, so writing drivers
for the others would be pure speculation.

	mjl