Quoting Manuel Bouyer,

> wdc0    at isa? port 0x1f0 irq 14
> wdc1	at isa? port 0x170 irq ?
> wd* at wdc? drive ?
> atapibus* at wdc?  #one atapibus per controller
> acd* at atapibus? drive ?   #The atapi cdrom, could be other devices

> This will require some changes to the wd/wdc driver, and i plan to 
> split wd.c into wdc.c (controller driver) and wd.c (ide drives drivers).

This is a good step to take, I think. The split wd.c -> wd.c + wdc.c is
also good for consistency, and makes for easier integration of stuff like
the Triton chipset, which would give us goodies like busmaster DMA.

I have thought of doing it more generally: making it "atabus" in stead
of "atapibus" and taking into account ATA-{1,2,3} optimally as well
as ATAPI. It would solve the possible confusion over a 'bus' being attached
to a controller was well as devices, but perhaps this is more trouble
then it's worth? Making a better use of the ATA-* specs would be a goal
worth pursueing in any case, I think (although I'm still happy the wd.c
we now have is stable; remember all the troubles we had before Charles
had a look and fixed it..)

- Frank