> } I beg to differ. fdthree, i.e. fdc@ebus, behaves just like a regular isa
> } fdc, and does real DMA, unlike fdc@sbus. The sparc fdc driver doesn't
>
>     In this case, I have to ask why the OpenBSD driver has the ebus
> attachment code in the same file as the sbus stuff, since the ebus
> driver is going to need a different back end?

A significant part of the code can be shared between sbus-style and ebus-style
drivers (ioctl interface, strategy routine, state machine, etc).

Miod