> A recent change to calls to map[io]dev that ommitted the use
> of a bustype seem to leave one with a few nodes that have a
> value of zero for the rr_iospace tag portion of register
> addresses.

The intent of the change was to always use the <iospace,paddr,len>
tuple to describe a device's location on some bus (as seen from the
processor) in stead of passing around `bustype' parameters. That
makes for fewer special cases based on system model (sun4/sun4c/sun4m).

I made the assumption that machines providing the "ROM properties"
(including SS2s) would always produce the correct iospace value.

-pk