> Additionally, I have done the following (after reset):
> 
> ok " /sbus" select-dev
> ok 40 2 100 map-in .s
> ffecf040
> ok 2 dump
> ......headerline......
> ffecf040  Data Access Error
> 
> The same happens with a dump of length 1 or length 4.
> 
> If I poke into the register, a Level 15 interrupt is
> generated. The 0x100-byte range has "holes", but the
> interrupt seems to be of which (valid) register I use.

A Level 15 interrupt is a data access error, or async fault.  

If you map in the area specified by the `reg' properties of the OBP
properly and still get async faults, either you are mapping cacheable
areas as un-cacheable (or visa-versa) or you are mis-interpreting the
register map.  Try toggling the cacheable attributes.

=========================================================================
Eduardo Horvath				eeh@netbsd.org
	"I need to find a pithy new quote." -- me