Manuel Bouyer wrote:
> 
> On Mon, Jan 14, 2002 at 09:03:14AM -0600, Dan Debertin wrote:
> > David Laight writes:
> >  >
> >  > For a moment I thought this was a discussion about sBus expansion
> >  > kits (card, cable and chassis with another half dozen sBus slots).
> >  > If you find one of those apply a large hammer to it before anyone
> >  > tries to use it.
> >
> > Oh damn. And here I thought that lack of sparc SMP support was the
> > only thing standing between me and running NetBSD on my main
> > machine. Now it appears that my xbox is not only unequivocally _not_
> > an engineering marvel, but also probably non-working in terms of
> > driver support.
> 
> No, I think it's supported by NetBSD. From a sparc GENERIC:
> ## SBus expander box
> xbox*  at sbus? slot ? offset ?
> sbus*  at xbox?
> 
> >
> > Oh well. Could someone at least remove the driver from the tree so
> > people don't get their hopes up, or put a note in GENERIC to the
> > effect that it's buggy?
> >
> >  > They don't (can't) work reliably.
> >
> > Working fine under SunOS 5.5.1...
> 
> So it should work fine under NetBSD too. If you ever find it doesn't work,
> please report with send-pr.

The difficulty with these sBus expansion boxes is that they don't work
RELIABLY, they may appear to work but under heavy load (and maybe
environmental stress) they fail to correctly forward bus cycles to the
slave bus.  They are playing with the tolerance margins in order to work
at all.

Driver support is a different issue.  Of course some cards (eg the FEPS
based 100M ethernet card) can't be used because the relevant address
space isn't available (unless there is some was of tweaking the address
mappings that the guys writing our driver were unaware of).

	David