On Fri, 9 Feb 1996, Per Fogelstrom wrote:

> > The hardware is broken, not the OS.
> 
> No, it's your idea about supporting hardware that is broken....

I absolutely agree with Per.  The idea that ISA bounce buffers
shouldn't be implemented because the ISA bus "is broken" is just plain
stupid.  Isn't that like saying that if a piece of hardware has
limitations and needs specific handling, that is a reason not to
support it?  That NetBSD will support perfect hardware only?  The ISA
bus has a very specific 16M physical address limitation.  That's a
known factor to be worked with.  What's so "broken" about that?

Heck (can I say "heck" here after Black Thursday?), this is like VAX
port-master Ragge saying that NetBSD/VAX will never support the
MicroVAX 2000 and VAXstation 2000 (probably _the_ most frequently
asked after VAX port) because of the way DEC decided to have disk and
tape I/O circuitry share a certain memory buffer, making disk and tape
drivers for the beasts a bit more difficult to implement.  Actually,
it's worse -- the MV/VS 2000 hardware _is_ broken, on purpose, whereas
the ISA bus just has a very simple, specific limitation that's easy to
work around.

-tih
-- 
Tom Ivar Helbekkmo
tih@Hamartun.Priv.NO