Let me say just one thing, or two....

I hope that bounce buffers will one day (soon) be added
to NetBSD.  I am a i386 port user and will selfishly say
that a arch/i386 fix would be great.

However, I would welcome a MI version also, so all ISA
sporting machines can reap the befits of having old
hardware yet using more memory.

Now I will make an observation.  Everytime this issue has
come up, the "core" has got very defensive and yelled MI.  
Then told us much less important bounce buffers are than
current projects, then the issue gets pushed back more.

Now may I ask, when a project or a fix get finished, why
doesn't bounce buffers get worked on instead of just pushed
off?  It would soothe the souls of many people.

I've been using BSD since the first i386BSD 0.1 and the two
issues that have been flamed the most are lack of hardware
support and bounce buffers.

Now I realize that the best way to ensure a feature not be
added is to gripe about it, bruise some egos.  NetBSD is a great
OS, stable, fast and clean, given.  I would hope the core would
take a moment, and for once, and consider adding a popular
feature request.

(sarcasm follows, please take it lightly.)
(Now place yourself in pre-Revolutionary France)
It seems, historically, that when the masses raised there voices
to request a popular feature, the core would look down upon
the masses and reply with a "NO, how dare you make demands of
us, the core.  You must suffer until we see fit to add your
request.(eat cake, etc.)"  Then there would be a punishment
consisting of a delay.
(now come back to reality)

In all fairness, there have been moments when the core has
thumbed down a feature request and one core member has stepped
forward and been kind enough to add the feature.

Anyway...

Make it MI, make it arch/i386, arch/alpha, etc. specific who cares!
The entire point of BSD is a research platform to help explore the
mysteries of an OS.  

Well here is a mystery, how to implement bounce buffers in a
MI manner?

If we are using playground rules:
  "Betcha can't do it."
or 
  "I double dog dare you to try to implement bounce buffers."

Well guys, I guess we'll wait another 5 years for
bounce buffers, I'm sure I pissed someone off...

jam