Subject: Re: The VM System
To: John Dyson <>
From: J.T. Conklin <>
List: tech-kern
Date: 11/23/1995 00:21:31
> > I've heard several times that the core team doesn't "like" John
> > Dyson's work for FreeBSD, without specific information. While
> > FreeBSD

> :-(.

:-( from me too.

It goes against the NetBSD mission for us to automatically dismiss
(or automatically accept) an idea and/or implementation based on its

I may not have been part of core when the above statement was made (or
perhaps I was, however I'm still very much a userlevel/standards dood
and sometimes I skim through kernel architecture discussions due to
lack of time).  However, I think it likely that there were technical
reasons for dismissing the FreeBSD VM but they may not have been 
communicated to the world.  

If we believe that improving the VM subsystem is a goal, then it is 
important for us to *understand* the problems, and to *understand* 
our options.  It's already been mentioned that this is a *hard* 
problem.  That is why that I'd like some study and discussion
has to take place before a decision is made.

It may turn out that the FreeBSD VM system has evolved since it was
last looked at and the prior objections are moot; there might be some
lingering problems that can be resolved easily by co-operating with
our FreeBSD counterparts; or it might be decided that although the
FreeBSD VM is not exactly what we want, it's a step forward from what
we've got so we'll adapt it while we make longer term plans.  Those
scenarios are all speculation, I say again that it is impossible for
us to know what course we'll take until some basic research is

And the above process of researching problems, possible approaches and
implementation doesn't just stop with VM.  I believe that we must
exercise the same care and diligence for other extensive changes.

This might slow down development of "new features" somewhat, but in my
opinion this will result in a vastly better system.  Plus we'll be
getting some good material for journal articles (good for publicity)
and internals documentation (necessary for NetBSD to be maintained or
ported by people not intimate with the code).

In summary, I think when the VM survey is begun that the FreeBSD VM
system needs to be considered.  And if we decide to experiment with
or adopt it, I hope that John is willing to offer us assistance and 


J.T. Conklin