[gawd I hate exchange.]

Apparently this was a funded project.  If it was funded, surely
it should have been completed by now, and funded or not, I think
the NetBSD community deserves to know what the progress is.  I
mean, not to be rude, but the official word that this was a funded
project came to us over a fortmoon ago:

[9808311608520 == 31 Aug 1998 16:08:52.0, near as I can tell.]

What good reason would the organization have not to make it
public?  This hush-hush stuff doesn't wash after nigh fifteen
months of zero results.  The only evidence I have seen of anything
remotely resembling threads in our kernel has been the softupdate
code, and as was indicated to me, they're not preemptive.  For
all intents and purposes, the work done on threading/SMP over the
last year or so has amounted to zero -- no experimental trees
for people to bash on, no mention of where we are with it,
no mention of even what team is working on it.  Curiously,
by whom is this operation "fully funded", and can we have any
reassurances that significant progress has been made?  I tend
to think not, given that the kernel is in a constant state of
flux, unless the threading will drop in seamlessly with
any given kernel.

Now that I've warranted asbestos jammies, I would like to say
that I comprehend how difficult it must be to write a decent
thread implementation, let alone MP, let alone SMP.  I've heard
REALLY GOOD kernel programmers indicate that scalable processors
really aren't, when it comes down to MP, and I do not pretend
to be a kernel programmer.

Sorry for my rant -- please don't smack me too hard.  I cannot
possibly be the only impatient person on this issue.



> > "Perry E. Metzger" <perry@piermont.com> writes:
> > > There is already a fully funded project that is working on
> > > multithreading our kernel, so there isn't really much cause to port
> > > the FreeBSD code.
> 
>  Andrew Gillham <gillhaa@ghost.whirlpool.com> wrote:
> 
>  > If there really is a fully funded project for SMP, why is it not
mentioned
>  > anywhere?  Personally I would love to have SMP for NetBSD, I will go
>  > out and buy an SMP motherboard as soon as alpha test code is available.
> 
> Because, perhaps, the organization does not wish to make it public yet?
> 
> Jason R. Thorpe                                       thorpej@nas.nasa.gov
> 
> Message-ID: <Pine.GSO.4.01.9808311608520.5190-100000@alpine.unicast.com>
> Well, I don't know much about the formal plans for SMP under NetBSD, but
> at Jason Thorpe's presentation on NetBSD at USENIX this year, he
> mentioned the following points:
> 
> 	* Work is fully under way to get SMP going under NetBSD.
>           presumably this is the `Well funded project' mentioned
>           above.
> 
>         * The current goal is to get a multi-threaded kernel going
>           first, and then allow scheduling of kernel and user
>           space threads across multiple processors.  Specific
>           mention was made of the fact that the FreeBSD project
>           took approximately the opposite approach, and is running
>           into trouble trying to integrate a multi-threaded kernel
>           into thir SMP work.
> 
> 
>       * The NetBSD-2.0 version mark has been reserved for the first
>           formal release with a multi-threaded kernel and SMP.
>                                 Jim Wise
>                                 jwise@unicast.com
-----Original Message-----
From: David Maxwell [mailto:david@fundy.ca]
Sent: Tuesday, November 23, 1999 11:11 AM
To: Charles M. Hannum
Cc: James Graham; 'Frank van der Linden'; Wenchi Liao;
port-sparc@netbsd.org
Subject: Re: unable to boot with 256MB of RAM


On Tue, Nov 23, 1999 at 01:44:59AM -0500, Charles M. Hannum wrote:
> David Maxwell <david@fundy.ca> writes:
> > Every comment I've seen on the subject said 2.0 == SMP
> 
> Really?  Every comment I've seen on the subject from a reputable
> source indicated that there was no plan to change the version
> numbering -- which would mean the next release after 1.5 would be 1.6.

As Ken said, I haven't heard any comments specifically in a while.
Here are some of the comments I've seen that helped me think that way...

(Apologies in advance for misattribution, missed messages, people
who didn't want others reminded of these comments....)

Perhaps all of this is fallout from Jason Thorpe's comments at USENIX.
No one has said anything contradictory until now however.

Every 'reputable source' must have commented on a list I'm not subscribed
to,
or don't have access to.

						David


> 
> > I heard once rumors of a funded SMP project, did anything come of that?
> > Was I misled?
> 
> The project is still running. You may have noticed a bunch of
multi-processor
> related commits to the kernel sources lately, which *may* be related.
> 
> Manuel Bouyer, LIP6, Universite Paris VI.           Manuel.Bouyer@lip6.fr
> 
> 
> From: "John F. Woods" <jfw@jfwhome.funhouse.com>
> I seem to vaguely recall suggestions that SMP would be a good candidate
for
> a major enough change to warrant calling it 2.0.
> 
> Of course, I'm not a member of core, and don't even play one on TV.  I
also
> don't have every current-users message for the past 8 years memorized, so
> I could be way, WAY off.
> 
> 
> From: Greg Hudson <ghudson@mit.edu>
> Nobody knows what will trigger the fabled major version bump, because
> nobody's made that decision.  There's some speculation that a
> preemptible kernel with SMP support is the critical change.
> 
> 
> From: Bill Studenmund <skippy@macro.Stanford.EDU>
> I believe that SMP will be the big new feature of 2.X.
> 
> From: Colin Wood <cwood@ichips.intel.com>
> not really.  there has been some mention that kernel threads/smp support
> will be the needed catalyst for the 2.0 version bump.
> 
> From: Calvin Vette <cvette@borders.com>
> This is actually a question for tech-kern. The timetable for threads/SMP
> isn't currently public.
> 
> From: "Perry E. Metzger" <perry@piermont.com>
> > 1.  Just curious, what's going on with the SMP stuff?
> I honestly don't know. Its in the hands of people who don't talk much
> about what they are doing.
> 
> From: christos@zoulas.com (Christos Zoulas)
> Date: Wed, 26 May 1999 15:05:49 -0400
> Since there have been a lot of questions lately about SMP and there
> is currently some kernel hacking activity related to SMP, I thought
> that it was appropriate to create a forum for SMP related discussions.
> 
> From: Stefan Grefen <grefen@hprc.tandem.com>
> Subject: port-i386/6928: i386 SMP
>        Bootstrap suuport for SMP system based on intel MP spec