strace broken in 7.0?

Dag-Erling Smørgrav des at
Fri Jan 11 06:49:50 PST 2008

Dag-Erling Smørgrav <des at> writes:
> I will give you the benefit of the doubt and assume that you did not
> realize how insulting this is to someone who has spent the last ten
> years of his life writing the operating system that you use for free.

Having gotten that off my chest and calmed down a little, I will try to
address your other points.

Timo Schoeler <timo.schoeler at> writes:
> Dag-Erling Smørgrav <des at> writes:
> > Plurium Interrogationum.  Your question presupposes that FreeBSD is or
> > will soon be unbuildable and / or that its quality decreases over
> > time.
> The quality decrease could be felt, and was felt by too many people.

No factual basis.

> > That is far from the case.  In fact, we have very strong QA
> > mechanisms in place to ensure that it does not happen.
> Better ones than those that were there years ago are *now* in place or
> being implemented.

So you admit that we have better QA mechanisms than we used to (though I
suspect you don't realize how long they've been in place).  Doesn't that
contradict your unsupported claim that quality has decreased?

> > If you believe otherwise, you are misinformed, deluded or malicious.
> I don't 'believe'.

A favorite statement of fundamentalists.  You cannot claim to own the
objective truth; neither can I, but having spent ten years working on
FreeBSD, I can rightfully claim to know what goes on in the project far
better than you do.

> > That is not how collaborative open source development works.
> On Linux it works like this, everyone does what (s)he wants, and a few
> gurus (Torvalds etc) have hard times to concentrate on the important
> things.

This is not how Linux development works.  I will not pursue it further
as it isn't relevant to this discussion.

> The BSDs have/had a different approach, and this was well known
> throughout the community and is still said in case somebody asks.

Mere conjecture on your part.  You do not have first-hand knowledge of
how the project works.

> You didn't get my point.
> [...]
> You didn't get my point, again.

I'm starting to wonder if you actually have a point, since you
conveniently left it out of your email.

> I did, and I still help to work on a BSD licensed exokernel that will be
> published when it's finished.

Good, I love vaporware.

> > No.  You don't get to use the L-word.  The license says so, in
> > capital letters.
> >
> > If what you really meant was "responsible", see above.  People will
> > only take responsibility for a piece of code if they have an
> > interest in it.
> You know the difference between ``I can't sue you for your dog pooing
> into my yard'' and the fact that you should be ashamed of your dog
> doing it and you allowing it?  Convert this, take it *(-1), then you
> get the picture.  I write code, commit it, whatever, then I should be
> 'liable' for it, take care it works, whatever.

OK, so you did mean "responsible" and not "liable", though you don't
seem to appreciate the difference in meaning.  Score one for the
language barrier.

I agree that people should feel a certain responsibility for the code
they write, but life doesn't always work like that.  Significant
portions of FreeBSD were written by people who are no longer around, for
a variety of reasons.  By your standards, all of that code should be
removed unless someone else was willing to take over.  If we did that
consistently, you would be even angrier than you already are (judging
from some people's reactions when we have been forced to do so in the

> It's a *moral* thing!

Morals are for people who don't have ethics.

> > > [...] I wouldn't be surprised about races on six core machines.
> > I would.  Very much.
> Welcome to reality.  Honestly.

Poor quoting on my part.  What I actually responded to was:

> > > I wouldn't be surprised about races on six core machines, that
> > > don't appear on quad or eight cores.

Once again, I would be very surprised if there turned out to be a race
condition in FreeBSD that only occurred on six-core machines.  I can
easily imagine that there might be bugs resulting from the incorrect
assumption that the number of cores is always a power of two, but I
doubt very much that they would manifest as races.  Race conditions tend
to stem from the "yoghurt problem", which is a zero-one-infinity type of
problem.  I won't bore you with the entire train of thought, but the
bottom line is that the number of cores doesn't matter beyond two.

By the way, as far as I know, FreeBSD already runs on machines with a
non-power-of-two number of cores:

 - The AMD Opteron architecture supports an arbitrary number of cores in
   a HyperTransport ring; there are triple-socket Opteron motherboards,
   and AMD is expected to launch a triple-core Phenom this quarter.

 - The Sun UltraSPARC T1 (which runs FreeBSD/sun4v) is available with
   four, six and eight physical cores, each with four concurrent
   execution threads, giving a total of sixteen, twenty-four or
   thirty-two logical cores.

Dag-Erling Smørgrav - des at

More information about the freebsd-chat mailing list