Subject: Re: Multiprocessor with NetBSD ?
To: Jason R Thorpe <firstname.lastname@example.org>
From: Greywolf <email@example.com>
Date: 06/06/2001 10:35:49
# > > This notwithstanding, if we *don't* attempt SMP, I think we're going
# > > to get completely buried. I'm intent on picking up a dual-proc or
# > > quad-proc box at some point, and I'd like to be able to use BSD --
# > > NetBSD, specifically -- once I do.
To which Mike Stone replied:
# > *grin*
Whereupon Jason Thorpe added:
# Grin, indeed... we *do* SMP -- as has been hashed out here, it is symmetric,
# there are only a few very special cases where the "boot" processor is treated
# differently from the "other" processors, and for the most part, these involve
# arbitrary assignments of work (e.g. we make the "boot" processor the one that
# makes sure the time-of-day is up-to-date, even though all processors handle
# the clock interrupt), where as some deal with the way the hardware likes to
# work ("boot" processor spins up the others, and also has to be treated
# specially in the halt procedure).
Of course, you realise that all I was trying to say was that if we were
to fall back and punt out of SMP, that's not a favourable move.
Questions (please observe from a less-clued individual's POV):
+ Is cpu0 always going to be the boot processor? If that's the case,
how is it I've seen a Sun start on one processor and halt on another one?
+ When you say that all processors handle the clock interrupt, do you mean
that they all trigger the clock interrupt at some point, or that they
read it and update internal things accordingly? (I assume the latter,
based on the premise that only the boot processor keeps the TODC updated.)
NetBSD is PAR -- Powerful, Advanced, Reliable. Is your OS up to PAR?