Skip site navigation (1)Skip section navigation (2)
Date:      Sat, 17 Aug 1996 12:37:38 -0700
From:      erich@uruk.org
To:        Steve Passe <smp@csn.net>
Cc:        freebsd-smp@freebsd.org
Subject:   Re: SMP on Intel MG15 
Message-ID:  <199608171937.MAA08515@uruk.org>
In-Reply-To: Your message of "Sat, 17 Aug 1996 13:08:14 MDT." <199608171908.NAA20026@clem.systemsix.com> 

next in thread | previous in thread | raw e-mail | index | archive | help

Hi all!

As someone who used an Xtended Xpress box as a primary implementation
platform for generalizing the Linux-SMP code, I think comments are in
order...

FWIW, I've got these mostly fixed in a version I'm playing with, and if I
can get synched up with the changes coming across and some other fixes in,
then I'll post them to the list to be merged in.  I still need to discuss
some important issues with the folks involved with FreeBSD development...
I've just been sorta lazy with it since I've been busy with some other
things.


Steve Passe <smp@csn.net> writes:

> >In article <4v4ddu$n7q@al.imforei.apana.org.au> you wrote:
> >
> >: 2) The machine locks up shortly after the SMP kernel load:
> >
> >: FreeBSD/SMP: Multiprocessor motherboard
> >: Boot cpunumber=0x0
> >: [ machines freezes ]
> >
> >: I presume this means that the machine is reporting the number of
> >: CPUs in a non-recognised format. APIC supports appears to be
> >: enabled in the BIOS.
> 
> "Boot cpunumber=0x0" is correct, this says that it is booting off CPU #0,
> ie the 1st CPU.
> 
> here's my dmesg output, note that the first 3 lines only
> appear because I modified machdep.c to print them after
> /var/log/messages was active (default prints too early):
> 
> FreeBSD/SMP: Multiprocessor motherboard
> Boot cpunumber=0x0

I might know your problem.  Some of the nastiest bugs in Linux-SMP before
I fixed it, and that is definitely still in FreeBSD-SMP, are the following
presumptions:

 (1) boot CPU is APIC id #0  (this is true on Xtended Xpress boxes, but not
	on many Pentium Pro machines).
 (2) APICs are numbered consecutively.

#2 is FALSE on an Xtended Xpress box.  The one I tested with was numbered:

	0, 2, 3, 4.

NOTE that FreeBSD-SMP is currently hard-coded to expect the next CPU after
the boot CPU to be #1.

I'd say this is your problem.

> Number of CPUS found: 2

This is probably reporting what the MP Config Table said, not how many
it was able to boot (so this is irrelevant from the point of view of
APIC numbering).

> >I wrote:
> >> I have just received an Intel Xtended Xpress/MG15 with dual
> 
> I would hate to think Intel can't make an Intel SMP compliant
> motherboard.  Is this particular board known to have ever run
> an SMP OS?

I don't know about the "MG15" part, but Intel is almost rabid about
getting Windows NT to run on their SMP boxes.

If you look in the BIOS settings, you'll note that the "MPS version"
is either 1.1 or 1.4 (in the version I had, you could set it).

For purposes of FreeBSD or Linux, this version number doesn't currently
matter.

For Windows NT, it does, and you need to set it to 1.4.  The reason is
that Windows NT uses all kinds of information about the bus structure
heirarchy.

> Can you read the s-spec of the chips?  this is a 3
> digit number starting with SX,SK,S?.  I believe I earlier wrote:
> > I am using it with 2 P5 133mHz parts, type SU022
> which is wrong, I am using 2 SY022 chips  (according to my
> notes of a call with the vendor, the chips are buried too
> deep in the system to check right now)

I'd be suspicious of FreeBSD unless I was sure it identified them
correctly, as Intel is always adding new designations when new steppings
of a chip come out.


--
  Erich Stefan Boleyn                 \_ E-mail (preferred):  <erich@uruk.org>
Mad Genius wanna-be, CyberMuffin        \__      (finger me for other stats)
Web:  http://www.uruk.org/~erich/     Motto: "I'll live forever or die trying"



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199608171937.MAA08515>