From owner-freebsd-smp  Sun Jan 19 01:29:01 1997
Return-Path: <owner-smp>
Received: (from root@localhost)
          by freefall.freebsd.org (8.8.4/8.8.4) id BAA24481
          for smp-outgoing; Sun, 19 Jan 1997 01:29:01 -0800 (PST)
Received: from clem.systemsix.com (clem.systemsix.com [198.99.86.131])
          by freefall.freebsd.org (8.8.4/8.8.4) with SMTP id BAA24476
          for <smp@freebsd.org>; Sun, 19 Jan 1997 01:28:57 -0800 (PST)
Received: from localhost (localhost [127.0.0.1]) by clem.systemsix.com (8.6.12/8.6.12) with SMTP id CAA02329; Sun, 19 Jan 1997 02:28:37 -0700
Message-Id: <199701190928.CAA02329@clem.systemsix.com>
X-Authentication-Warning: clem.systemsix.com: Host localhost didn't use HELO protocol
X-Mailer: exmh version 1.6.5 12/11/95
From: Steve Passe <smp@csn.net>
To: Kenneth Merry <ken@housing1.stucen.gatech.edu>
cc: smp@freebsd.org
Subject: Re: success!! 
In-reply-to: Your message of "Sun, 19 Jan 1997 04:04:10 EST."
             <199701190904.EAA20041@housing1.stucen.gatech.edu> 
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Date: Sun, 19 Jan 1997 02:28:37 -0700
Sender: owner-smp@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

Hi,

>	Well, I'm writing this from my machine, running X, with all of the
>cards installed an autoconfigured by the BIOS, with an APIC_IO kernel,
>using version 1.4 of the MP spec.

this is great news!!!

---
> Also, I ran mptable tonight, with all the
>cards installed, and noticed that NINTR=15 had changed to NINTR=17.
> ...
> I installed an additional ethernet card last
>night, and it wasn't working properly with the SMP kernels.  That was most
>likely because of the number of interrutps.

if there are too few INTR slots compiled into the kernel it SHOULD panic
during the boot:

i386/i386/mp_machdep.c:

intEntry( IntEntry entry, int* intr )
{
    int	x;

    if ( (x = (*intr)++) == NINTR )
	panic( "too many INTs, increase 'NINTR'\n" );
 ...
}

so when you "hung", I wonder if it was just a panic() call that didn't print
properly?  I thought I was seeing this sort of behaviour in the past, and
coded all my panic calls to look like:

 printf( "message\n" );
 panic( "\n" );

a few cleanups ago I removed all that stuff because it looked so ugly and
unnecessary.  Why don't you try decrementing NINTR by 1 or 2 and rerunnning
config/make and see if it panic()s or if it "hangs".

I may have gotten "too clever" by counting INT sources with mptable.  If
NINTR is undefined the kernel source defaults to 24, which works fine with
all IO APICs, just wastes a little space.

--
Steve Passe	| powered by
smp@csn.net	|            FreeBSD

-----BEGIN PGP PUBLIC KEY BLOCK-----
Version: 2.6.2

mQCNAzHe7tEAAAEEAM274wAEEdP+grIrV6UtBt54FB5ufifFRA5ujzflrvlF8aoE
04it5BsUPFi3jJLfvOQeydbegexspPXL6kUejYt2OeptHuroIVW5+y2M2naTwqtX
WVGeBP6s2q/fPPAS+g+sNZCpVBTbuinKa/C4Q6HJ++M9AyzIq5EuvO0a8Rr9AAUR
tBlTdGV2ZSBQYXNzZSA8c21wQGNzbi5uZXQ+
=ds99
-----END PGP PUBLIC KEY BLOCK-----