From owner-freebsd-current Thu Mar 5 18:32:19 1998 Return-Path: Received: (from majordom@localhost) by hub.freebsd.org (8.8.8/8.8.8) id SAA02007 for freebsd-current-outgoing; Thu, 5 Mar 1998 18:32:19 -0800 (PST) (envelope-from owner-freebsd-current@FreeBSD.ORG) Received: from danberlin.resnet.rochester.edu (danberlin.resnet.rochester.edu [128.151.91.212]) by hub.freebsd.org (8.8.8/8.8.8) with ESMTP id SAA02000 for ; Thu, 5 Mar 1998 18:32:17 -0800 (PST) (envelope-from root@danberlin.resnet.rochester.edu) Received: from localhost (root@localhost) by danberlin.resnet.rochester.edu (8.8.8/8.8.8) with SMTP id VAA13218; Thu, 5 Mar 1998 21:27:39 -0500 (EST) (envelope-from root@danberlin.resnet.rochester.edu) Date: Thu, 5 Mar 1998 21:27:39 -0500 (EST) From: Daniel Berlin To: Chris Timmons cc: freebsd-current@FreeBSD.ORG Subject: Re: SMP & pci bridged cards? In-Reply-To: Message-ID: MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Sender: owner-freebsd-current@FreeBSD.ORG Precedence: bulk X-Loop: FreeBSD.ORG I took a look at that. The 3940 is special because they cheat by including a pci bridge on the card to handle the extra channel, instead of using some weird method. So it's effectively an extra controller, AFAIK. IE the controllers don't know they are on the same card,etc. I'm doing this from memory of what a friend at adaptec told me once. the 2940 is single channel. I haven't had time to delve to deeply into the source as to why it happens, but it indeed only happens on 1.24 bioses (i only have access to 1.24,1.25,1.32 and 1.34). I would rather figure out if it's adaptec's fault before trying to fix something that ain't broke. --Dan On Thu, 5 Mar 1998, Chris Timmons wrote: > > Are you sure this isn't the known problem about pci<->pci bridges with smp > kernels? Do you see the problem with 3940's but not 2940's? See > http://www.freebsd.org/~fsmp/SMP/pcibridge.html. > > I saw similar behavior which was caused by just this phenomenon - > following the procedure and hacking mp_machdep.c by hand per the www page > got things going on my tyan s1662D with a 3940. > > Perhaps this problem is one that could receive some attention in Steve > Passe's abscence? Some hints on an architectural direction wrt the pci > code would be welcome. With an architectural template, some proto-SMP > hackers might emerge from the woodwork. > > -Chris To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-current" in the body of the message