From owner-freebsd-smp Wed Sep 25 7:55:27 2002 Delivered-To: freebsd-smp@freebsd.org Received: from mx1.FreeBSD.org (mx1.FreeBSD.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 05DB137B401; Wed, 25 Sep 2002 07:55:26 -0700 (PDT) Received: from smtp2.enst.fr (matrix2.enst.fr [137.194.2.14]) by mx1.FreeBSD.org (Postfix) with ESMTP id 49CD143E86; Wed, 25 Sep 2002 07:55:25 -0700 (PDT) (envelope-from beyssac@bofh.enst.fr) Received: from bofh.enst.fr (bofh.enst.fr [137.194.32.191]) (using TLSv1 with cipher EDH-RSA-DES-CBC3-SHA (168/168 bits)) (Client CN "bofh.enst.fr", Issuer CN "ENST CA" (verified OK)) by smtp2.enst.fr (Postfix) with ESMTP id 466361F00E; Wed, 25 Sep 2002 16:55:23 +0200 (MEST) Received: from bofh.enst.fr (localhost [127.0.0.1]) by bofh.enst.fr (8.12.5/8.12.5) with ESMTP id g8PEtNgA058588 (version=TLSv1/SSLv3 cipher=EDH-RSA-DES-CBC3-SHA bits=168 verify=NO); Wed, 25 Sep 2002 16:55:23 +0200 (CEST) (envelope-from beyssac@bofh.enst.fr) Received: (from beyssac@localhost) by bofh.enst.fr (8.12.5/8.12.5/Submit) id g8PEtNx3058587; Wed, 25 Sep 2002 16:55:23 +0200 (CEST) Date: Wed, 25 Sep 2002 16:55:23 +0200 From: Pierre Beyssac To: Terry Lambert Cc: John Baldwin , smp@FreeBSD.ORG, beemern Subject: Re: PATCH: start_ap(), and P4 SMP hack to try Message-ID: <20020925165523.A58257@bofh.enst.fr> References: <3D8F48FB.F3DBE738@mindspring.com> <3D8F5AB2.8857A73B@mindspring.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.2.5i In-Reply-To: <3D8F5AB2.8857A73B@mindspring.com>; from tlambert2@mindspring.com on Mon, Sep 23, 2002 at 11:17:22AM -0700 Sender: owner-freebsd-smp@FreeBSD.ORG Precedence: bulk List-ID: List-Archive: (Web Archive) List-Help: (List Instructions) List-Subscribe: List-Unsubscribe: X-Loop: FreeBSD.org On Mon, Sep 23, 2002 at 11:17:22AM -0700, Terry Lambert wrote: > What it does is correct the start_ap routine to take a physical > instead of a logical CPU as an argument. FWIW, I've tested a similar patch of mine a few days ago, because I'm having the exact same problem as everyone else in this thread on a bi-Xeon Intel motherboard. My patch just changes the start_ap routine to use a physical ID, and start_ap is called in a loop with CPU varying from 8 back to 1 until we get one processor started. It doesn't work. I even added additional delays in start_ap in case this would just be a timing problem, and it didn't work either. I'm not a SMP specialist (actually I was not subscribed to -smp until I found it searching "PHY FreeBSD panic y/n" through Google) so I have no idea what to try next. -- Pierre Beyssac pb@enst.fr To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-smp" in the body of the message