From owner-freebsd-smp Thu Feb 7 15:56:25 2002 Delivered-To: freebsd-smp@freebsd.org Received: from blue.gradwell.net (blue.gradwell.net [195.149.39.10]) by hub.freebsd.org (Postfix) with SMTP id A733137B417 for ; Thu, 7 Feb 2002 15:56:15 -0800 (PST) Received: (qmail 2619 invoked from network); 7 Feb 2002 23:56:08 -0000 Received: from stonkin.dcs.aber.ac.uk (HELO vaio.gradwell.com) (193.60.15.4) by pop3.gradwell.net with SMTP; 7 Feb 2002 23:56:08 -0000 Message-Id: <5.1.0.14.0.20020207231134.05284c60@indigo.gradwell.net> X-Sender: postmaster%pop3.peterg.org.uk@indigo.gradwell.net X-Mailer: QUALCOMM Windows Eudora Version 5.1 Date: Thu, 07 Feb 2002 23:16:36 +0000 To: freebsd-smp@FreeBSD.ORG From: Peter Gradwell Subject: Re: Troubles with MP on ServerWorks NB6635 platform In-Reply-To: <3C63085B.C585BF04@mindspring.com> References: Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii"; format=flowed 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 At 15:06 07/02/2002 -0800, Terry Lambert wrote: > > On Mon, 4 Feb 2002, Igor Lepechin wrote: > > > > > Feb 4 02:44:49 web /kernel: APIC_IO: Testing 8254 interrupt delivery > > > Feb 4 02:44:49 web /kernel: APIC_IO: Broken MP table detected: 8254 > is not connected to IOAPIC #0 intpin 2 > > > Feb 4 02:44:49 web /kernel: APIC_IO: routing 8254 via 8259 and > IOAPIC #0 intpin 0 [...] >There are two possible breakages: > >1) The MP Table does not accurately reflect how the > hardware works. > > That's its job; if it doesn't do its job, then it > might as well be replaced with one that daoes (i.e. > reflash your MB with a working BIOS, please) > >2) The 8254 is not wired into the IO APIC correctly. > > This is more serious. It's either a motherboard > layout flaw, or it's a chipset flaw, and it's > going to take an incredible synchronization > penalty to work around the problem, to the point > that you are much better off with a different > motherboard, since the benefit you gain from the > additional processor is being spent on workarounds. I have looked at the BIOS updates/fixes for Intel's SCB2 mobo, which (a) does this and (b) does not reboot in FreeBSD 4.5-STABLE except after a cold reboot. http://support.intel.com/support/motherboards/server/scb2/latest_bios.htm However, none of them seem to suggest that they fix anything like this. How do I best report it to the manufacturer? Does the FreeBSD-SMP team have a contact inside intel? thanks peter -- peter gradwell. gradwell dot com Ltd. http://www.gradwell.com/ engineering & hosting services for email, usenet, sms and web. To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-smp" in the body of the message