From owner-freebsd-smp Wed Dec 4 17:30:53 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 42EB637B401; Wed, 4 Dec 2002 17:30:52 -0800 (PST) Received: from pintail.mail.pas.earthlink.net (pintail.mail.pas.earthlink.net [207.217.120.122]) by mx1.FreeBSD.org (Postfix) with ESMTP id D72FE43E4A; Wed, 4 Dec 2002 17:30:51 -0800 (PST) (envelope-from tlambert2@mindspring.com) Received: from pool0353.cvx40-bradley.dialup.earthlink.net ([216.244.43.98] helo=mindspring.com) by pintail.mail.pas.earthlink.net with esmtp (Exim 3.33 #1) id 18Jkqp-0004M7-00; Wed, 04 Dec 2002 17:30:51 -0800 Message-ID: <3DEEABFD.A40A735E@mindspring.com> Date: Wed, 04 Dec 2002 17:29:33 -0800 From: Terry Lambert X-Mailer: Mozilla 4.79 [en] (Win98; U) X-Accept-Language: en MIME-Version: 1.0 To: John Baldwin Cc: freebsd-smp@freebsd.org, Michael Richards Subject: Re: Intel SE7500CW2 narrowed down... References: Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 7bit 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 John Baldwin wrote: > Terry. We are having problems sending _interprocess_interrupts_. > IPI's do _not_ go through the IO APIC. They are sent from one > local APIC over to another local APIC. The problems we are seeing > have nothing to do with I/O interrupts. The box runs fine with one > processor and using the I/O APIC if you ignore the failure of the > AP to respond to the startup IPI. This is the "ServerwWorks related lockup", not the "Second processor does not start, and so does not grab Giant, so system does not lock up" thing, right? The easy way to prove your thesis is to send a broadcast IPI. You only have 2 CPUs in that box, anyway. If it starts working then you are right. I think the serverworks chipset can only run in 2 of the 4 modes, at least as programmed by the BIOS, and goes off in the weeds, otherwise (the first thing an AP does is grab giant, which means it tries to grab the interrupts, as a side effect). It may be that the chipset is plain busted. Note that one of the OS's that works doesn't run in APIC I/O mode at all, and the other OS that works runs in virtual wire mode. Other than tonight, I won't be able to engage in a deep discussion on this for about a week, I think, so get your shots in now. 8-). -- Terry To Unsubscribe: send mail to majordomo@FreeBSD.org with "unsubscribe freebsd-smp" in the body of the message