From owner-freebsd-current@FreeBSD.ORG Tue May 16 05:09:37 2006 Return-Path: X-Original-To: current@FreeBSD.org Delivered-To: freebsd-current@FreeBSD.ORG Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 142CC16A504; Tue, 16 May 2006 05:09:37 +0000 (UTC) (envelope-from sobomax@FreeBSD.org) Received: from sippysoft.com (gk.360sip.com [72.236.70.226]) by mx1.FreeBSD.org (Postfix) with ESMTP id 37DA043D58; Tue, 16 May 2006 05:09:11 +0000 (GMT) (envelope-from sobomax@FreeBSD.org) Received: from [192.168.0.38] ([204.244.149.125]) (authenticated bits=0) by sippysoft.com (8.13.6/8.13.6) with ESMTP id k4G58sOk030209 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO); Mon, 15 May 2006 22:08:59 -0700 (PDT) (envelope-from sobomax@FreeBSD.org) Message-ID: <44695E63.6030507@FreeBSD.org> Date: Mon, 15 May 2006 22:08:51 -0700 From: Maxim Sobolev Organization: Sippy Software, Inc. User-Agent: Thunderbird 1.5.0.2 (Windows/20060308) MIME-Version: 1.0 To: Nate Lawson References: <4467B75F.2030406@FreeBSD.org> <446945FA.7020103@root.org> In-Reply-To: <446945FA.7020103@root.org> Content-Type: text/plain; charset=KOI8-U; format=flowed Content-Transfer-Encoding: 7bit Cc: "current@freebsd.org" , John Baldwin Subject: Re: SMP on MacBook Pro, some new info X-BeenThere: freebsd-current@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Discussions about the use of FreeBSD-current List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 16 May 2006 05:09:40 -0000 Nate Lawson wrote: > Maxim Sobolev wrote: >> Hi, >> >> I have discovered that it's possible with a very good repeatability >> (90% or more) to start the second core by pressing power button >> promptly during the 5 second period when BSP have sent startup IPIs >> already and is waiting for the AP going online. I am not sure how to >> interpret this info - perhaps there is some problem with caching BSPs >> LAPIC writes, and interrupt generated by the power button causes the >> cache to be flushed. >> >> Any ideas? > > My completely unresearched opinion is that the IPI is not reaching the > AP or the response is getting lost. I don't know why. Looks like IPI cached somewhere and simply doesn't reach LAPIC or AP since if I enable checkpoints and don't press power key, I see message about failure to start the AP and a trace consisting of 99 99 99 99 99 99. At the same time, if I "help" SMP startup by pressing the power key in time, I see normal trace 1 2 3 4 5 6, which suggests that checkpoints work as expected and in the former case the AP simply doesn't get the message. Is it possible that one of the cores is put by BIOS into some deep sleep state. BTW, Intel has introduced some new "deeper sleep state" (aka "deep C4") in Core Duo and for that reason it doesn't see IPI? Maybe activity generated by power button press wakes it up. -Maxim