From owner-freebsd-smp@FreeBSD.ORG Thu Dec 20 12:15:16 2007 Return-Path: Delivered-To: freebsd-smp@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id B939016A417 for ; Thu, 20 Dec 2007 12:15:16 +0000 (UTC) (envelope-from bug@camisano.net) Received: from mail.skytek.it (mail.skytek.it [217.194.176.18]) by mx1.freebsd.org (Postfix) with ESMTP id 4E1F413C457 for ; Thu, 20 Dec 2007 12:15:16 +0000 (UTC) (envelope-from bug@camisano.net) Received: from [192.168.30.100] ([192.168.30.100]) by mail.skytek.it (Skytek Mail Server v.9.47) with ASMTP id ZYV87848 for ; Thu, 20 Dec 2007 13:01:48 +0100 Message-ID: <476A59B0.2020902@camisano.net> Date: Thu, 20 Dec 2007 13:01:52 +0100 From: Daniel Ponticello User-Agent: Thunderbird 2.0.0.9 (Windows/20071031) MIME-Version: 1.0 To: freebsd-smp@freebsd.org Content-Type: text/plain; charset=ISO-8859-15; format=flowed Content-Transfer-Encoding: 7bit Subject: SMP kernel test on Freebsd7.0-BETA4 inside VMWARE: Good results! X-BeenThere: freebsd-smp@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: FreeBSD SMP implementation group List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 20 Dec 2007 12:15:16 -0000 Hello Everyone, I'm writing this to the list to let you know about some tests on SMP kernel of 7.0: i'm testing both FreeBSD 7.0-BETA4 and FreeBSD6.3 on VMWARE ESX Server 3.5, using 2 virtual cpu (SMP Kernel). On freebsd 6.2 and FreeBSD6.3 in ESX Server, SMP kernel is very unstable and many processes (like compiling something from source or installing from the ports) will end with segmentation fault. Also, the VM won't survive a buildworld (the process just freezes). I've noticed that the problem is no longer present in FreeBSD7.0-BETA4: the VM i'm testing is up since 2 days and survived several buildworlds and installation from the ports (SMP kernel with 2 processors, APIC enabled). Also the clock derive problem is no longer present: while in freebsd6.2 the clock was loosing 5 seconds per day, on freebsd 7 it is loosing only 0.9 seconds per day. Looks promising. I guess this is due to several rewrites of SMP in freebsd7 kernel. Anyone can confirm these results? Thanks, Daniel Ponticello