From owner-freebsd-current@FreeBSD.ORG Wed Oct 20 21:33:26 2004 Return-Path: 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 B8FCC16A4CE for ; Wed, 20 Oct 2004 21:33:26 +0000 (GMT) Received: from parati.mdbrasil.com.br (parati.mdbrasil.com.br [200.210.70.4]) by mx1.FreeBSD.org (Postfix) with SMTP id 6B42D43D1D for ; Wed, 20 Oct 2004 21:33:25 +0000 (GMT) (envelope-from eksffa@freebsdbrasil.com.br) Received: (qmail 11187 invoked by uid 1003); 20 Oct 2004 21:33:18 -0000 Received: from eksffa@freebsdbrasil.com.br by parati.mdbrasil.com.br by uid 82 with qmail-scanner-1.20 (uvscan: v4.1.60/v4288. Clear:RC:1(200.166.90.82):. Processed in 0.015473 secs); 20 Oct 2004 21:33:18 -0000 Received: from unknown (HELO freebsdbrasil.com.br) (200.166.90.82) by parati.mdbrasil.com.br with SMTP; 20 Oct 2004 21:33:18 -0000 Message-ID: <4176D9A3.9030500@freebsdbrasil.com.br> Date: Wed, 20 Oct 2004 18:33:23 -0300 From: Patrick Tracanelli Organization: FreeBSD Brasil LTDA User-Agent: Mozilla/5.0 (X11; U; FreeBSD i386; en-US; rv:1.5) Gecko/20031207 X-Accept-Language: en-us, en MIME-Version: 1.0 To: current@freebsd.org Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit cc: Valmir Consoni Subject: -RC1 more stable when compared to -STABLE X-BeenThere: freebsd-current@freebsd.org X-Mailman-Version: 2.1.1 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: Wed, 20 Oct 2004 21:33:27 -0000 Recently I was running a number of tests against 5.3-STABLE (btw, RELENG_5 labeling the system as 5.3-STABLE might be very confusing in a situation where there's not even a 5.3-RELEASE yet, if one doesnt follow how -STABLE are treated compared to "security branchs"), in a SMP system with multiple SATA RAID, and it was not responding well, had a number of freezes under stress tests; Some of those problems could be minimized running BSD as the scheduller instead of ULE; Now -RC1 shows the same performance 5.2.1 used to, also w/ SCHED_BSD, but with ULE it stops responding well sometimes. This first system is a dual-P4 3.0; but with older SMP systems I get a problem on 5.3 that does not happens on 5.2.1; The system is a 4-proc Pentium Pro, it installs fine, but when we start using more CPU (say, running stress(8) or building a port) it panics w/ Panic: APIC: Previous IPI is stuck cpuid: 0 boot() called on cpu#0 Uptime: 3h44m23s Cannot dump. No dump device defined Automatic reboot in 15 second - press a key ... cpu_reset called on cpu#0 cpu_reset: stopping other CPUs Panic: APIC: Previous IPI is stuck cpuid=0 boot() called on cpu#0 Under 5.2.1 it runs fine. What kind of more detailed info can I send to help tracking this behaviour? -- Atenciosamente, Patrick Tracanelli