From owner-freebsd-current@FreeBSD.ORG Thu Jul 29 03:30:06 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 B44AD16A4CF for ; Thu, 29 Jul 2004 03:30:06 +0000 (GMT) Received: from cpanel.ezone.ru (cpanel.ezone.ru [213.85.31.234]) by mx1.FreeBSD.org (Postfix) with ESMTP id 9E3F443D67 for ; Thu, 29 Jul 2004 03:30:05 +0000 (GMT) (envelope-from mcsi@mcsi.pp.ru) Received: from [81.195.3.106] (ppp3-106.pppoe.mtu-net.ru [81.195.3.106]) (authenticated bits=0) by cpanel.ezone.ru (8.13.0/8.12.11) with ESMTP id i6T3TkmQ002195 for ; Thu, 29 Jul 2004 07:29:47 +0400 (MSD) (envelope-from mcsi@mcsi.pp.ru) Message-ID: <41086F25.3070707@mcsi.pp.ru> Date: Thu, 29 Jul 2004 07:29:41 +0400 From: Maxim Maximov User-Agent: Mozilla/5.0 (X11; U; FreeBSD i386; en-US; rv:1.7.1) Gecko/20040720 X-Accept-Language: ru, en-us, en MIME-Version: 1.0 To: current@freebsd.org References: <200407281329.27271.jhb@FreeBSD.org> In-Reply-To: <200407281329.27271.jhb@FreeBSD.org> Content-Type: text/plain; charset=us-ascii; format=flowed Content-Transfer-Encoding: 7bit X-Spam-Status: No, hits=-4.9 required=5.0 tests=BAYES_00 autolearn=ham version=2.63 X-Spam-Checker-Version: SpamAssassin 2.63 (2004-01-11) on mail3.ezone.ru Subject: Re: A couple of SMP deadlock patches 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: Thu, 29 Jul 2004 03:30:06 -0000 John Baldwin wrote: > Included below are two patches for two different potential deadlocks on SMP. > The first one I have sent to the list before, but it didn't seem to help > anyone. I do believe it is a fix for a possible deadlock though so I'd like > some more testing feedback with the aim of committing it. The second patch > works around a lock order reversal introduced with the sleep queue code > (since sched_lock's scope was reduced) at the expense of increasing the > latency between a swapped out thread becoming runnable and the swapper waking > up to swap the process back in. I'm sorry, but for me PREEMPTION defined in kernel with these patches is still causing deadlocks. SMP, HTT enabled. -- Maxim Maximov