From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S964948AbWAXIPa (ORCPT ); Tue, 24 Jan 2006 03:15:30 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S932466AbWAXIPa (ORCPT ); Tue, 24 Jan 2006 03:15:30 -0500 Received: from mustang.oldcity.dca.net ([216.158.38.3]:57067 "HELO mustang.oldcity.dca.net") by vger.kernel.org with SMTP id S932406AbWAXIP3 (ORCPT ); Tue, 24 Jan 2006 03:15:29 -0500 Subject: Re: RCU latency regression in 2.6.16-rc1 From: Lee Revell To: Ingo Molnar Cc: linux-kernel , Linus Torvalds , "Paul E. McKenney" In-Reply-To: <20060124081301.GC25855@elte.hu> References: <1138089139.2771.78.camel@mindpipe> <20060124075640.GA24806@elte.hu> <1138089483.2771.81.camel@mindpipe> <20060124080157.GA25855@elte.hu> <1138090078.2771.88.camel@mindpipe> <20060124081301.GC25855@elte.hu> Content-Type: text/plain Date: Tue, 24 Jan 2006 03:15:27 -0500 Message-Id: <1138090527.2771.91.camel@mindpipe> Mime-Version: 1.0 X-Mailer: Evolution 2.5.4 Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Tue, 2006-01-24 at 09:13 +0100, Ingo Molnar wrote: > * Lee Revell wrote: > > > On Tue, 2006-01-24 at 09:01 +0100, Ingo Molnar wrote: > > > * Lee Revell wrote: > > > > > > > On Tue, 2006-01-24 at 08:56 +0100, Ingo Molnar wrote: > > > > > * Lee Revell wrote: > > > > > > > > > > > I ported the latency tracer to 2.6.16 and got this 13ms latency within > > > > > > a few hours. This is a regression from 2.6.15. > > > > > > > > > > > > It appears that RCU can invoke ipv4_dst_destroy thousands of times in > > > > > > a single batch. > > > > > > > > > > could you try the PREEMPT_RCU patch below? > > > > > > > > Sure. If it works do you see this making it in 2.6.16? Otherwise we > > > > still would have a regression... > > > > > > nope, that likely wont make v2.6.16, which is frozen already. > > > > > > > How about just lowering maxbatch to 1000? > > does that fix the latency for you? I think "maxbatch=1000" should work > as a boot parameter too. > Have not tested yet but it appears that will reduce it substantially: $ grep "dst_destroy (dst_rcu_free)" /proc/latency_trace | wc -l 3027 This implies the latency would be reduced to ~4ms, still not great but it will be overshadowed by rt_run_flush/rt_garbage_collect. Lee