From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1753440AbbIJCJ1 (ORCPT ); Wed, 9 Sep 2015 22:09:27 -0400 Received: from g1t6225.austin.hp.com ([15.73.96.126]:42000 "EHLO g1t6225.austin.hp.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751429AbbIJCJZ (ORCPT ); Wed, 9 Sep 2015 22:09:25 -0400 X-Greylist: delayed 448 seconds by postgrey-1.27 at vger.kernel.org; Wed, 09 Sep 2015 22:09:25 EDT Message-ID: <55F0E650.1030207@hpe.com> Date: Wed, 09 Sep 2015 22:09:20 -0400 From: Waiman Long User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:10.0.12) Gecko/20130109 Thunderbird/10.0.12 MIME-Version: 1.0 To: Dave Chinner CC: Peter Zijlstra , Linus Torvalds , Linux Kernel Mailing List , Waiman Long , Ingo Molnar Subject: Re: [4.2, Regression] Queued spinlocks cause major XFS performance regression References: <20150904054820.GY3902@dastard> <20150904073917.GA18489@twins.programming.kicks-ass.net> <20150904081234.GA3902@dastard> <20150904113233.GT3644@twins.programming.kicks-ass.net> <20150906234708.GP26895@dastard> In-Reply-To: <20150906234708.GP26895@dastard> Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On 09/06/2015 07:47 PM, Dave Chinner wrote: > On Fri, Sep 04, 2015 at 01:32:33PM +0200, Peter Zijlstra wrote: >> On Fri, Sep 04, 2015 at 06:12:34PM +1000, Dave Chinner wrote: >>> You probably don't even need a VM to reproduce it - that would >>> certainly be an interesting counterpoint if it didn't.... >> Even though you managed to restore your DEBUG_SPINLOCK performance by >> changing virt_queued_spin_lock() to use __delay(1), I ran the thing on >> actual hardware just to test. >> >> [ Note: In any case, I would recommend you use (or at least try) >> PARAVIRT_SPINLOCKS if you use VMs, as that is where we were looking for >> performance, the test-and-set fallback really wasn't meant as a >> performance option (although it clearly sucks worse than expected). > FSUse% Count Size Files/sec App Overhead > 0 1600000 0 319431.5 10116018 > 0 3200000 0 307824.5 10054299 > 0 4800000 0 296971.5 10770197 > 0 6400000 0 281653.6 11748423 > .... > > PARAVIRT_SPINLOCKS seems to work OK these days, too. I'll leave that > set so I'll end up testing whatever comes along down that pipe... > > Cheers, > > Dave. I am working on patches to improve PV qspinlock performance and will run your fstest to verify that there will be no regression. Cheers, Longman