From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S262570AbTKDVkU (ORCPT ); Tue, 4 Nov 2003 16:40:20 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S262598AbTKDVkU (ORCPT ); Tue, 4 Nov 2003 16:40:20 -0500 Received: from 209-166-240-202.cust.walrus.com ([209.166.240.202]:19945 "EHLO ti3.telemetry-investments.com") by vger.kernel.org with ESMTP id S262570AbTKDVkQ (ORCPT ); Tue, 4 Nov 2003 16:40:16 -0500 Date: Tue, 4 Nov 2003 16:39:59 -0500 From: "Bill Rugolsky Jr." To: Linus Torvalds Cc: Paul Venezia , linux-kernel@vger.kernel.org Subject: Re: ext3 performance inconsistencies, 2.4/2.6 Message-ID: <20031104213959.GD30612@ti19.telemetry-investments.com> Reply-To: "Bill Rugolsky Jr." Mail-Followup-To: "Bill Rugolsky Jr." , Linus Torvalds , Paul Venezia , linux-kernel@vger.kernel.org References: <20031104202037.GB30612@ti19.telemetry-investments.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.4i Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Tue, Nov 04, 2003 at 12:30:23PM -0800, Linus Torvalds wrote: > But there really should be zero contention on the stdio data structures, > so the locking would have to be _seriously_ broken to make that kind o > fdifference (not necessarily buggy, but seriously badly implemented). > > A non-contended lock should be at most one locked instruction if well > done, both on LinuxThreads and NPTL. The results that I just posted are also for Red Hat 9, kernel 2.4.20-20.9. rugolsky@ti31: getconf GNU_LIBPTHREAD_VERSION NPTL 0.34 Ulrich's release notes for nptl-0.57 says: The changes are numerous and most of them were made by Jakub: ... ~ better stdio locking I don't have my laptop running Fedora handy, but that's the next thing to test. Regards, Bill Rugolsky