From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S262563AbTD3XuD (ORCPT ); Wed, 30 Apr 2003 19:50:03 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S262568AbTD3XuD (ORCPT ); Wed, 30 Apr 2003 19:50:03 -0400 Received: from [12.47.58.20] ([12.47.58.20]:3703 "EHLO pao-ex01.pao.digeo.com") by vger.kernel.org with ESMTP id S262563AbTD3XuD (ORCPT ); Wed, 30 Apr 2003 19:50:03 -0400 Date: Wed, 30 Apr 2003 16:59:14 -0700 From: Andrew Morton To: viro@parcelfarce.linux.theplanet.co.uk Cc: ricklind@us.ibm.com, solt@dns.toxicfilms.tv, linux-kernel@vger.kernel.org, frankeh@us.ibm.com Subject: Re: must-fix list for 2.6.0 Message-Id: <20030430165914.2facc464.akpm@digeo.com> In-Reply-To: <20030430234746.GW10374@parcelfarce.linux.theplanet.co.uk> References: <20030430121105.454daee1.akpm@digeo.com> <200304302311.h3UNB2H27134@owlet.beaverton.ibm.com> <20030430162108.09dbd019.akpm@digeo.com> <20030430234746.GW10374@parcelfarce.linux.theplanet.co.uk> X-Mailer: Sylpheed version 0.8.9 (GTK+ 1.2.10; i586-pc-linux-gnu) Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit X-OriginalArrivalTime: 01 May 2003 00:02:17.0403 (UTC) FILETIME=[EDA348B0:01C30F74] Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org viro@parcelfarce.linux.theplanet.co.uk wrote: > > On Wed, Apr 30, 2003 at 04:21:08PM -0700, Andrew Morton wrote: > > menu if there was a kernel build happening at the same time. That is just > > utterly broken, so if we're going to leave the sched.c code as-is then we > > *require* that all applications be updated to not spin on sched_yield. > > Excuse me, but WTF do they spin on the sched_yield() in the first place? > _That_ sounds like utterly broken... I think it's happening down inside the old linuxthreads library. No idea who, what, where or why. There are quite a few places in the kernel which do it, too. Usually when waiting for memory to come free. These are being gradually removed, in favour of blk_congestion_wait() calls. That leaves behind the very performance-critical sched_yield() in ext3 transaction batching. That was designed to allow other processes to join a transaction before the calling one closes the transaction. With the new yield() it was causing horrid starvation and was lamely replaced with a schedule(). It needs to be resurrected for real, but I'm not sure how. Probably just a sleep(0.01).