From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S262540AbTHWK2R (ORCPT ); Sat, 23 Aug 2003 06:28:17 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S262552AbTHWK2R (ORCPT ); Sat, 23 Aug 2003 06:28:17 -0400 Received: from mail3.ithnet.com ([217.64.64.7]:11432 "HELO heather-ng.ithnet.com") by vger.kernel.org with SMTP id S262540AbTHWK2P (ORCPT ); Sat, 23 Aug 2003 06:28:15 -0400 X-Sender-Authentication: SMTPafterPOP by from 217.64.64.14 Date: Sat, 23 Aug 2003 12:28:13 +0200 From: Stephan von Krawczynski To: TeJun Huh Cc: linux-kernel@vger.kernel.org Subject: Re: Race condition in 2.4 tasklet handling (cli() broken?) Message-Id: <20030823122813.0c90e241.skraw@ithnet.com> In-Reply-To: <20030823052633.GA4307@atj.dyndns.org> References: <20030823025448.GA32547@atj.dyndns.org> <20030823040931.GA3872@atj.dyndns.org> <20030823052633.GA4307@atj.dyndns.org> Organization: ith Kommunikationstechnik GmbH X-Mailer: Sylpheed version 0.9.4 (GTK+ 1.2.10; i686-pc-linux-gnu) Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Sat, 23 Aug 2003 14:26:34 +0900 TeJun Huh wrote: > Oops, Sorry. Only bh handling is relevant. Softirq and tasklet are > not of concern to cli(). > > On Sat, Aug 23, 2003 at 01:09:31PM +0900, TeJun Huh wrote: > > Additional suspicious things. > > > > 1. tasklet_kill() has similar race condition. mb() required before > > tasklet_unlock_wait(). > > Corrected 2. > [...] > If bh hanlding is not guaranteed to be blocked during cli() - sti() > critical section, global_bh_lock test inside wait_on_irq() is > redundant and if it should be guaranteed, current implmentation seems > broken. If we follow your analysis and say it is broken, do you have a suggestion/patch how to fix both? I am willing to try your proposals, as it seems I am one of very few who really experience stability issues on SMP with the current implementation. Regards, Stephan