From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S264075AbTDWPRT (ORCPT ); Wed, 23 Apr 2003 11:17:19 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S264080AbTDWPRT (ORCPT ); Wed, 23 Apr 2003 11:17:19 -0400 Received: from [81.80.245.157] ([81.80.245.157]:15551 "EHLO smtp.alcove-fr") by vger.kernel.org with ESMTP id S264075AbTDWPRR (ORCPT ); Wed, 23 Apr 2003 11:17:17 -0400 Date: Wed, 23 Apr 2003 17:29:14 +0200 From: Stelian Pop To: Ben Collins Cc: Linux Kernel Mailing List , Marcelo Tosatti Subject: Re: IEEE-1394 problem on init [ was Re: Linux 2.4.21-rc1 ] Message-ID: <20030423152914.GM820@hottah.alcove-fr> Reply-To: Stelian Pop Mail-Followup-To: Stelian Pop , Ben Collins , Linux Kernel Mailing List , Marcelo Tosatti References: <20030423125315.GH820@hottah.alcove-fr> <20030423130139.GD354@phunnypharm.org> <20030423132227.GI820@hottah.alcove-fr> <20030423133256.GG354@phunnypharm.org> <20030423135814.GJ820@hottah.alcove-fr> <20030423135448.GI354@phunnypharm.org> <20030423142131.GK820@hottah.alcove-fr> <20030423142353.GL354@phunnypharm.org> <20030423145122.GL820@hottah.alcove-fr> <20030423144857.GN354@phunnypharm.org> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20030423144857.GN354@phunnypharm.org> User-Agent: Mutt/1.3.25i Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Wed, Apr 23, 2003 at 10:48:58AM -0400, Ben Collins wrote: > > As for 2.4.21, well, we want something pretty well tested. Will this > > be the case with your new mega-patch ? I don't think so. The safest > > is to go back to a version which worked. At least the bugs of that > > version are known, which is not the case for the new version. > > BTW, have you even tested the patch? I can almost guarantee is is more > stable than what was in -pre7 (outside of the one small fix I had to > apply for the IRM looping). The -pre7 code has loads of irq disabling > problems and dead lock issues, not to mention the race conditions. > > The problem you see with the irq disabling around kernel_thread() may > not be there in -pre7, but that's only because the shared data with the > thread was not protected from a race condition that causes an oops in > some not-so-rare cases. I confirm that your patch at least solves the initialisation issues. I'll test later with some ieee devices and I'll report back if I found other issues. Stelian. -- Stelian Pop Alcove - http://www.alcove.com