From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S265421AbTLSBBo (ORCPT ); Thu, 18 Dec 2003 20:01:44 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S265422AbTLSBBo (ORCPT ); Thu, 18 Dec 2003 20:01:44 -0500 Received: from palrel10.hp.com ([156.153.255.245]:27821 "EHLO palrel10.hp.com") by vger.kernel.org with ESMTP id S265421AbTLSBBn (ORCPT ); Thu, 18 Dec 2003 20:01:43 -0500 Date: Thu, 18 Dec 2003 17:01:42 -0800 To: Martin Diehl Cc: Andrew Morton , Linux kernel mailing list Subject: Re: Linux 2.6.0 Message-ID: <20031219010142.GA1907@bougret.hpl.hp.com> Reply-To: jt@hpl.hp.com References: <20031218231650.GA828@bougret.hpl.hp.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: User-Agent: Mutt/1.3.28i Organisation: HP Labs Palo Alto Address: HP Labs, 1U-17, 1501 Page Mill road, Palo Alto, CA 94304, USA. E-mail: jt@hpl.hp.com From: Jean Tourrilhes Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org On Fri, Dec 19, 2003 at 01:31:01AM +0100, Martin Diehl wrote: > > I have converted dongle drivers on disk here. So far I was holding back > basically due to: > * 2.6.0 freeze > * neither hardware nor testers to do even basic validation > * wait for tx-lock and rawmode fixes to get applied first so we would > hopefully get some more feedback and trust in the infrastructure. Yep. Hopefully this will change as more user migrate to 2.6.X. > If you would like to accept it now I'd sent patches this weekend. I'm flying tommorow, and when I'm away, I never read my e-mail. > > o IrCOMM disconnect race condition Oops ; todo > > After some user provided helpful information off-list I found out it gets > triggered when the app doesn't close the ircomm-fd before exiting. > Just do "cat /dev/ircomm0" and kill -9 the cat to see. > > Below my current experimental patch. It works for me and I've got an user > confirmation meanwhile. It's definedly an improvement and I do think it is > as safe as we can get without reworking the whole ircomm-locking... Yes, that is more or less what I had suggested. Note that I'm unhappy about (self->flags & ASYNC_INITIALIZED), either it goes under spinlock or use test_bit(). > Martin Jean