From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1750707AbWEKS5r (ORCPT ); Thu, 11 May 2006 14:57:47 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1750713AbWEKS5r (ORCPT ); Thu, 11 May 2006 14:57:47 -0400 Received: from dsl027-180-168.sfo1.dsl.speakeasy.net ([216.27.180.168]:40883 "EHLO sunset.davemloft.net") by vger.kernel.org with ESMTP id S1750707AbWEKS5q (ORCPT ); Thu, 11 May 2006 14:57:46 -0400 Date: Thu, 11 May 2006 11:57:40 -0700 (PDT) Message-Id: <20060511.115740.30658747.davem@davemloft.net> To: akpm@osdl.org Cc: stern@rowland.harvard.edu, sekharan@us.ibm.com, jes@sgi.com, linux-kernel@vger.kernel.org Subject: Re: [PATCH] Allow raw_notifier callouts to unregister themselves From: "David S. Miller" In-Reply-To: <20060511112509.6c9db883.akpm@osdl.org> References: <20060511112509.6c9db883.akpm@osdl.org> X-Mailer: Mew version 4.2.53 on Emacs 21.4 / Mule 5.0 (SAKAKI) 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 From: Andrew Morton Date: Thu, 11 May 2006 11:25:09 -0700 > Alan Stern wrote: > > > > Since raw_notifier chains don't benefit from any centralized locking > > protections, they shouldn't suffer from the associated limitations. > > Under some circumstances it might make sense for a raw_notifier callout > > routine to unregister itself from the notifier chain. This patch (as678) > > changes the notifier core to allow for such things. > > ok... Can you see any reason why 2.6.17 needs this? If this patch makes raw notifiers behave more closely to the way notifiers did before the notifier patch went into 2.6.17, we should seriouly consider it. We've had enough regressions from that patch, and anything which minimizes any possible other such regressions would be a plus.