From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1752175Ab2IYBqv (ORCPT ); Mon, 24 Sep 2012 21:46:51 -0400 Received: from out1-smtp.messagingengine.com ([66.111.4.25]:53688 "EHLO out1-smtp.messagingengine.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751754Ab2IYBqt (ORCPT ); Mon, 24 Sep 2012 21:46:49 -0400 X-Sasl-enc: RZFVoTn6P1lUVr7Y1d64bqWz17ZQcloH7DZ+qTw3ejrX 1348537607 Message-ID: <1348537604.2307.8.camel@perseus.themaw.net> Subject: Re: [PATCH 0/2] struct pid-ify autofs4 From: Ian Kent To: Miklos Szeredi Cc: autofs@vger.kernel.org, linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org, sukadev@linux.vnet.ibm.com, serge.hallyn@canonical.com, ebiederm@xmission.com Date: Tue, 25 Sep 2012 09:46:44 +0800 In-Reply-To: <87d31bczo7.fsf@tucsk.pomaz.szeredi.hu> References: <1348068576-21363-1-git-send-email-miklos@szeredi.hu> <87haqrwfbj.fsf@tucsk.pomaz.szeredi.hu> <1348447130.2318.1.camel@perseus.themaw.net> <87d31bczo7.fsf@tucsk.pomaz.szeredi.hu> Content-Type: text/plain; charset="UTF-8" X-Mailer: Evolution 3.2.3 (3.2.3-3.fc16) Content-Transfer-Encoding: 7bit Mime-Version: 1.0 Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org On Mon, 2012-09-24 at 15:34 +0200, Miklos Szeredi wrote: > Ian Kent writes: > > > On Fri, 2012-09-21 at 17:44 +0200, Miklos Szeredi wrote: > >> Miklos Szeredi writes: > >> > >> > These two patches change autofs4 to store struct pid pointers instead of pid_t > >> > values. > >> > > >> > Fixed various issues with the previous post. Not tested, handle with > >> > care! > >> > >> Customer gave positive test results. > > > > For what exactly, there's no problem description in these patches? > > From what I understand (and I'm not an expert by any means) is that > autofs doesn't work if containers are used. The first patch fixes this. Yeah, the problem with that is that "autofs doesn't work if containers are used" is ill defined since there are use cases where it does, I believe. At the very least, ill defined in my view of things. But I can't even sensibly discuss it because of the lack of specified use cases and requirements for each. So, there's a chance this will break another case that does work. All I can do is ask annoying questions each time time I see related patches. > > Both the patches replace pid_t with a refcounted struct pid object, > which has better lifetime properties: you don't know whether a pid_t is > valid, because pid numbers are reused, while pid objects remain valid > until there are no more references to them. Yep, at least I got that. > > Thanks, > Miklos