From mboxrd@z Thu Jan 1 00:00:00 1970 From: Tejun Heo Subject: Re: [RFC] fs, proc: Introduce the /proc//map_files/ directory v2 Date: Thu, 25 Aug 2011 23:39:31 +0200 Message-ID: <20110825213931.GR2803@mtj.dyndns.org> References: <20110824085329.GL29452@sun> <20110824111806.GA7191@albatros> <20110825082944.GH10030@sun> <20110825170147.GM2803@mtj.dyndns.org> <20110825170705.GA6387@sun> <20110825205426.GO2803@mtj.dyndns.org> <20110825211213.GP2803@mtj.dyndns.org> <20110825213459.GA1929@sun> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: Vasiliy Kulikov , Nathan Lynch , Oren Laadan , Daniel Lezcano , Andrew Morton , Pavel Emelyanov , linux-kernel@vger.kernel.org, James Bottomley , LINUXFS-ML , containers@lists.osdl.org, Zan Lynx , Andi Kleen To: Cyrill Gorcunov Return-path: Content-Disposition: inline In-Reply-To: <20110825213459.GA1929@sun> Sender: linux-kernel-owner@vger.kernel.org List-Id: linux-fsdevel.vger.kernel.org Hello, On Fri, Aug 26, 2011 at 01:34:59AM +0400, Cyrill Gorcunov wrote: > > Unfortunately, not quite as easy as I expected. The information still > > seems redundant but it seems we'll need to change > > proc_inode->get_link() to take dentry instead of inode before doing > > away with proc_inode->fd, but, at any rate, I don't think this is a > > big deal one way or the other. > > Hohum... picking up an additional reference to dentry might be dangerous > I think. How exactly you imagine we would do that? (without this problem > I guess we indeed may drop or rather not change proc-inode). Why would you need an extra reference? All these data structures are created dynamically on access and dentry is always available while any operation on the inode is in progress so it's guaranteed to be available and there's no reason to diddle with reference count. Anyways, we can deal with this optimization later, I think. Thanks. -- tejun