From mboxrd@z Thu Jan 1 00:00:00 1970 From: Christoph Hellwig Subject: Re: [RFD 1/4] Pass no useless nameidata to the create, lookup, and permission IOPs Date: Sat, 30 Jun 2007 10:14:52 +0100 Message-ID: <20070630091452.GB21784@infradead.org> References: <20070626231541.697783295@suse.de> <200706270011.l5R0BfPO011424@agora.fsl.cs.sunysb.edu> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Cc: jjohansen@suse.de, linux-kernel@vger.kernel.org, linux-security-module@vger.kernel.org, linux-fsdevel@vger.kernel.org, Andreas Gruenbacher To: Erez Zadok Return-path: Content-Disposition: inline In-Reply-To: <200706270011.l5R0BfPO011424@agora.fsl.cs.sunysb.edu> Sender: linux-security-module-owner@vger.kernel.org List-Id: linux-fsdevel.vger.kernel.org On Tue, Jun 26, 2007 at 08:11:41PM -0400, Erez Zadok wrote: > Perhaps it is also time to put the dentry + mnt into a single struct path? > It's a small change, but it emphasizes that the two items here, dentry+mnt, > really define a single path to be passed around: No. The vfsmount will go away completely medium-term once we get right of the atomic open nfs hack that paases a file back in the open intent. A normal filesystem has not business looking at the vfsmount instance at all, and the raw open/create we need to replace the atomic open hack will need the full nameidata per definition, also to allow things like a multi-component lookup that various cluster/network filesystems really want.