From: Trond Myklebust Subject: Re: [PATCH 0/2] vfs: ensure that dentries are revalidated on open (try #2) Date: Wed, 11 Nov 2009 17:26:10 +0900 Message-ID: <1257927970.5138.24.camel@heimdal.trondhjem.org> References: <1257870456-31188-1-git-send-email-jlayton@redhat.com> Mime-Version: 1.0 Content-Type: text/plain Cc: Jeff Layton , linux-kernel@vger.kernel.org, linux-fsdevel@vger.kernel.org, linux-nfs@vger.kernel.org, ebiederm@xmission.com, adobriyan@gmail.com, viro@ZenIV.linux.org.uk, jamie@shareable.org To: Miklos Szeredi Return-path: In-Reply-To: Sender: linux-fsdevel-owner@vger.kernel.org List-ID: On Wed, 2009-11-11 at 08:57 +0100, Miklos Szeredi wrote: > On Tue, 10 Nov 2009, Jeff Layton wrote: > > This is the second attempt to fix this problem. The first one attempted > > to fix this in procfs, but Eric Biederman pointed out that file bind > > mounts have a similar problem. This set attempts to fix the issue at a > > higher level, in the generic VFS layer. > > I suspect the correct fix would be to clean up the open API so that > NFSv4 doesn't have to hack its stateful open routine into the > ->lookup() and ->d_revalidate() methods. I've been working on that. I hope to have patches soon... > Having said that, doing revalidation for proc symlinks and bind mounts > (and not just for opens) might make sense. This is something similar > to FS_REVAL_DOT, so perhaps make it conditional on this flag (or a > new, appropriately named one). Aren't both proc symlinks and bind mounts pretty much guaranteed to point to a valid dentry? Once we fix the open case, I can't see that we need to do much more. Networked filesystems may want to revalidate the inode attributes, but not the dentry itself... Cheers Trond