From mboxrd@z Thu Jan 1 00:00:00 1970 From: Trond Myklebust Subject: Re: How to handle non-local renames? Date: Thu, 28 Sep 2006 08:57:08 -0400 Message-ID: <1159448228.5439.26.camel@lade.trondhjem.org> References: <1158597517.6297.10.camel@lade.trondhjem.org> <1158883241.5535.7.camel@lade.trondhjem.org> <20060928100223.GY29920@ftp.linux.org.uk> <1159446678.5439.23.camel@lade.trondhjem.org> <20060928124247.GD29920@ftp.linux.org.uk> Mime-Version: 1.0 Content-Type: text/plain Content-Transfer-Encoding: 7bit Cc: Miklos Szeredi , dhowells@redhat.com, linux-fsdevel@vger.kernel.org Return-path: Received: from pat.uio.no ([129.240.10.4]:12026 "EHLO pat.uio.no") by vger.kernel.org with ESMTP id S1161107AbWI1M5a (ORCPT ); Thu, 28 Sep 2006 08:57:30 -0400 To: Al Viro In-Reply-To: <20060928124247.GD29920@ftp.linux.org.uk> Sender: linux-fsdevel-owner@vger.kernel.org List-Id: linux-fsdevel.vger.kernel.org On Thu, 2006-09-28 at 13:42 +0100, Al Viro wrote: > On Thu, Sep 28, 2006 at 08:31:18AM -0400, Trond Myklebust wrote: > > > We don't _care_ if lookup() is not from rename. That's OK. > > > > You can't assume that you can always kill the subdirectories: they may > > be in use. Even if no actual processes are using them, I may have > > submounts. > > So what? Actual processes will get -ESTALE and STFU, submounts can > bloody well stay where they are; since we don't free dentries, WTF > would VFS care? Mark inodes so that nothing would be done to those > directories, unhash them and be done with it. umount -l will be > able to take them out just fine afterwards. *Or* we could even > try and play with detaching them (needed on invalidation anyway). That would mean that I can in practice unmount directories on your client simply by renaming a directory on my client. I can't see how that would be acceptable either.