From mboxrd@z Thu Jan 1 00:00:00 1970 From: Andreas Dilger Subject: Re: [PATCH] vfs: new O_NODE open flag Date: Fri, 25 Sep 2009 13:02:22 -0600 Message-ID: <20090925190222.GX10562@webber.adilger.int> References: <200909250223.58664.agruen@suse.de> <20090925123747.GA31228@gallifrey> <9988.1253899252@turing-police.cc.vt.edu> Mime-Version: 1.0 Content-Type: text/plain; CHARSET=US-ASCII Content-Transfer-Encoding: 7BIT Cc: "Dr. David Alan Gilbert" , Miklos Szeredi , Andreas Gruenbacher , linux-fsdevel@vger.kernel.org, linux-kernel@vger.kernel.org To: Valdis.Kletnieks@vt.edu Return-path: Received: from sca-es-mail-1.Sun.COM ([192.18.43.132]:51690 "EHLO sca-es-mail-1.sun.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1750935AbZIYTCY (ORCPT ); Fri, 25 Sep 2009 15:02:24 -0400 Content-disposition: inline In-reply-to: <9988.1253899252@turing-police.cc.vt.edu> Sender: linux-fsdevel-owner@vger.kernel.org List-ID: On Sep 25, 2009 13:20 -0400, Valdis.Kletnieks@vt.edu wrote: > On Fri, 25 Sep 2009 13:37:47 BST, "Dr. David Alan Gilbert" said: > > > Given an fd opened in this way is it possible to reopen it normally and > > be guarenteed to get the same object? > > It's not possible even without this flag. Consider: > > fd1 = open("/tmp/foo",flags); > rc = rename("/tmp/foo","/tmp/bar"); > fd2 = open("/tmp/foo",flags); > > Or were you asking if *absent that sort of tomfoolery* if it would work? No, the point is that we HAVE an fd that points to the original "/tmp/foo" opened with O_NODE, and now (after an ioctl, stat, etc) we decide it is safe to open the file read and/or write without releasing the existing fd. The whole point is to AVOID this kind of tomfoolery. Cheers, Andreas -- Andreas Dilger Sr. Staff Engineer, Lustre Group Sun Microsystems of Canada, Inc.