1From: Jeff Layton <jlayton-eUNUBHrolfbYtjvyW6yDsg@public.gmane.org> 2Subject: Re: [RFC][PATCH 06/10] cifs: define inode-level cache object and 3 register them 4Date: Fri, 25 Jun 2010 12:53:06 -0400 5Lines: 36 6Message-ID: <20100625125306.7f9b1966@tlielax.poochiereds.net> 7References: <4C24A606.5040001@suse.de> 8 <1277220214-3597-1-git-send-email-sjayaraman@suse.de> 9 <yes> 10 <9822.1277312573@redhat.com> 11 <22697.1277470549@redhat.com> 12Mime-Version: 1.0 13Content-Type: text/plain; charset=US-ASCII 14Content-Transfer-Encoding: 7bit 15Cc: Suresh Jayaraman <sjayaraman-l3A5Bk7waGM@public.gmane.org>, 16 Steve French <smfrench-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>, linux-cifs-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, 17 linux-fsdevel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, linux-kernel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, 18 samba-technical-w/Ol4Ecudpl8XjKLYN78aQ@public.gmane.org 19To: David Howells <dhowells-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org> 20X-From: linux-cifs-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org Fri Jun 25 18:53:12 2010 21Return-path: <linux-cifs-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org> 22Envelope-to: glkc-linux-cifs-1dZseelyfdZg9hUCZPvPmw@public.gmane.org 23Received: from vger.kernel.org ([209.132.180.67]) 24 by lo.gmane.org with esmtp (Exim 4.69) 25 (envelope-from <linux-cifs-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org>) 26 id 1OSC9P-0005Eb-SU 27 for glkc-linux-cifs-1dZseelyfdZg9hUCZPvPmw@public.gmane.org; Fri, 25 Jun 2010 18:53:12 +0200 28Received: (majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org) by vger.kernel.org via listexpand 29 id S932199Ab0FYQxK (ORCPT <rfc822;glkc-linux-cifs@m.gmane.org>); 30 Fri, 25 Jun 2010 12:53:10 -0400 31Received: from cdptpa-omtalb.mail.rr.com ([75.180.132.122]:53512 "EHLO 32 cdptpa-omtalb.mail.rr.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org 33 with ESMTP id S932187Ab0FYQxJ (ORCPT 34 <rfc822;linux-cifs-u79uwXL29TY76Z2rM5mHXA@public.gmane.org>); Fri, 25 Jun 2010 12:53:09 -0400 35X-Authority-Analysis: v=1.0 c=1 a=iVNVO0OCT3kA:10 a=yQWWgrYGNuUA:10 a=kj9zAlcOel0A:10 a=20KFwNOVAAAA:8 a=hGzw-44bAAAA:8 a=f0L6POiToRdS6aViIA4A:9 a=tdNtT7bw1iHNm6ggrCkIte35EhAA:4 a=CjuIK1q_8ugA:10 a=jEp0ucaQiEUA:10 a=0kPLrQdw3YYA:10 a=dowx1zmaLagA:10 a=00U40p1LBqVLw4jT:21 a=gh7LVOPznGai4vo_:21 36X-Cloudmark-Score: 0 37X-Originating-IP: 71.70.153.3 38Received: from [71.70.153.3] ([71.70.153.3:42266] helo=mail.poochiereds.net) 39 by cdptpa-oedge01.mail.rr.com (envelope-from <jlayton-eUNUBHrolfbYtjvyW6yDsg@public.gmane.org>) 40 (ecelerity 2.2.2.39 r()) with ESMTP 41 id 2D/E0-24471-3FED42C4; Fri, 25 Jun 2010 16:53:08 +0000 42Received: from tlielax.poochiereds.net (tlielax.poochiereds.net [192.168.1.3]) 43 by mail.poochiereds.net (Postfix) with ESMTPS id E9B19580FA; 44 Fri, 25 Jun 2010 12:53:06 -0400 (EDT) 45In-Reply-To: <22697.1277470549-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org> 46X-Mailer: Claws Mail 3.7.6 (GTK+ 2.20.1; x86_64-redhat-linux-gnu) 47Sender: linux-cifs-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org 48Precedence: bulk 49List-ID: <linux-cifs.vger.kernel.org> 50X-Mailing-List: linux-cifs-u79uwXL29TY76Z2rM5mHXA@public.gmane.org 51Archived-At: <http://permalink.gmane.org/gmane.linux.kernel/1003033> 52 53On Fri, 25 Jun 2010 13:55:49 +0100 54David Howells <dhowells-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org> wrote: 55 56> Suresh Jayaraman <sjayaraman-l3A5Bk7waGM@public.gmane.org> wrote: 57> 58> > I think the creation time is currently being ignored as we won't be able 59> > to accomodate in POSIX stat struct. 60> 61> The FS-Cache interface doesn't use the POSIX stat struct, but it could be 62> really useful to save it and use it for cache coherency inside the kernel. 63> 64> Out of interest, what does Samba do when it comes to generating a creation time 65> for UNIX where one does not exist? 66> 67 68(cc'ing samba-technical since we're talking about the create time) 69 70Looks like it mostly uses the ctime. IMO, the mtime would be a better 71choice since it changes less frequently, but I don't guess that it 72matters very much. 73 74I have a few patches that make the cifs_iget code do more stringent 75checks. One of those makes it use the create time like an i_generation 76field to guard against matching inodes that have the same number but 77that have undergone a delete/create cycle. They need a bit more testing 78but I'm planning to post them in time for 2.6.36. 79 80Because of how samba generates this number, it could be somewhat 81problematic to do this. What may save us though is that Linux<->Samba 82mostly uses unix extensions unless someone has specifically disabled 83them on either end. The unix extension calls don't generally send any 84sort of create time field, so we can't rely on it in those codepaths 85anyway. 86 87-- 88Jeff Layton <jlayton-eUNUBHrolfbYtjvyW6yDsg@public.gmane.org> 89 90 91