]> git.cworth.org Git - notmuch/blobdiff - test/corpora/lkml/cur/1382298587.001970:2,
test: add 'lkml' corpus
[notmuch] / test / corpora / lkml / cur / 1382298587.001970:2,
diff --git a/test/corpora/lkml/cur/1382298587.001970:2, b/test/corpora/lkml/cur/1382298587.001970:2,
new file mode 100644 (file)
index 0000000..707d0ad
--- /dev/null
@@ -0,0 +1,103 @@
+From: Suresh Jayaraman <sjayaraman-l3A5Bk7waGM@public.gmane.org>
+Subject: Re: [RFC][PATCH 04/10] cifs: define server-level cache index objects
+ and register them with FS-Cache
+Date: Wed, 23 Jun 2010 11:04:39 +0530
+Lines: 61
+Message-ID: <4C219CEF.5000003@suse.de>
+References: <yes>      <1277220198-3522-1-git-send-email-sjayaraman@suse.de> <20100622175214.4c56234f@corrin.poochiereds.net>
+Mime-Version: 1.0
+Content-Type: text/plain; charset=UTF-8
+Content-Transfer-Encoding: 7bit
+Cc: Steve French <smfrench-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>, linux-cifs-u79uwXL29TY76Z2rM5mHXA@public.gmane.org,
+       linux-fsdevel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org, linux-kernel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org,
+       David Howells <dhowells-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
+To: Jeff Layton <jlayton-vpEMnDpepFuMZCB2o+C8xQ@public.gmane.org>
+X-From: linux-cifs-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org Wed Jun 23 07:34:50 2010
+Return-path: <linux-cifs-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org>
+Envelope-to: glkc-linux-cifs-1dZseelyfdZg9hUCZPvPmw@public.gmane.org
+Received: from vger.kernel.org ([209.132.180.67])
+       by lo.gmane.org with esmtp (Exim 4.69)
+       (envelope-from <linux-cifs-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org>)
+       id 1ORIbp-0002v4-3W
+       for glkc-linux-cifs-1dZseelyfdZg9hUCZPvPmw@public.gmane.org; Wed, 23 Jun 2010 07:34:49 +0200
+Received: (majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org) by vger.kernel.org via listexpand
+       id S1750954Ab0FWFes (ORCPT <rfc822;glkc-linux-cifs@m.gmane.org>);
+       Wed, 23 Jun 2010 01:34:48 -0400
+Received: from cantor2.suse.de ([195.135.220.15]:58263 "EHLO mx2.suse.de"
+       rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP
+       id S1750809Ab0FWFes (ORCPT <rfc822;linux-cifs-u79uwXL29TY76Z2rM5mHXA@public.gmane.org>);
+       Wed, 23 Jun 2010 01:34:48 -0400
+Received: from relay1.suse.de (charybdis-ext.suse.de [195.135.221.2])
+       by mx2.suse.de (Postfix) with ESMTP id 8C18386A2E;
+       Wed, 23 Jun 2010 07:34:46 +0200 (CEST)
+User-Agent: Mozilla/5.0 (X11; U; Linux x86_64; en-US; rv:1.9.1.5) Gecko/20091130 SUSE/3.0.0-1.1.1 Thunderbird/3.0
+In-Reply-To: <20100622175214.4c56234f-4QP7MXygkU+dMjc06nkz3ljfA9RmPOcC@public.gmane.org>
+Sender: linux-cifs-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
+Precedence: bulk
+List-ID: <linux-cifs.vger.kernel.org>
+X-Mailing-List: linux-cifs-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
+Archived-At: <http://permalink.gmane.org/gmane.linux.kernel/1001998>
+
+On 06/23/2010 03:22 AM, Jeff Layton wrote:
+> On Tue, 22 Jun 2010 20:53:18 +0530
+> Suresh Jayaraman <sjayaraman-l3A5Bk7waGM@public.gmane.org> wrote:
+> 
+>> Define server-level cache index objects (as managed by TCP_ServerInfo structs).
+>> Each server object is created in the CIFS top-level index object and is itself
+>> an index into which superblock-level objects are inserted.
+>>
+>> Currently, the server objects are keyed by hostname.
+>>
+>> Signed-off-by: Suresh Jayaraman <sjayaraman-l3A5Bk7waGM@public.gmane.org>
+>> ---
+>>  fs/cifs/Makefile   |    2 +-
+>>  fs/cifs/cache.c    |   25 +++++++++++++++++++++++++
+>>  fs/cifs/cifsglob.h |    3 +++
+>>  fs/cifs/connect.c  |    4 ++++
+>>  fs/cifs/fscache.c  |   47 +++++++++++++++++++++++++++++++++++++++++++++++
+>>  fs/cifs/fscache.h  |   12 ++++++++++++
+>>  6 files changed, 92 insertions(+), 1 deletion(-)
+>>  create mode 100644 fs/cifs/fscache.c
+>>
+>> Index: cifs-2.6/fs/cifs/Makefile
+>> ===================================================================
+>> --- cifs-2.6.orig/fs/cifs/Makefile
+>> +++ cifs-2.6/fs/cifs/Makefile
+>> @@ -12,4 +12,4 @@ cifs-$(CONFIG_CIFS_UPCALL) += cifs_spneg
+>>  
+>>  cifs-$(CONFIG_CIFS_DFS_UPCALL) += dns_resolve.o cifs_dfs_ref.o
+>>  
+>> -cifs-$(CONFIG_CIFS_FSCACHE) += cache.o
+>> +cifs-$(CONFIG_CIFS_FSCACHE) += fscache.o cache.o
+>> Index: cifs-2.6/fs/cifs/cache.c
+>> ===================================================================
+>> --- cifs-2.6.orig/fs/cifs/cache.c
+>> +++ cifs-2.6/fs/cifs/cache.c
+>> @@ -51,3 +51,28 @@ void cifs_fscache_unregister(void)
+>>     fscache_unregister_netfs(&cifs_fscache_netfs);
+>>  }
+>>  
+>> +/*
+>> + * Server object currently keyed by hostname
+>> + */
+>> +static uint16_t cifs_server_get_key(const void *cookie_netfs_data,
+>> +                              void *buffer, uint16_t maxbuf)
+>> +{
+>> +   const struct TCP_Server_Info *server = cookie_netfs_data;
+>> +   uint16_t len = strnlen(server->hostname, sizeof(server->hostname));
+>> +
+> 
+> Would a tuple of address/family/port be a better choice here? Imagine I
+> mount "foo" and then later mount "foor.bar.baz". If they are the same
+> address and only the UNC differs, then you won't get the benefit of
+> the cache, right?
+> 
+
+Good point. I'll fix it up when I do a respin.
+
+Thanks,
+
+-- 
+Suresh Jayaraman
+
+