From mboxrd@z Thu Jan 1 00:00:00 1970 From: Alex Elder Subject: Re: [PATCH 0/7] rbd: use common code for probe and refresh Date: Wed, 08 May 2013 15:39:56 -0500 Message-ID: <518AB81C.9060307@inktank.com> References: <51885E06.8020201@inktank.com> <518AA788.7000801@inktank.com> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1 Content-Transfer-Encoding: 7bit Return-path: Received: from mail-ie0-f169.google.com ([209.85.223.169]:58925 "EHLO mail-ie0-f169.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1755781Ab3EHUj5 (ORCPT ); Wed, 8 May 2013 16:39:57 -0400 Received: by mail-ie0-f169.google.com with SMTP id u16so4192870iet.28 for ; Wed, 08 May 2013 13:39:57 -0700 (PDT) In-Reply-To: <518AA788.7000801@inktank.com> Sender: ceph-devel-owner@vger.kernel.org List-ID: To: Josh Durgin Cc: ceph-devel@vger.kernel.org On 05/08/2013 02:29 PM, Josh Durgin wrote: > On 05/06/2013 06:51 PM, Alex Elder wrote: >> This is some work I had nearly done a long time ago. It didn't even >> have a bug associated with it. I resurrected it over the weekend >> and ported it to the new code. >> >> It's basically cleanup though. For format 1 rbd images, when >> probing an image, header information for it is read in and >> translated directly into the rbd_dev->header structure. >> >> For an image refresh, instead, we use a stack structure to >> hold the translated header, and then in a second step we >> copy that into rbd_dev->header. >> >> This series gets rid of the local variable, and always just >> puts things directly into rbd_dev->header. It also simplifies >> probe and refresh for both format 1 and format 2, using a >> common rbd_dev_vX_header_info() function for both purposes. >> >> This set of patches, as well as the two single patches >> and series of six I just posted, are available in the >> "review/wip-rbd-cleanup-1" branch of the ceph-client git >> repository. >> >> -Alex >> >> [PATCH 1/7] rbd: set the mapping size and features later >> [PATCH 2/7] rbd: zero format 1 header structure earlier >> [PATCH 3/7] rbd: refactor rbd_header_from_disk() >> [PATCH 4/7] rbd: update in-core header directly >> [PATCH 5/7] rbd: simplify rbd_dev_v1_probe() >> [PATCH 6/7] rbd: get rid of trivial v1 header wrappers >> [PATCH 7/7] rbd: define rbd_dev_v1_header_info() > > These all look good. > The last one leaves the only call to rbd_dev_v2_parent_info() in > rbd_dev_v2_header_onetime(), but I'm guessing you already moved it > in your upcoming flatten-handling patches. It took me a minute to figure out your point, but yes, I move that into rbd_dev_v2_header_info() in an upcoming patch. -Alex > Reviewed-by: Josh Durgin