From mboxrd@z Thu Jan  1 00:00:00 1970
From: Christoph Hellwig <hch-wEGCiKHe2LqWVfeAwA7xHQ@public.gmane.org>
Subject: Re: [PATCH v5 8/9] vfs: Add vfs_copy_file_range() support for
 pagecache copies
Date: Wed, 14 Oct 2015 11:49:46 -0700
Message-ID: <20151014184946.GA15214@infradead.org>
References: <1443634014-3026-1-git-send-email-Anna.Schumaker@Netapp.com>
 <1443634014-3026-9-git-send-email-Anna.Schumaker@Netapp.com>
 <20151011142203.GA31867@infradead.org>
 <20151012231749.GC11398@birch.djwong.org>
 <561E980C.9010509@Netapp.com>
 <CALCETrU0-Q-q0u0BAPwu+mbyS+LVp6M1Hqro3xwzSm7eNCbEmg@mail.gmail.com>
 <20151014182701.GC31225@infradead.org>
 <CALCETrWrHs6x+aZ7c=sySRCdzwdfUQZH_DL++vLL78NXXN+7FQ@mail.gmail.com>
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Cc: Christoph Hellwig <hch-wEGCiKHe2LqWVfeAwA7xHQ@public.gmane.org>,
	Anna Schumaker <Anna.Schumaker-HgOvQuBEEgTQT0dZR+AlfA@public.gmane.org>,
	"Darrick J. Wong" <darrick.wong-QHcLZuEGTsvQT0dZR+AlfA@public.gmane.org>,
	linux-nfs-u79uwXL29TY76Z2rM5mHXA@public.gmane.org,
	Linux btrfs Developers List <linux-btrfs-u79uwXL29TY76Z2rM5mHXA@public.gmane.org>,
	Linux FS Devel <linux-fsdevel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org>,
	Linux API <linux-api-u79uwXL29TY76Z2rM5mHXA@public.gmane.org>,
	Zach Brown <zab-ugsP4Wv/S6ZeoWH0uzbU5w@public.gmane.org>, Al Viro <viro-RmSDqhL/yNMiFSDQTTA3OLVCufUGDwFn@public.gmane.org>,
	Chris Mason <clm-b10kYP2dOMg@public.gmane.org>,
	Michael Kerrisk-manpages <mtk.manpages-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>,
	andros-HgOvQuBEEgTQT0dZR+AlfA@public.gmane.org
To: Andy Lutomirski <luto-kltTT9wpgjJwATOyAt5JVQ@public.gmane.org>
Return-path: <linux-api-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org>
Content-Disposition: inline
In-Reply-To: <CALCETrWrHs6x+aZ7c=sySRCdzwdfUQZH_DL++vLL78NXXN+7FQ-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
Sender: linux-api-owner-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
List-Id: linux-fsdevel.vger.kernel.org

On Wed, Oct 14, 2015 at 11:38:13AM -0700, Andy Lutomirski wrote:
> One might argue that reflink is like copy + immediate dedupe.

Not, it's not.  It's all that and more, because it is an operation that
is atomic vs other writes to the file and it's an operation that either
clones the whole range or nothing.  That's a very important difference.

> Also, I
> can imagine there being network protocols over which you can't really
> tell the difference between reflink and server-to-server copy.

For NFS we specificly have a CLONE and a COPY operations so that smart
servers can support the proper clone, and dumb servers still get copy
offload.  Other protocols might only be able to support COPY if they
don't have a CLONE primitive.  Note that a clone also always is a valid
copy, just with much simpler an at the same time more useful semantics.
Take a look at the NFSv4.2 sections for CLONE vs COPY if you're
interested.