From: "J. Bruce Fields" <bfields@fieldses.org>
To: Zach Brown <zab@redhat.com>
Cc: Anna Schumaker <schumaker.anna@gmail.com>,
Szeredi Miklos <miklos@szeredi.hu>,
linux-kernel@vger.kernel.org, linux-fsdevel@vger.kernel.org,
"linux-nfs@vger.kernel.org" <linux-nfs@vger.kernel.org>,
Trond Myklebust <Trond.Myklebust@netapp.com>,
Bryan Schumaker <bjschuma@netapp.com>,
"Martin K. Petersen" <mkp@mkp.net>, Jens Axboe <axboe@kernel.dk>,
Mark Fasheh <mfasheh@suse.com>, Joel Becker <jlbec@evilplan.org>,
Eric Wong <normalperson@yhbt.net>
Subject: Re: [RFC] extending splice for copy offloading
Date: Wed, 25 Sep 2013 15:55:26 -0400 [thread overview]
Message-ID: <20130925195526.GA18971@fieldses.org> (raw)
In-Reply-To: <20130925190620.GB30372@lenny.home.zabbo.net>
On Wed, Sep 25, 2013 at 12:06:20PM -0700, Zach Brown wrote:
> On Wed, Sep 25, 2013 at 03:02:29PM -0400, Anna Schumaker wrote:
> > On Wed, Sep 25, 2013 at 2:38 PM, Zach Brown <zab@redhat.com> wrote:
> > >
> > > Hrmph. I had composed a reply to you during Plumbers but.. something
> > > happened to it :). Here's another try now that I'm back.
> > >
> > >> > Some things to talk about:
> > >> > - I really don't care about the naming here. If you do, holler.
> > >> > - We might want different flags for file-to-file splicing and acceleration
> > >>
> > >> Yes, I think "copy" and "reflink" needs to be differentiated.
> > >
> > > I initially agreed but I'm not so sure now. The problem is that we
> > > can't know whether the acceleration is copying or not. XCOPY on some
> > > array may well do some shared referencing tricks. The nfs COPY op can
> > > have a server use btrfs reflink, or ext* and XCOPY, or .. who knows. At
> > > some point we have to admit that we have no way to determine the
> > > relative durability of writes. Storage can do a lot to make writes more
> > > or less fragile that we have no visibility of. SSD FTLs can log a bunch
> > > of unrelated sectors on to one flash failure domain.
> > >
> > > And if such a flag couldn't *actually* guarantee anything for a bunch of
> > > storage topologies, well, let's not bother with it.
> > >
> > > The only flag I'm in favour of now is one that has splice return rather
> > > than falling back to manual page cache reads and writes. It's more like
> > > O_NONBLOCK than any kind of data durability hint.
> >
> > For reference, I'm planning to have the NFS server do the fallback
> > when it copies since any local copy will be faster than a read and
> > write over the network.
>
> Agreed, this is definitely the reasonable thing to do.
A client-side copy will be slower, but I guess it does have the
advantage that the application can track progress to some degree, and
abort it fairly quickly without leaving the file in a totally undefined
state--and both might be useful if the copy's not a simple constant-time
operation.
So maybe a way to pass your NONBLOCKy flag to the server would be
useful?
FWIW the protocol doesn't seem frozen yet, so I assume we could still
add an extra flag field if you think it would be worthwhile.
--b.
WARNING: multiple messages have this Message-ID (diff)
From: "J. Bruce Fields" <bfields-uC3wQj2KruNg9hUCZPvPmw@public.gmane.org>
To: Zach Brown <zab-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
Cc: Anna Schumaker
<schumaker.anna-Re5JQEeQqe8AvxtiuMwx3w@public.gmane.org>,
Szeredi Miklos <miklos-sUDqSbJrdHQHWmgEVkV9KA@public.gmane.org>,
linux-kernel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org,
linux-fsdevel-u79uwXL29TY76Z2rM5mHXA@public.gmane.org,
"linux-nfs-u79uwXL29TY76Z2rM5mHXA@public.gmane.org"
<linux-nfs-u79uwXL29TY76Z2rM5mHXA@public.gmane.org>,
Trond Myklebust
<Trond.Myklebust-HgOvQuBEEgTQT0dZR+AlfA@public.gmane.org>,
Bryan Schumaker
<bjschuma-HgOvQuBEEgTQT0dZR+AlfA@public.gmane.org>,
"Martin K. Petersen" <mkp-30zCAauEzIw@public.gmane.org>,
Jens Axboe <axboe-tSWWG44O7X1aa/9Udqfwiw@public.gmane.org>,
Mark Fasheh <mfasheh-IBi9RG/b67k@public.gmane.org>,
Joel Becker <jlbec-aKy9MeLSZ9dg9hUCZPvPmw@public.gmane.org>,
Eric Wong <normalperson-rMlxZR9MS24@public.gmane.org>
Subject: Re: [RFC] extending splice for copy offloading
Date: Wed, 25 Sep 2013 15:55:26 -0400 [thread overview]
Message-ID: <20130925195526.GA18971@fieldses.org> (raw)
In-Reply-To: <20130925190620.GB30372-fypN+1c5dIyjpB87vu3CluTW4wlIGRCZ@public.gmane.org>
On Wed, Sep 25, 2013 at 12:06:20PM -0700, Zach Brown wrote:
> On Wed, Sep 25, 2013 at 03:02:29PM -0400, Anna Schumaker wrote:
> > On Wed, Sep 25, 2013 at 2:38 PM, Zach Brown <zab-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org> wrote:
> > >
> > > Hrmph. I had composed a reply to you during Plumbers but.. something
> > > happened to it :). Here's another try now that I'm back.
> > >
> > >> > Some things to talk about:
> > >> > - I really don't care about the naming here. If you do, holler.
> > >> > - We might want different flags for file-to-file splicing and acceleration
> > >>
> > >> Yes, I think "copy" and "reflink" needs to be differentiated.
> > >
> > > I initially agreed but I'm not so sure now. The problem is that we
> > > can't know whether the acceleration is copying or not. XCOPY on some
> > > array may well do some shared referencing tricks. The nfs COPY op can
> > > have a server use btrfs reflink, or ext* and XCOPY, or .. who knows. At
> > > some point we have to admit that we have no way to determine the
> > > relative durability of writes. Storage can do a lot to make writes more
> > > or less fragile that we have no visibility of. SSD FTLs can log a bunch
> > > of unrelated sectors on to one flash failure domain.
> > >
> > > And if such a flag couldn't *actually* guarantee anything for a bunch of
> > > storage topologies, well, let's not bother with it.
> > >
> > > The only flag I'm in favour of now is one that has splice return rather
> > > than falling back to manual page cache reads and writes. It's more like
> > > O_NONBLOCK than any kind of data durability hint.
> >
> > For reference, I'm planning to have the NFS server do the fallback
> > when it copies since any local copy will be faster than a read and
> > write over the network.
>
> Agreed, this is definitely the reasonable thing to do.
A client-side copy will be slower, but I guess it does have the
advantage that the application can track progress to some degree, and
abort it fairly quickly without leaving the file in a totally undefined
state--and both might be useful if the copy's not a simple constant-time
operation.
So maybe a way to pass your NONBLOCKy flag to the server would be
useful?
FWIW the protocol doesn't seem frozen yet, so I assume we could still
add an extra flag field if you think it would be worthwhile.
--b.
--
To unsubscribe from this list: send the line "unsubscribe linux-nfs" in
the body of a message to majordomo-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
next prev parent reply other threads:[~2013-09-25 19:55 UTC|newest]
Thread overview: 111+ messages / expand[flat|nested] mbox.gz Atom feed top
2013-09-11 17:06 [RFC] extending splice for copy offloading Zach Brown
2013-09-11 17:06 ` [PATCH 1/3] splice: add DIRECT flag for splicing between files Zach Brown
2013-09-11 17:06 ` Zach Brown
2013-09-11 17:06 ` [PATCH 2/3] splice: add f_op->splice_direct Zach Brown
2013-09-11 17:06 ` Zach Brown
2013-09-11 17:06 ` [PATCH 3/3] btrfs: implement .splice_direct extent copying Zach Brown
2013-09-11 17:06 ` Zach Brown
2013-09-11 21:17 ` [RFC] extending splice for copy offloading Eric Wong
2013-09-16 19:44 ` Rob Landley
2013-09-16 19:44 ` Rob Landley
2013-09-19 12:59 ` Jeff Layton
2013-09-20 9:49 ` Szeredi Miklos
2013-09-20 9:49 ` Szeredi Miklos
2013-09-25 18:38 ` Zach Brown
2013-09-25 18:38 ` Zach Brown
2013-09-25 19:02 ` Anna Schumaker
2013-09-25 19:06 ` Zach Brown
2013-09-25 19:06 ` Zach Brown
2013-09-25 19:55 ` J. Bruce Fields [this message]
2013-09-25 19:55 ` J. Bruce Fields
2013-09-25 21:07 ` Zach Brown
2013-09-26 8:58 ` Miklos Szeredi
2013-09-26 15:34 ` J. Bruce Fields
2013-09-26 16:46 ` Ric Wheeler
2013-09-26 16:46 ` Ric Wheeler
2013-09-26 18:06 ` Miklos Szeredi
2013-09-26 19:06 ` Zach Brown
2013-09-26 19:53 ` Miklos Szeredi
2013-09-26 21:23 ` Ric Wheeler
2013-09-26 21:23 ` Ric Wheeler
2013-09-27 4:47 ` Miklos Szeredi
2013-09-27 4:47 ` Miklos Szeredi
2013-09-27 14:00 ` Ric Wheeler
2013-09-27 14:39 ` Miklos Szeredi
2013-10-06 8:42 ` Rob Landley
2013-10-06 8:42 ` Rob Landley
2013-09-26 18:55 ` Zach Brown
2013-09-26 21:26 ` Ric Wheeler
2013-09-26 21:26 ` Ric Wheeler
2013-09-27 20:05 ` J. Bruce Fields
2013-09-27 20:05 ` J. Bruce Fields
2013-09-27 20:50 ` Zach Brown
2013-09-28 5:49 ` Miklos Szeredi
2013-09-28 15:20 ` Myklebust, Trond
2013-09-28 15:20 ` Myklebust, Trond
2013-09-28 15:20 ` Myklebust, Trond
2013-09-28 21:20 ` Ric Wheeler
2013-09-30 12:20 ` Miklos Szeredi
2013-09-30 12:20 ` Miklos Szeredi
2013-09-30 14:34 ` J. Bruce Fields
2013-09-30 14:34 ` J. Bruce Fields
2013-09-30 14:48 ` Ric Wheeler
2013-09-30 14:51 ` Miklos Szeredi
2013-09-30 14:52 ` Ric Wheeler
2013-09-30 14:52 ` Ric Wheeler
2013-09-30 15:24 ` Miklos Szeredi
2013-09-30 14:28 ` Ric Wheeler
2013-09-30 15:33 ` Myklebust, Trond
2013-09-30 15:33 ` Myklebust, Trond
2013-09-30 15:33 ` Myklebust, Trond
2013-09-30 15:38 ` Miklos Szeredi
2013-09-30 15:38 ` Miklos Szeredi
2013-09-30 14:41 ` Ric Wheeler
2013-09-30 14:41 ` Ric Wheeler
2013-09-30 15:46 ` Miklos Szeredi
2013-09-30 15:46 ` Miklos Szeredi
2013-09-30 14:49 ` Ric Wheeler
2013-09-30 14:49 ` Ric Wheeler
2013-09-30 15:57 ` Miklos Szeredi
2013-09-30 15:57 ` Miklos Szeredi
2013-09-30 16:31 ` Miklos Szeredi
2013-09-30 16:31 ` Miklos Szeredi
2013-09-30 17:17 ` Bernd Schubert
2013-09-30 17:44 ` Myklebust, Trond
2013-09-30 17:44 ` Myklebust, Trond
2013-09-30 17:44 ` Myklebust, Trond
2013-09-30 17:48 ` Bernd Schubert
2013-09-30 17:48 ` Bernd Schubert
2013-09-30 18:02 ` Myklebust, Trond
2013-09-30 18:02 ` Myklebust, Trond
2013-09-30 18:02 ` Myklebust, Trond
2013-09-30 18:49 ` Bernd Schubert
2013-09-30 19:34 ` Myklebust, Trond
2013-09-30 19:34 ` Myklebust, Trond
2013-09-30 19:34 ` Myklebust, Trond
2013-09-30 20:00 ` Bernd Schubert
2013-09-30 20:00 ` Bernd Schubert
2013-09-30 20:08 ` Ric Wheeler
2013-09-30 20:08 ` Ric Wheeler
2013-09-30 20:27 ` Myklebust, Trond
2013-09-30 20:27 ` Myklebust, Trond
2013-09-30 20:27 ` Myklebust, Trond
2013-09-30 20:10 ` Myklebust, Trond
2013-09-30 20:10 ` Myklebust, Trond
2013-09-30 20:10 ` Myklebust, Trond
2013-10-01 18:42 ` J. Bruce Fields
2013-10-01 18:42 ` J. Bruce Fields
2013-10-01 19:58 ` Zach Brown
2013-10-01 19:58 ` Zach Brown
2013-10-02 12:58 ` Jan Kara
2013-10-02 12:58 ` Jan Kara
2013-10-02 13:31 ` David Lang
2013-12-18 12:41 ` Christoph Hellwig
2013-12-18 12:41 ` Christoph Hellwig
2013-12-18 17:10 ` Zach Brown
2013-12-18 17:26 ` Anna Schumaker
-- strict thread matches above, loose matches on Subject: below --
2013-09-26 17:22 Steve French
[not found] ` <CAH2r5muBuTK7ZZ+aKGC4q35gqaSWF4o07eoHypLKiNn5Y83RbQ-JsoAwUIsXosN+BqQ9rBEUg@public.gmane.org>
2013-09-26 19:34 ` David Disseldorp
2013-10-10 2:18 ` Steve French
2013-10-01 21:05 ` J. Bruce Fields
[not found] ` <20131001210531.GA7093-uC3wQj2KruNg9hUCZPvPmw@public.gmane.org>
2013-10-02 1:19 ` Steve French
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20130925195526.GA18971@fieldses.org \
--to=bfields@fieldses.org \
--cc=Trond.Myklebust@netapp.com \
--cc=axboe@kernel.dk \
--cc=bjschuma@netapp.com \
--cc=jlbec@evilplan.org \
--cc=linux-fsdevel@vger.kernel.org \
--cc=linux-kernel@vger.kernel.org \
--cc=linux-nfs@vger.kernel.org \
--cc=mfasheh@suse.com \
--cc=miklos@szeredi.hu \
--cc=mkp@mkp.net \
--cc=normalperson@yhbt.net \
--cc=schumaker.anna@gmail.com \
--cc=zab@redhat.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.