From mboxrd@z Thu Jan  1 00:00:00 1970
Content-Type: multipart/mixed; boundary="===============4075506957751719290=="
MIME-Version: 1.0
From: Walker, Benjamin <benjamin.walker at intel.com>
Subject: [SPDK] Re: SPDK socket abstraction layer
Date: Thu, 31 Oct 2019 18:54:50 +0000
Message-ID: <7c17efd207274a60451415e26dee896d2a7af22b.camel@intel.com>
In-Reply-To: f5415d8f-7633-244c-15c6-453bf240bdd3@dev.mellanox.co.il
List-ID: <spdk@lists.01.org>
To: spdk@lists.01.org

--===============4075506957751719290==
Content-Type: text/plain; charset="utf-8"
MIME-Version: 1.0
Content-Transfer-Encoding: quoted-printable

On Thu, 2019-10-31 at 16:21 +0200, Sasha Kotchubievsky wrote:
> > However, I just figured out what's wrong, so maybe you can help me fix =
it.
> > The
> > zero copy is all working mechanically, except when I get the zero copy
> > completion notification, ee_code is set to SO_EE_CODE_ZEROCOPY_COPIED. =
So
> > something is not configured correctly in my networking stack and the ke=
rnel
> > is
> > doing deferred copies instead. Any ideas what I'd need to do in order to
> > enable
> > this? I'm fairly certain I have it working at my desk on Fedora 30 in
> > loopback,
> > based on the CPU traces I'm seeing, so maybe it's just a matter of
> > installing
> > Fedora 30 on the benchmark system instead.
> =

> I don't know about any specific setting for enabling ZERO-COPY. But, let =

> me double check that.
> =

> > Is it enough to apply those two patches for zero-copy in target ?
> > =

> > Asynchronous writevhttps://review.gerrithub.io/c/spdk/spdk/+/470523
> > =

> > MSG_ZEROCOPY use in the posix implementation
> > https://review.gerrithub.io/c/spdk/spdk/+/471752
> > Let me get the patches sorted out in a nice series and then you can gra=
b the
> > top
> > of the series for testing.
> =

> This will be great.  On Sunday, we will create an environment like =

> yours  (except OS) and will check the feature.


Here's the patch at the top of the series:
https://review.gerrithub.io/c/spdk/spdk/+/471752

Zero copy is getting enabled on the socket and I do see completion
notifications, but it's always doing a deferred copy. If there was some
description somewhere of what causes the kernel to end up doing a deferred =
copy
instead of page pinning that would be really useful.

> =

> _______________________________________________
> SPDK mailing list -- spdk(a)lists.01.org
> To unsubscribe send an email to spdk-leave(a)lists.01.org


--===============4075506957751719290==--