Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Message-ID: <c493ef8fa5b735fe32be0c720be77db18e660dfb.camel@linux.ibm.com>
Subject: Re: [PATCH net-next v3 0/2] net/smc: Fix effective buffer size
From:   Gerd Bayer <gbayer@linux.ibm.com>
To:     Wenjia Zhang <wenjia@linux.ibm.com>,
        Jan Karcher <jaka@linux.ibm.com>,
        Tony Lu <tonylu@linux.alibaba.com>,
        Paolo Abeni <pabeni@redhat.com>
Cc:     Karsten Graul <kgraul@linux.ibm.com>,
        "D . Wythe" <alibuda@linux.alibaba.com>,
        Wen Gu <guwen@linux.alibaba.com>,
        "David S . Miller" <davem@davemloft.net>,
        Eric Dumazet <edumazet@google.com>,
        Jakub Kicinski <kuba@kernel.org>, linux-s390@vger.kernel.org,
        netdev@vger.kernel.org, linux-kernel@vger.kernel.org
Date:   Fri, 04 Aug 2023 18:55:26 +0200
In-Reply-To: <20230804163049.937185-1-gbayer@linux.ibm.com>
References: <20230804163049.937185-1-gbayer@linux.ibm.com>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
User-Agent: Evolution 3.48.4 (3.48.4-1.module_f38+17164+63eeee4a) 
MIME-Version: 1.0
Precedence: bulk

On Fri, 2023-08-04 at 18:30 +0200, Gerd Bayer wrote:
> Hi all,
>=20
> commit 0227f058aa29 ("net/smc: Unbind r/w buffer size from clcsock
> and make them tunable") started to derive the effective buffer size
> for
> SMC connections inconsistently in case a TCP fallback was used and
> memory consumption of SMC with the default settings was doubled when
> a connection negotiated SMC. That was not what we want.
>=20
> This series consolidates the resulting effective buffer size that is
> used with SMC sockets, which is based on Jan Karcher's effort (see=20
> [1]). For all TCP exchanges (in particular in case of a fall back
> when
> no SMC connection was possible) the values from net.ipv4.tcp_[rw]mem
> are used. If SMC succeeds in establishing a SMC connection, the newly
> introduced values from net.smc.[rw]mem are used.
>=20
> net.smc.[rw]mem is initialized to 64kB, respectively. Internal test=20
> have show this to be a good compromise between throughput/latency=20
> and memory consumption. Also net.smc.[rw]mem is now decoupled
> completely
> from any tuning through net.ipv4.tcp_[rw]mem.
>=20
> If a user chose to tune a socket's receive or send buffer size with
> setsockopt, this tuning is now consistently applied to either fall-
> back
> TCP or proper SMC connections over the socket.
>=20
> Thanks,
> Gerd=20
>=20
> v2 - v3:
> =C2=A0- Rebase to and resolve conflict of second patch with latest
> net/master.
> v1 - v2:
> =C2=A0- In second patch, use sock_net() helper as suggested by Tony and
> demanded
> =C2=A0=C2=A0 by kernel test robot.
>=20
>=20
> Gerd Bayer (2):
> =C2=A0 net/smc: Fix setsockopt and sysctl to specify same buffer size
> again
> =C2=A0 net/smc: Use correct buffer sizes when switching between TCP and
> SMC
>=20
> =C2=A0net/smc/af_smc.c=C2=A0=C2=A0=C2=A0=C2=A0 | 77 +++++++++++++++++++++=
+++++++++------------
> --
> =C2=A0net/smc/smc.h=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0=C2=A0 |=C2=A0 2 +=
-
> =C2=A0net/smc/smc_clc.c=C2=A0=C2=A0=C2=A0 |=C2=A0 4 +--
> =C2=A0net/smc/smc_core.c=C2=A0=C2=A0 | 25 +++++++-------
> =C2=A0net/smc/smc_sysctl.c | 10 ++++--
> =C2=A05 files changed, 76 insertions(+), 42 deletions(-)
>=20
>=20
> base-commit: 1733d0be68ab1b89358a3b0471ef425fd61de7c5

Oh boy,

this should have gone as v3 against "net" instead of "net-next".
Resending ASAP.

Sorry for the noise,
Gerd