From mboxrd@z Thu Jan  1 00:00:00 1970
Content-Type: multipart/mixed; boundary="===============1330087012963576425=="
MIME-Version: 1.0
From: Florian Westphal <fw at strlen.de>
To: mptcp at lists.01.org
Subject: [MPTCP] Re: [PATCH mptcp-next 2/2] mptcp: add receive buffer
 auto-tuning
Date: Thu, 28 May 2020 19:55:18 +0200
Message-ID: <20200528175518.GN2915@breakpoint.cc>
In-Reply-To: 9d1f3c59c12898b7bf8c7fe23cc8a9ae30ea7340.camel@redhat.com
X-Status: 
X-Keywords:                  
X-UID: 4537

--===============1330087012963576425==
Content-Type: text/plain; charset="utf-8"
MIME-Version: 1.0
Content-Transfer-Encoding: quoted-printable

Paolo Abeni <pabeni(a)redhat.com> wrote:
> On Thu, 2020-05-28 at 15:02 +0200, Florian Westphal wrote:
> > +static void mptcp_rcv_space_adjust(struct mptcp_sock *msk, int copied)
> > +{
> > +	struct mptcp_subflow_context *subflow;
> > +	struct sock *sk =3D (struct sock *)msk;
> > +	u32 time, advmss =3D 1;
> > +	u64 rtt_us, mstamp;
> > +
> > +	sock_owned_by_me(sk);
> > +
> > +	if (copied <=3D 0)
> > +		return;
> > +
> > +	msk->rcvq_space.copied +=3D copied;
> > +
> > +	rtt_us =3D 0;
> > +	mptcp_for_each_subflow(msk, subflow) {
> > +		const struct tcp_sock *tp;
> > +		u64 sf_rtt_us;
> > +		u32 sf_advmss;
> > +
> > +		tp =3D tcp_sk(mptcp_subflow_tcp_sock(subflow));
> > +
> > +		sf_rtt_us =3D READ_ONCE(tp->rcv_rtt_est.rtt_us);
> > +		sf_advmss =3D READ_ONCE(tp->advmss);
> > +
> > +		rtt_us =3D max(sf_rtt_us, rtt_us);
> > +		advmss =3D max(sf_advmss, advmss);
> > +	}
> > +
> > +	mstamp =3D div_u64(tcp_clock_ns(), NSEC_PER_USEC);
> > +
> > +	time =3D tcp_stamp_us_delta(mstamp, msk->rcvq_space.time);
> > +	if (time < (rtt_us >> 3) || rtt_us =3D=3D 0)
> > +		return;
> =

> what if you cache the rtt_ns value, too, and use it to move the above
> test before the loop? So that the receive path will loop on the subflow
> at most once every rtt_us/8.

Which rtt_us should I cache?  The one used during last measurement
window?
--===============1330087012963576425==--