Message-ID: <795aed3f0e433a89fb72a8af3fc736f58dea1bf1.camel@redhat.com>
Subject: Re: [PATCH net] udp: fix memory schedule error
From:   Paolo Abeni <pabeni@redhat.com>
To:     Jason Xing <kerneljasonxing@gmail.com>
Cc:     willemdebruijn.kernel@gmail.com, davem@davemloft.net,
        dsahern@kernel.org, edumazet@google.com, kuba@kernel.org,
        netdev@vger.kernel.org, linux-kernel@vger.kernel.org,
        bpf@vger.kernel.org, Jason Xing <kernelxing@tencent.com>
Date:   Thu, 23 Feb 2023 09:39:17 +0100
In-Reply-To: <CAL+tcoBGFkXea-GyzbO41Ve8_wUF3PT=YF43TxuzgM+adVa8gw@mail.gmail.com>
References: <20230221110344.82818-1-kerneljasonxing@gmail.com>
         <48429c16fdaee59867df5ef487e73d4b1bf099af.camel@redhat.com>
         <CAL+tcoD8PzL4khHq44z27qSHHGkcC4YUa91E3h+ki7O0u3SshQ@mail.gmail.com>
         <aaf3d11ea5b247ab03d117dadae682fe2180d38a.camel@redhat.com>
         <CAL+tcoBZFFwOnUqzcDtSsNyfPgHENAOv0bPcvncxuMPwCn40+Q@mail.gmail.com>
         <CAL+tcoBGFkXea-GyzbO41Ve8_wUF3PT=YF43TxuzgM+adVa8gw@mail.gmail.com>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
User-Agent: Evolution 3.46.4 (3.46.4-1.fc37) 
MIME-Version: 1.0
Precedence: bulk

On Wed, 2023-02-22 at 11:47 +0800, Jason Xing wrote:
> On Tue, Feb 21, 2023 at 11:46 PM Jason Xing <kerneljasonxing@gmail.com> w=
rote:
> >=20
> > On Tue, Feb 21, 2023 at 10:46 PM Paolo Abeni <pabeni@redhat.com> wrote:
> > >=20
> > > On Tue, 2023-02-21 at 21:39 +0800, Jason Xing wrote:
> > > > On Tue, Feb 21, 2023 at 8:27 PM Paolo Abeni <pabeni@redhat.com> wro=
te:
> > > > >=20
> > > > > On Tue, 2023-02-21 at 19:03 +0800, Jason Xing wrote:
> > > > > > From: Jason Xing <kernelxing@tencent.com>
> > > > > >=20
> > > > > > Quoting from the commit 7c80b038d23e ("net: fix sk_wmem_schedul=
e()
> > > > > > and sk_rmem_schedule() errors"):
> > > > > >=20
> > > > > > "If sk->sk_forward_alloc is 150000, and we need to schedule 150=
001 bytes,
> > > > > > we want to allocate 1 byte more (rounded up to one page),
> > > > > > instead of 150001"
> > > > >=20
> > > > > I'm wondering if this would cause measurable (even small) perform=
ance
> > > > > regression? Specifically under high packet rate, with BH and user=
-space
> > > > > processing happening on different CPUs.
> > > > >=20
> > > > > Could you please provide the relevant performance figures?
> > > >=20
> > > > Sure, I've done some basic tests on my machine as below.
> > > >=20
> > > > Environment: 16 cpus, 60G memory
> > > > Server: run "iperf3 -s -p [port]" command and start 500 processes.
> > > > Client: run "iperf3 -u -c 127.0.0.1 -p [port]" command and start 50=
0 processes.
> > >=20
> > > Just for the records, with the above command each process will send
> > > pkts at 1mbs - not very relevant performance wise.
> > >=20
> > > Instead you could do:
> > >=20
> >=20
> > > taskset 0x2 iperf -s &
> > > iperf -u -c 127.0.0.1 -b 0 -l 64
> > >=20
> >=20
> > Thanks for your guidance.
> >=20
> > Here're some numbers according to what you suggested, which I tested
> > several times.
> > ----------|IFACE   rxpck/s   txpck/s    rxkB/s    txkB/s
> > Before: lo 411073.41 411073.41  36932.38  36932.38
> > After:   lo 410308.73 410308.73  36863.81  36863.81
> >=20
> > Above is one of many results which does not mean that the original
> > code absolutely outperforms.
> > The output is not that constant and stable, I think.
>=20
> Today, I ran the same test on other servers, it looks the same as
> above. Those results fluctuate within ~2%.
>=20
> Oh, one more thing I forgot to say is the output of iperf itself which
> doesn't show any difference.
> Before: Bitrate is 211 - 212 Mbits/sec
> After: Bitrate is 211 - 212 Mbits/sec
> So this result is relatively constant especially if we keep running
> the test over 2 minutes.

Thanks for the testing. My personal take on this one is that is more a
refactor than a bug fix - as the amount forward allocated memory should
always be negligible for UDP.=20

Still it could make sense keep the accounting schema consistent across
different protocols. I suggest to repost for net-next, when it will re-
open, additionally introducing __sk_mem_schedule() usage to avoid code
duplication.

Thanks,

Paolo