Message-ID: <ede0b4ea92187ca7b6303f3c1c98c26f513a3ce9.camel@redhat.com>
Subject: Re: [PATCH v2 3/4] selftests: net: udpgso_bench: Fix racing bug
 between the rx/tx programs
From:   Paolo Abeni <pabeni@redhat.com>
To:     Willem de Bruijn <willemb@google.com>,
        Andrei Gherzan <andrei.gherzan@canonical.com>
Cc:     "David S. Miller" <davem@davemloft.net>,
        Eric Dumazet <edumazet@google.com>,
        Jakub Kicinski <kuba@kernel.org>,
        Shuah Khan <shuah@kernel.org>, netdev@vger.kernel.org,
        linux-kselftest@vger.kernel.org, linux-kernel@vger.kernel.org
Date:   Tue, 31 Jan 2023 15:45:04 +0100
In-Reply-To: <CA+FuTSdtzFXWWDLk=LOdrkS00oH4HGvtoYYQh7YQd2ADsp0UbA@mail.gmail.com>
References: <20230131130412.432549-1-andrei.gherzan@canonical.com>
         <20230131130412.432549-3-andrei.gherzan@canonical.com>
         <CA+FuTSdtzFXWWDLk=LOdrkS00oH4HGvtoYYQh7YQd2ADsp0UbA@mail.gmail.com>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
User-Agent: Evolution 3.46.3 (3.46.3-1.fc37) 
MIME-Version: 1.0
Precedence: bulk

On Tue, 2023-01-31 at 08:33 -0500, Willem de Bruijn wrote:
> On Tue, Jan 31, 2023 at 8:06 AM Andrei Gherzan
> <andrei.gherzan@canonical.com> wrote:
> >=20
> > "udpgro_bench.sh" invokes udpgso_bench_rx/udpgso_bench_tx programs
> > subsequently and while doing so, there is a chance that the rx one is n=
ot
> > ready to accept socket connections. This racing bug could fail the test
> > with at least one of the following:
> >=20
> > ./udpgso_bench_tx: connect: Connection refused
> > ./udpgso_bench_tx: sendmsg: Connection refused
> > ./udpgso_bench_tx: write: Connection refused
> >=20
> > This change addresses this by making udpgro_bench.sh wait for the rx
> > program to be ready before firing off the tx one - with an exponential =
back
> > off algorithm from 1s to 10s.
> >=20
> > Signed-off-by: Andrei Gherzan <andrei.gherzan@canonical.com>
>=20
> please CC: reviewers of previous revisions on new revisions
>=20
> also for upcoming patches: please clearly mark net or net-next.
> > ---
> >  tools/testing/selftests/net/udpgso_bench.sh | 18 ++++++++++++++++++
> >  1 file changed, 18 insertions(+)
> >=20
> > diff --git a/tools/testing/selftests/net/udpgso_bench.sh b/tools/testin=
g/selftests/net/udpgso_bench.sh
> > index dc932fd65363..20b5db8fcbde 100755
> > --- a/tools/testing/selftests/net/udpgso_bench.sh
> > +++ b/tools/testing/selftests/net/udpgso_bench.sh
> > @@ -7,6 +7,7 @@ readonly GREEN=3D'\033[0;92m'
> >  readonly YELLOW=3D'\033[0;33m'
> >  readonly RED=3D'\033[0;31m'
> >  readonly NC=3D'\033[0m' # No Color
> > +readonly TESTPORT=3D8000 # Keep this in sync with udpgso_bench_rx/tx
>=20
> then also pass explicit -p argument to the processes to keep all three
> consistent
>=20
> >=20
> >  readonly KSFT_PASS=3D0
> >  readonly KSFT_FAIL=3D1
> > @@ -56,10 +57,27 @@ trap wake_children EXIT
> >=20
> >  run_one() {
> >         local -r args=3D$@
> > +       local -r init_delay_s=3D1
> > +       local -r max_delay_s=3D10
> > +       local delay_s=3D0
> > +       local nr_socks=3D0
> >=20
> >         ./udpgso_bench_rx &
> >         ./udpgso_bench_rx -t &
> >=20
> > +       # Wait for the above test program to get ready to receive conne=
ctions.
> > +       delay_s=3D"${init_delay_s}"
> > +       while [ "$delay_s" -lt "$max_delay_s" ]; do
> > +               nr_socks=3D"$(ss -lnHi | grep -c "\*:${TESTPORT}")"
> > +               [ "$nr_socks" -eq 2 ] && break
> > +               sleep "$delay_s"
> > +               delay=3D"$((delay*2))"
>=20
> I don't think we need exponential back-off for something this simple

Agreed. Additionally you could use constant, sub-second delay (say 0.1)
to keep the runtime delta relatively low.

Cheers,

Paolo