Message-ID: <8a2ff1bd-52dc-421d-87b7-fc2f56e81da2@kernel.org>
Date: Wed, 14 Aug 2024 12:04:31 +0200
From: Matthieu Baerts <matttbe@...nel.org>
To: Martin KaFai Lau <martin.lau@...ux.dev>
Cc: mptcp@...ts.linux.dev, Mat Martineau <martineau@...nel.org>,
 Geliang Tang <geliang@...nel.org>, Andrii Nakryiko <andrii@...nel.org>,
 Eduard Zingerman <eddyz87@...il.com>, Mykola Lysenko <mykolal@...com>,
 Alexei Starovoitov <ast@...nel.org>, Daniel Borkmann <daniel@...earbox.net>,
 Song Liu <song@...nel.org>, Yonghong Song <yonghong.song@...ux.dev>,
 John Fastabend <john.fastabend@...il.com>, KP Singh <kpsingh@...nel.org>,
 Stanislav Fomichev <sdf@...ichev.me>, Hao Luo <haoluo@...gle.com>,
 Jiri Olsa <jolsa@...nel.org>, Shuah Khan <shuah@...nel.org>,
 linux-kernel@...r.kernel.org, netdev@...r.kernel.org, bpf@...r.kernel.org,
 linux-kselftest@...r.kernel.org, Daniel Xu <dxu@...uu.xyz>,
 Manu Bretelle <chantra@...a.com>
Subject: Re: [PATCH bpf-next v4 2/2] selftests/bpf: Add mptcp subflow subtest

Hi Martin,

Thank you for your reply!

On 14/08/2024 03:12, Martin KaFai Lau wrote:
> On 8/5/24 2:52 AM, Matthieu Baerts (NGI0) wrote:
>> +static int endpoint_init(char *flags)
>> +{
>> +    SYS(fail, "ip -net %s link add veth1 type veth peer name veth2",
>> NS_TEST);
>> +    SYS(fail, "ip -net %s addr add %s/24 dev veth1", NS_TEST, ADDR_1);
>> +    SYS(fail, "ip -net %s link set dev veth1 up", NS_TEST);
>> +    SYS(fail, "ip -net %s addr add %s/24 dev veth2", NS_TEST, ADDR_2);
>> +    SYS(fail, "ip -net %s link set dev veth2 up", NS_TEST);
>> +    if (SYS_NOFAIL("ip -net %s mptcp endpoint add %s %s", NS_TEST,
>> ADDR_2, flags)) {
>> +        printf("'ip mptcp' not supported, skip this test.\n");
>> +        test__skip();
> 
> It is always a skip now in bpf CI:
> 
> #171/3   mptcp/subflow:SKIP
> 
> This test is a useful addition for the bpf CI selftest.
> 
> It can't catch regression if it is always a skip in bpf CI though.

Indeed, for the moment, this test is skipped in bpf CI.

The MPTCP CI checks the MPTCP BPF selftests that are on top of net and
net-next at least once a day. It is always running with the last stable
version of iproute2, so this test is not skipped:

   #169/3   mptcp/subflow:OK

https://github.com/multipath-tcp/mptcp_net-next/actions/runs/10384566794/job/28751869426#step:7:11080

> iproute2 needs to be updated (cc: Daniel Xu and Manu, the outdated
> iproute2 is something that came up multiple times).
> 
> Not sure when the iproute2 can be updated. In the mean time, your v3 is
> pretty close to getting pm_nl_ctl compiled. Is there other blocker on this?

I will try to find some time to check the modifications I suggested in
the v3, but I don't know how long it will take to have them ready, as
they might require some adaptations of the CI side as well, I need to
check. On the other hand, I understood adding a duplicated version of
the mptcp.h UAPI header is not an option either.

So not to block this (already old) series, I thought it would help to
first focus on this version using 'ip mptcp', while I'm looking at the
selftests modifications. If these modifications are successful, I can
always resend the patch 2/3 from the v3 later, and using 'pm_nl_ctl'
instead of 'ip mptcp', to be able to work with IPRoute2 5.5.

Do you think that could work like that?

>> +        goto fail;
>> +    }
>> +
>> +    return 0;
>> +fail:
>> +    return -1;
>> +}
>> +
>> +static int _ss_search(char *src, char *dst, char *port, char *keyword)
>> +{
>> +    return SYS_NOFAIL("ip netns exec %s ss -enita src %s dst %s %s %d
>> | grep -q '%s'",
>> +              NS_TEST, src, dst, port, PORT_1, keyword);
>> +}
>> +
>> +static int ss_search(char *src, char *keyword)
>> +{
>> +    return _ss_search(src, ADDR_1, "dport", keyword);
>> +}
>> +
>> +static void run_subflow(char *new)
>> +{
>> +    int server_fd, client_fd, err;
>> +    char cc[TCP_CA_NAME_MAX];
>> +    socklen_t len = sizeof(cc);
>> +
>> +    server_fd = start_mptcp_server(AF_INET, ADDR_1, PORT_1, 0);
>> +    if (!ASSERT_GE(server_fd, 0, "start_mptcp_server"))
>> +        return;
>> +
>> +    client_fd = connect_to_fd(server_fd, 0);
>> +    if (!ASSERT_GE(client_fd, 0, "connect to fd"))
>> +        goto fail;
>> +
>> +    err = getsockopt(server_fd, SOL_TCP, TCP_CONGESTION, cc, &len);
>> +    if (!ASSERT_OK(err, "getsockopt(srv_fd, TCP_CONGESTION)"))
>> +        goto fail;
>> +
>> +    send_byte(client_fd);
>> +
>> +    ASSERT_OK(ss_search(ADDR_1, "fwmark:0x1"), "ss_search fwmark:0x1");
>> +    ASSERT_OK(ss_search(ADDR_2, "fwmark:0x2"), "ss_search fwmark:0x2");
>> +    ASSERT_OK(ss_search(ADDR_1, new), "ss_search new cc");
>> +    ASSERT_OK(ss_search(ADDR_2, cc), "ss_search default cc");
> 
> Is there a getsockopt way instead of ss + grep?

No there isn't: from the userspace, the app communicates with the MPTCP
socket, which can have multiple paths (subflows, a TCP socket). To keep
the compatibility with TCP, [gs]etsockopt() will look at/modify the
whole MPTCP connection. For example, in some cases, a setsockopt() will
propagate the option to all the subflows. Depending on the option, the
modification might only apply to the first subflow, or to the
user-facing socket.

For advanced users who want to have different options set to the
different subflows of an MPTCP connection, they can use BPF: that's what
is being validated here. In other words, doing a 'getsockopt()' from the
userspace program here will not show all the different marks and TCP CC
that can be set per subflow with BPF. We can see that in the test: a
getsockopt() is done on the MPTCP socket to retrieve the default TCP CC
('cc' which is certainly 'cubic'), but we expect to find another one
('new' which is 'reno'), set by the BPF program from patch 1/2. I guess
we could use bpf to do a getsockopt() per subflow, but that's seems a
bit cheated to have the BPF test program setting something and checking
if it is set. Here, it is an external way. Because it is done from a
dedicated netns, it sounds OK to do that, no?

Cheers,
Matt
-- 
Sponsored by the NGI0 Core fund.

Powered by blists - more mailing lists