Re: [PATCH net] tcp: md5: fix IPv4-mapped support

[Eric Dumazet <edumazet@google.com>]

On Tue, Jul 26, 2022 at 5:06 PM David Ahern <dsahern@kernel.org> wrote:
>
> On 7/26/22 5:57 AM, Eric Dumazet wrote:
> > After the blamed commit, IPv4 SYN packets handled
> > by a dual stack IPv6 socket are dropped, even if
> > perfectly valid.
> >
> > $ nstat | grep MD5
> > TcpExtTCPMD5Failure             5                  0.0
> >
> > For a dual stack listener, an incoming IPv4 SYN packet
> > would call tcp_inbound_md5_hash() with @family == AF_INET,
> > while tp->af_specific is pointing to tcp_sock_ipv6_specific.
> >
> > Only later when an IPv4-mapped child is created, tp->af_specific
> > is changed to tcp_sock_ipv6_mapped_specific.
> >
> > Fixes: 7bbb765b7349 ("net/tcp: Merge TCP-MD5 inbound callbacks")
> > Reported-by: Brian Vazquez <brianvv@google.com>
> > Signed-off-by: Eric Dumazet <edumazet@google.com>
> > Cc: Dmitry Safonov <dima@arista.com>
> > Cc: David Ahern <dsahern@kernel.org>
> > Cc: Leonard Crestez <cdleonard@gmail.com>
> > ---
> >  net/ipv4/tcp.c | 15 ++++++++++++---
> >  1 file changed, 12 insertions(+), 3 deletions(-)
> >
> > diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
> > index 002a4a04efbe076ba603d7d42eb85e60d9bf4fb8..766881775abb795c884d048d51c361e805b91989 100644
> > --- a/net/ipv4/tcp.c
> > +++ b/net/ipv4/tcp.c
> > @@ -4459,9 +4459,18 @@ tcp_inbound_md5_hash(const struct sock *sk, const struct sk_buff *skb,
> >               return SKB_DROP_REASON_TCP_MD5UNEXPECTED;
> >       }
> >
> > -     /* check the signature */
> > -     genhash = tp->af_specific->calc_md5_hash(newhash, hash_expected,
> > -                                              NULL, skb);
> > +     /* Check the signature.
> > +      * To support dual stack listeners, we need to handle
> > +      * IPv4-mapped case.
> > +      */
> > +     if (family == AF_INET)
> > +             genhash = tcp_v4_md5_hash_skb(newhash,
> > +                                           hash_expected,
> > +                                           NULL, skb);
> > +     else
> > +             genhash = tp->af_specific->calc_md5_hash(newhash,
> > +                                                      hash_expected,
> > +                                                      NULL, skb);
> >
> >       if (genhash || memcmp(hash_location, newhash, 16) != 0) {
> >               NET_INC_STATS(sock_net(sk), LINUX_MIB_TCPMD5FAILURE);
>
> We should get v4-mapped cases added to the fcnal-test.sh permutations.
>
> Reviewed-by: David Ahern <dsahern@kernel.org>

Although fcnal-test.sh uses ~45 minutes currently :/
Maybe we should make it multi netns and multi threaded to speed up things.

And/or replace various "sleep 1" with more appropriate sync to make
this faster and not flaky in case of system load.