On Thu, Dec 10, 2020 at 10:24 PM Eric Dumazet <eduma...@google.com> wrote:
>
> On Fri, Dec 11, 2020 at 2:55 AM Alexander Duyck
> <alexander.du...@gmail.com> wrote:
> >
> > From: Alexander Duyck <alexanderdu...@fb.com>
> >
> > In the case of a fastopen SYN there are cases where it may trigger either a
> > ICMP_TOOBIG message in the case of IPv6 or a fragmentation request in the
> > case of IPv4. This results in the socket stalling for a second or more as
> > it does not respond to the message by retransmitting the SYN frame.
> >
> > Normally a SYN frame should not be able to trigger a ICMP_TOOBIG or
> > ICMP_FRAG_NEEDED however in the case of fastopen we can have a frame that
> > makes use of the entire MTU. In the case of fastopen it does, and an
> > additional complication is that the retransmit queue doesn't contain the
> > original frames. As a result when tcp_simple_retransmit is called and
> > walks the list of frames in the queue it may not mark the frames as lost
> > because both the SYN and the data packet each individually are smaller than
> > the MSS size after the adjustment. This results in the socket being stalled
> > until the retransmit timer kicks in and forces the SYN frame out again
> > without the data attached.
> >
> > In order to resolve this we need to mark the SYN frame as lost if it is the
> > first packet in the queue. Doing this allows the socket to recover much
> > more quickly without the retransmit timeout stall.
> >
> > Signed-off-by: Alexander Duyck <alexanderdu...@fb.com>
>
>
> I do not think it is net candidate, but net-next
>
> Yuchung might correct me, but I think TCP Fastopen standard was very
> conservative about payload len in the SYN packet
>
> So receiving an ICMP was never considered.
That's fine. I can target this for net-next. I had just selected net
since I had considered it a fix, but I suppose it could be considered
a behavioral change.
> > ---
> > include/net/tcp.h | 1 +
> > net/ipv4/tcp_input.c | 8 ++++++++
> > net/ipv4/tcp_ipv4.c | 6 ++++++
> > net/ipv6/tcp_ipv6.c | 4 ++++
> > 4 files changed, 19 insertions(+)
> >
> > diff --git a/include/net/tcp.h b/include/net/tcp.h
> > index d4ef5bf94168..6181ad98727a 100644
> > --- a/include/net/tcp.h
>
>
> > +++ b/net/ipv4/tcp_ipv4.c
> > @@ -546,6 +546,12 @@ int tcp_v4_err(struct sk_buff *skb, u32 info)
> > if (sk->sk_state == TCP_LISTEN)
> > goto out;
> >
> > + /* fastopen SYN may have triggered the fragmentation
> > + * request. Mark the SYN or SYN/ACK as lost.
> > + */
> > + if (sk->sk_state == TCP_SYN_SENT)
> > + tcp_mark_syn_lost(sk);
>
> This is going to crash in some cases, you do not know if you own the socket.
> (Look a few lines below)
Okay, I will look into moving this down into the block below since I
assume if it is owned by user we cannot make these changes.
> > +
> > tp->mtu_info = info;
> > if (!sock_owned_by_user(sk)) {
> > tcp_v4_mtu_reduced(sk);
> > diff --git a/net/ipv6/tcp_ipv6.c b/net/ipv6/tcp_ipv6.c
> > index 992cbf3eb9e3..d7b1346863e3 100644
> > --- a/net/ipv6/tcp_ipv6.c
> > +++ b/net/ipv6/tcp_ipv6.c
> > @@ -443,6 +443,10 @@ static int tcp_v6_err(struct sk_buff *skb, struct
> > inet6_skb_parm *opt,
> > if (!ip6_sk_accept_pmtu(sk))
> > goto out;
> >
> > + /* fastopen SYN may have triggered TOOBIG, mark it lost. */
> > + if (sk->sk_state == TCP_SYN_SENT)
> > + tcp_mark_syn_lost(sk);
>
>
> Same issue here.
I'll move this one too.
> > +
> > tp->mtu_info = ntohl(info);
> > if (!sock_owned_by_user(sk))
> > tcp_v6_mtu_reduced(sk);
> >
> >
