Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
MIME-Version: 1.0
References: <1577699681-14748-1-git-send-email-yangpc@wangsu.com>
In-Reply-To: <1577699681-14748-1-git-send-email-yangpc@wangsu.com>
From:   Eric Dumazet <edumazet@google.com>
Date:   Mon, 30 Dec 2019 05:40:49 -0800
Message-ID: <CANn89iLq6Tt5TZj9SXxK02y=3f35kRSsR2zGBL=1QmFmJBpvXQ@mail.gmail.com>
Subject: Re: [PATCH] tcp: fix "old stuff" D-SACK causing SACK to be treated as D-SACK
To:     Pengcheng Yang <yangpc@wangsu.com>
Cc:     David Miller <davem@davemloft.net>,
        Alexey Kuznetsov <kuznet@ms2.inr.ac.ru>,
        Hideaki YOSHIFUJI <yoshfuji@linux-ipv6.org>,
        Alexei Starovoitov <ast@kernel.org>,
        Daniel Borkmann <daniel@iogearbox.net>,
        Martin KaFai Lau <kafai@fb.com>,
        Song Liu <songliubraving@fb.com>, Yonghong Song <yhs@fb.com>,
        andriin@fb.com, netdev <netdev@vger.kernel.org>,
        LKML <linux-kernel@vger.kernel.org>
Content-Type: text/plain; charset="UTF-8"
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Mon, Dec 30, 2019 at 1:55 AM Pengcheng Yang <yangpc@wangsu.com> wrote:
>
> When we receive a D-SACK, where the sequence number satisfies:
>         undo_marker <= start_seq < end_seq <= prior_snd_una
> we consider this is a valid D-SACK and tcp_is_sackblock_valid()
> returns true, then this D-SACK is discarded as "old stuff",
> but the variable first_sack_index is not marked as negative
> in tcp_sacktag_write_queue().
>
> If this D-SACK also carries a SACK that needs to be processed
> (for example, the previous SACK segment was lost),

What do you mean by ' previous sack segment was lost'  ?

 this SACK
> will be treated as a D-SACK in the following processing of
> tcp_sacktag_write_queue(), which will eventually lead to
> incorrect updates of undo_retrans and reordering.
>
> Fixes: fd6dad616d4f ("[TCP]: Earlier SACK block verification & simplify access to them")
> Signed-off-by: Pengcheng Yang <yangpc@wangsu.com>
> ---
>  net/ipv4/tcp_input.c | 5 ++++-
>  1 file changed, 4 insertions(+), 1 deletion(-)
>
> diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
> index 88b987c..0238b55 100644
> --- a/net/ipv4/tcp_input.c
> +++ b/net/ipv4/tcp_input.c
> @@ -1727,8 +1727,11 @@ static int tcp_sack_cache_ok(const struct tcp_sock *tp, const struct tcp_sack_bl
>                 }
>
>                 /* Ignore very old stuff early */
> -               if (!after(sp[used_sacks].end_seq, prior_snd_una))
> +               if (!after(sp[used_sacks].end_seq, prior_snd_una)) {
> +                       if (i == 0)
> +                               first_sack_index = -1;
>                         continue;
> +               }
>
>                 used_sacks++;
>         }


Hi Pengcheng Yang

This corner case deserves a packetdrill test so that we understand the
issue, can you provide one ?

Thanks.