netdev - Re: [PATCH net] tcp: fix SYNACK RTT estimation in Fast Open

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Thu, 24 Oct 2013 13:04:48 -0700
From:	Eric Dumazet <eric.dumazet@...il.com>
To:	Yuchung Cheng <ycheng@...gle.com>
Cc:	davem@...emloft.net, ncardwell@...gle.com, edumazet@...gle.com,
	panweiping3@...il.com, netdev@...r.kernel.org
Subject: Re: [PATCH net] tcp: fix SYNACK RTT estimation in Fast Open

On Thu, 2013-10-24 at 08:44 -0700, Yuchung Cheng wrote:
> tp->lsndtime may not always be the SYNACK timestamp if a passive
> Fast Open socket sends data before handshake completes. And if the
> remote acknowledges both the data and the SYNACK, the RTT sample
> is already taken in tcp_ack(), so no need to call
> tcp_update_ack_rtt() in tcp_synack_rtt_meas() aagain.
> 
> Signed-off-by: Yuchung Cheng <ycheng@...gle.com>
> ---
>  net/ipv4/tcp_input.c | 18 +++++++++++++-----
>  1 file changed, 13 insertions(+), 5 deletions(-)
> 
> diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
> index a16b01b..305cd05 100644
> --- a/net/ipv4/tcp_input.c
> +++ b/net/ipv4/tcp_input.c
> @@ -2871,14 +2871,19 @@ static inline bool tcp_ack_update_rtt(struct sock *sk, const int flag,
>  }
>  
>  /* Compute time elapsed between (last) SYNACK and the ACK completing 3WHS. */
> -static void tcp_synack_rtt_meas(struct sock *sk, struct request_sock *req)
> +static void tcp_synack_rtt_meas(struct sock *sk, const u32 synack_stamp)
>  {
>  	struct tcp_sock *tp = tcp_sk(sk);
>  	s32 seq_rtt = -1;
>  
> -	if (tp->lsndtime && !tp->total_retrans)
> -		seq_rtt = tcp_time_stamp - tp->lsndtime;
> -	tcp_ack_update_rtt(sk, FLAG_SYN_ACKED, seq_rtt, -1);
> +	if (synack_stamp && !tp->total_retrans)
> +		seq_rtt = tcp_time_stamp - synack_stamp;
> +
> +	/* If the ACK acks both the SYNACK and the (Fast Open'd) data packets
> +	 * sent in SYN_RECV, SYNACK RTT is the smooth RTT computed in tcp_ack()
> +	 */
> +	if (!tp->srtt)
> +		tcp_ack_update_rtt(sk, FLAG_SYN_ACKED, seq_rtt, -1);
>  }
>  
>  static void tcp_cong_avoid(struct sock *sk, u32 ack, u32 in_flight)
> @@ -5587,6 +5592,7 @@ int tcp_rcv_state_process(struct sock *sk, struct sk_buff *skb,
>  	struct request_sock *req;
>  	int queued = 0;
>  	bool acceptable;
> +	u32 synack_stamp;
>  
>  	tp->rx_opt.saw_tstamp = 0;
>  
> @@ -5669,9 +5675,11 @@ int tcp_rcv_state_process(struct sock *sk, struct sk_buff *skb,
>  		 * so release it.
>  		 */
>  		if (req) {
> +			synack_stamp = tcp_rsk(req)->snt_synack;
> 
Alternative would have been to set here :

	tp->lsndtime = tcp_rsk(req)->snt_synack;

Because it seems TCP_INFO can return quite bogus data anyway in 

info->tcpi_last_data_sent = jiffies_to_msecs(now - tp->lsndtime);



>  			tp->total_retrans = req->num_retrans;
>  			reqsk_fastopen_remove(sk, req, false);
>  		} else {
> +			synack_stamp = tp->lsndtime;
>  			/* Make sure socket is routed, for correct metrics. */
>  			icsk->icsk_af_ops->rebuild_header(sk);
>  			tcp_init_congestion_control(sk);
> @@ -5694,7 +5702,7 @@ int tcp_rcv_state_process(struct sock *sk, struct sk_buff *skb,
>  		tp->snd_una = TCP_SKB_CB(skb)->ack_seq;
>  		tp->snd_wnd = ntohs(th->window) << tp->rx_opt.snd_wscale;
>  		tcp_init_wl(tp, TCP_SKB_CB(skb)->seq);
> -		tcp_synack_rtt_meas(sk, req);
> +		tcp_synack_rtt_meas(sk, synack_stamp);
>  
>  		if (tp->rx_opt.tstamp_ok)
>  			tp->advmss -= TCPOLEN_TSTAMP_ALIGNED;

Acked-by: Eric Dumazet <edumazet@...gle.com>



--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html