tcp: Tail loss probe (TLP)

This patch series implement the Tail loss probe (TLP) algorithm described
in http://tools.ietf.org/html/draft-dukkipati-tcpm-tcp-loss-probe-01. The
first patch implements the basic algorithm.

TLP's goal is to reduce tail latency of short transactions. It achieves
this by converting retransmission timeouts (RTOs) occuring due
to tail losses (losses at end of transactions) into fast recovery.
TLP transmits one packet in two round-trips when a connection is in
Open state and isn't receiving any ACKs. The transmitted packet, aka
loss probe, can be either new or a retransmission. When there is tail
loss, the ACK from a loss probe triggers FACK/early-retransmit based
fast recovery, thus avoiding a costly RTO. In the absence of loss,
there is no change in the connection state.

PTO stands for probe timeout. It is a timer event indicating
that an ACK is overdue and triggers a loss probe packet. The PTO value
is set to max(2*SRTT, 10ms) and is adjusted to account for delayed
ACK timer when there is only one oustanding packet.

TLP Algorithm

On transmission of new data in Open state:
  -> packets_out > 1: schedule PTO in max(2*SRTT, 10ms).
  -> packets_out == 1: schedule PTO in max(2*RTT, 1.5*RTT + 200ms)
  -> PTO = min(PTO, RTO)

Conditions for scheduling PTO:
  -> Connection is in Open state.
  -> Connection is either cwnd limited or no new data to send.
  -> Number of probes per tail loss episode is limited to one.
  -> Connection is SACK enabled.

When PTO fires:
  new_segment_exists:
    -> transmit new segment.
    -> packets_out++. cwnd remains same.

  no_new_packet:
    -> retransmit the last segment.
       Its ACK triggers FACK or early retransmit based recovery.

ACK path:
  -> rearm RTO at start of ACK processing.
  -> reschedule PTO if need be.

In addition, the patch includes a small variation to the Early Retransmit
(ER) algorithm, such that ER and TLP together can in principle recover any
N-degree of tail loss through fast recovery. TLP is controlled by the same
sysctl as ER, tcp_early_retrans sysctl.
tcp_early_retrans==0; disables TLP and ER.
		 ==1; enables RFC5827 ER.
		 ==2; delayed ER.
		 ==3; TLP and delayed ER. [DEFAULT]
		 ==4; TLP only.

The TLP patch series have been extensively tested on Google Web servers.
It is most effective for short Web trasactions, where it reduced RTOs by 15%
and improved HTTP response time (average by 6%, 99th percentile by 10%).
The transmitted probes account for <0.5% of the overall transmissions.

Signed-off-by: Nandita Dukkipati <nanditad@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Acked-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

1 files changed, 124 insertions, 4 deletions

diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
index e2b4461074da..beb63dbc85f5 100644
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -74,6 +74,7 @@ static bool tcp_write_xmit(struct sock *sk, unsigned int mss_now, int nonagle,
 /* Account for new data that has been sent to the network. */
 static void tcp_event_new_data_sent(struct sock *sk, const struct sk_buff *skb)
 {
+	struct inet_connection_sock *icsk = inet_csk(sk);
 	struct tcp_sock *tp = tcp_sk(sk);
 	unsigned int prior_packets = tp->packets_out;
 
@@ -85,7 +86,8 @@ static void tcp_event_new_data_sent(struct sock *sk, const struct sk_buff *skb)
 		tp->frto_counter = 3;
 
 	tp->packets_out += tcp_skb_pcount(skb);
-	if (!prior_packets || tp->early_retrans_delayed)
+	if (!prior_packets || icsk->icsk_pending == ICSK_TIME_EARLY_RETRANS ||
+	    icsk->icsk_pending == ICSK_TIME_LOSS_PROBE)
 		tcp_rearm_rto(sk);
 }
 
@@ -1959,6 +1961,9 @@ static int tcp_mtu_probe(struct sock *sk)
  * snd_up-64k-mss .. snd_up cannot be large. However, taking into
  * account rare use of URG, this is not a big flaw.
  *
+ * Send at most one packet when push_one > 0. Temporarily ignore
+ * cwnd limit to force at most one packet out when push_one == 2.
+
  * Returns true, if no segments are in flight and we have queued segments,
  * but cannot send anything now because of SWS or another problem.
  */
@@ -1994,8 +1999,13 @@ static bool tcp_write_xmit(struct sock *sk, unsigned int mss_now, int nonagle,
 			goto repair; /* Skip network transmission */
 
 		cwnd_quota = tcp_cwnd_test(tp, skb);
-		if (!cwnd_quota)
-			break;
+		if (!cwnd_quota) {
+			if (push_one == 2)
+				/* Force out a loss probe pkt. */
+				cwnd_quota = 1;
+			else
+				break;
+		}
 
 		if (unlikely(!tcp_snd_wnd_test(tp, skb, mss_now)))
 			break;
@@ -2049,10 +2059,120 @@ repair:
 	if (likely(sent_pkts)) {
 		if (tcp_in_cwnd_reduction(sk))
 			tp->prr_out += sent_pkts;
+
+		/* Send one loss probe per tail loss episode. */
+		if (push_one != 2)
+			tcp_schedule_loss_probe(sk);
 		tcp_cwnd_validate(sk);
 		return false;
 	}
-	return !tp->packets_out && tcp_send_head(sk);
+	return (push_one == 2) || (!tp->packets_out && tcp_send_head(sk));
+}
+
+bool tcp_schedule_loss_probe(struct sock *sk)
+{
+	struct inet_connection_sock *icsk = inet_csk(sk);
+	struct tcp_sock *tp = tcp_sk(sk);
+	u32 timeout, tlp_time_stamp, rto_time_stamp;
+	u32 rtt = tp->srtt >> 3;
+
+	if (WARN_ON(icsk->icsk_pending == ICSK_TIME_EARLY_RETRANS))
+		return false;
+	/* No consecutive loss probes. */
+	if (WARN_ON(icsk->icsk_pending == ICSK_TIME_LOSS_PROBE)) {
+		tcp_rearm_rto(sk);
+		return false;
+	}
+	/* Don't do any loss probe on a Fast Open connection before 3WHS
+	 * finishes.
+	 */
+	if (sk->sk_state == TCP_SYN_RECV)
+		return false;
+
+	/* TLP is only scheduled when next timer event is RTO. */
+	if (icsk->icsk_pending != ICSK_TIME_RETRANS)
+		return false;
+
+	/* Schedule a loss probe in 2*RTT for SACK capable connections
+	 * in Open state, that are either limited by cwnd or application.
+	 */
+	if (sysctl_tcp_early_retrans < 3 || !rtt || !tp->packets_out ||
+	    !tcp_is_sack(tp) || inet_csk(sk)->icsk_ca_state != TCP_CA_Open)
+		return false;
+
+	if ((tp->snd_cwnd > tcp_packets_in_flight(tp)) &&
+	     tcp_send_head(sk))
+		return false;
+
+	/* Probe timeout is at least 1.5*rtt + TCP_DELACK_MAX to account
+	 * for delayed ack when there's one outstanding packet.
+	 */
+	timeout = rtt << 1;
+	if (tp->packets_out == 1)
+		timeout = max_t(u32, timeout,
+				(rtt + (rtt >> 1) + TCP_DELACK_MAX));
+	timeout = max_t(u32, timeout, msecs_to_jiffies(10));
+
+	/* If RTO is shorter, just schedule TLP in its place. */
+	tlp_time_stamp = tcp_time_stamp + timeout;
+	rto_time_stamp = (u32)inet_csk(sk)->icsk_timeout;
+	if ((s32)(tlp_time_stamp - rto_time_stamp) > 0) {
+		s32 delta = rto_time_stamp - tcp_time_stamp;
+		if (delta > 0)
+			timeout = delta;
+	}
+
+	inet_csk_reset_xmit_timer(sk, ICSK_TIME_LOSS_PROBE, timeout,
+				  TCP_RTO_MAX);
+	return true;
+}
+
+/* When probe timeout (PTO) fires, send a new segment if one exists, else
+ * retransmit the last segment.
+ */
+void tcp_send_loss_probe(struct sock *sk)
+{
+	struct sk_buff *skb;
+	int pcount;
+	int mss = tcp_current_mss(sk);
+	int err = -1;
+
+	if (tcp_send_head(sk) != NULL) {
+		err = tcp_write_xmit(sk, mss, TCP_NAGLE_OFF, 2, GFP_ATOMIC);
+		goto rearm_timer;
+	}
+
+	/* Retransmit last segment. */
+	skb = tcp_write_queue_tail(sk);
+	if (WARN_ON(!skb))
+		goto rearm_timer;
+
+	pcount = tcp_skb_pcount(skb);
+	if (WARN_ON(!pcount))
+		goto rearm_timer;
+
+	if ((pcount > 1) && (skb->len > (pcount - 1) * mss)) {
+		if (unlikely(tcp_fragment(sk, skb, (pcount - 1) * mss, mss)))
+			goto rearm_timer;
+		skb = tcp_write_queue_tail(sk);
+	}
+
+	if (WARN_ON(!skb || !tcp_skb_pcount(skb)))
+		goto rearm_timer;
+
+	/* Probe with zero data doesn't trigger fast recovery. */
+	if (skb->len > 0)
+		err = __tcp_retransmit_skb(sk, skb);
+
+rearm_timer:
+	inet_csk_reset_xmit_timer(sk, ICSK_TIME_RETRANS,
+				  inet_csk(sk)->icsk_rto,
+				  TCP_RTO_MAX);
+
+	if (likely(!err))
+		NET_INC_STATS_BH(sock_net(sk),
+				 LINUX_MIB_TCPLOSSPROBES);
+	return;
 }
 
 /* Push out any pending frames which were held back due to