[RFC PATCHv2 net-next 1/2] tcp: RTO Restart (RTOR)

[Per Hurtig <per.hurtig@kau.se>]

This patch implements the RTO restart modification (RTOR). When data is
ACKed, and the RTO timer is restarted, the time elapsed since the last
outstanding segment was transmitted is subtracted from the calculated RTO
value. This way, the RTO timer will expire after exactly RTO seconds, and
not RTO + RTT [+ delACK] seconds.

This patch also implements a new sysctl (tcp_timer_restart) that is used
to control the timer restart behavior.

Signed-off-by: Per Hurtig <per.hurtig@kau.se>
---
 Documentation/networking/ip-sysctl.txt | 12 ++++++++++++
 include/net/tcp.h                      |  6 ++++++
 net/ipv4/sysctl_net_ipv4.c             | 10 ++++++++++
 net/ipv4/tcp_input.c                   | 29 +++++++++++++++++++++++++++++
 4 files changed, 57 insertions(+)

diff --git a/Documentation/networking/ip-sysctl.txt b/Documentation/networking/ip-sysctl.txt
index 2ea4c45..4094128 100644
--- a/Documentation/networking/ip-sysctl.txt
+++ b/Documentation/networking/ip-sysctl.txt
@@ -591,6 +591,18 @@ tcp_syn_retries - INTEGER
 	with the current initial RTO of 1second. With this the final timeout
 	for an active TCP connection attempt will happen after 127seconds.
 
+tcp_timer_restart - INTEGER
+	Controls how the RTO and PTO timers are restarted (RTOR and TLPR).
+	If set (per timer or combined) the timers are restarted with
+	respect to the earliest outstanding segment, to not extend tail loss
+	latency unnecessarily.
+	Possible values:
+		0 disables RTOR and TLPR.
+		1 enables RTOR.
+		2 enables TLPR.
+		3 enables RTOR and TLPR.
+	Default: 3
+
 tcp_timestamps - BOOLEAN
 	Enable timestamps as defined in RFC1323.
 
diff --git a/include/net/tcp.h b/include/net/tcp.h
index f80e74c..833efb7 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -76,6 +76,11 @@ void tcp_time_wait(struct sock *sk, int state, int timeo);
 /* After receiving this amount of duplicate ACKs fast retransmit starts. */
 #define TCP_FASTRETRANS_THRESH 3
 
+/* Disable RTO Restart if the number of outstanding segments is at least. */
+#define TCP_TIMER_RTORESTART	1
+#define TCP_TIMER_TLPRESTART	2
+#define TCP_RTORESTART_THRESH	4
+
 /* Maximal number of ACKs sent quickly to accelerate slow-start. */
 #define TCP_MAX_QUICKACKS	16U
 
@@ -284,6 +289,7 @@ extern int sysctl_tcp_autocorking;
 extern int sysctl_tcp_invalid_ratelimit;
 extern int sysctl_tcp_pacing_ss_ratio;
 extern int sysctl_tcp_pacing_ca_ratio;
+extern int sysctl_tcp_timer_restart;
 
 extern atomic_long_t tcp_memory_allocated;
 extern struct percpu_counter tcp_sockets_allocated;
diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c
index a0bd7a5..dfb6968 100644
--- a/net/ipv4/sysctl_net_ipv4.c
+++ b/net/ipv4/sysctl_net_ipv4.c
@@ -28,6 +28,7 @@
 
 static int zero;
 static int one = 1;
+static int three = 3;
 static int four = 4;
 static int thousand = 1000;
 static int gso_max_segs = GSO_MAX_SEGS;
@@ -745,6 +746,15 @@ static struct ctl_table ipv4_table[] = {
 		.extra2		= &thousand,
 	},
 	{
+		.procname	= "tcp_timer_restart",
+		.data		= &sysctl_tcp_timer_restart,
+		.maxlen		= sizeof(int),
+		.mode		= 0644,
+		.proc_handler	= proc_dointvec_minmax,
+		.extra1		= &zero,
+		.extra2		= &three,
+	},
+	{
 		.procname	= "tcp_autocorking",
 		.data		= &sysctl_tcp_autocorking,
 		.maxlen		= sizeof(int),
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 2d656ee..2870af3 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -101,6 +101,8 @@ int sysctl_tcp_thin_dupack __read_mostly;
 
 int sysctl_tcp_moderate_rcvbuf __read_mostly = 1;
 int sysctl_tcp_early_retrans __read_mostly = 3;
+int sysctl_tcp_timer_restart __read_mostly = TCP_TIMER_RTORESTART |
+					     TCP_TIMER_TLPRESTART;
 int sysctl_tcp_invalid_ratelimit __read_mostly = HZ/2;
 
 #define FLAG_DATA		0x01 /* Incoming frame contained data.		*/
@@ -2997,6 +2999,22 @@ static void tcp_cong_avoid(struct sock *sk, u32 ack, u32 acked)
 	tcp_sk(sk)->snd_cwnd_stamp = tcp_time_stamp;
 }
 
+static u32 tcp_unsent_pkts(const struct sock *sk, u32 ulimit)
+{
+	struct sk_buff *skb = tcp_send_head(sk);
+	u32 pkts = 0;
+
+	if (skb)
+		tcp_for_write_queue_from(skb, sk) {
+			pkts += tcp_skb_pcount(skb);
+
+			if (ulimit && pkts >= ulimit)
+				return ulimit;
+		}
+
+	return pkts;
+}
+
 /* Restart timer after forward progress on connection.
  * RFC2988 recommends to restart timer to now+rto.
  */
@@ -3027,6 +3045,17 @@ void tcp_rearm_rto(struct sock *sk)
 			 */
 			if (delta > 0)
 				rto = delta;
+		} else if (icsk->icsk_pending == ICSK_TIME_RETRANS &&
+			   (sysctl_tcp_timer_restart & TCP_TIMER_RTORESTART) &&
+			   (tp->packets_out +
+			    tcp_unsent_pkts(sk, TCP_RTORESTART_THRESH) <
+			    TCP_RTORESTART_THRESH)) {
+			struct sk_buff *skb = tcp_write_queue_head(sk);
+			const u32 rto_time_stamp = tcp_skb_timestamp(skb);
+			s32 delta = (s32)(tcp_time_stamp - rto_time_stamp);
+
+			if (delta > 0 && rto > delta)
+				rto -= delta;
 		}
 		inet_csk_reset_xmit_timer(sk, ICSK_TIME_RETRANS, rto,
 					  TCP_RTO_MAX);
-- 
1.9.1