Skip to contentSkip to conversation

tcp: do not rearm RTO when future data are sacked

PATCH In review Yuchung Cheng 4 participants 4 messages Oct 24, 2013 – Oct 27, 2013

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 tcp: do not rearm RTO when future data are sacked

Patch ed08495c3 "tcp: use RTT from SACK for RTO" always re-arms RTO upon obtaining a RTT sample from newly sacked data.

But technically RTO should only be re-armed when the data sent before the last (re)transmission of write queue head are (s)acked. Otherwise the RTO may continue to extend during loss recovery on data sent in the future.

Note that RTTs from ACK or timestamps do not have this problem, as the RTT source must be from data sent before.

The new RTO re-arm policy is
1) Always re-arm RTO if SND.UNA is advanced
2) Re-arm RTO if sack RTT is available, provided the sacked data was sent before the last time write_queue_head was sent.

---
 net/ipv4/tcp_input.c | 13 ++++++++++---
 1 file changed, 10 insertions(+), 3 deletions(-)
net/ipv4/tcp_input.c +10 -3
index 6ffe41a..068c8fb 100644
@@ -2987,6 +2987,7 @@ static int tcp_clean_rtx_queue(struct sock *sk, int prior_fackets,
 	s32 seq_rtt = -1;
 	s32 ca_seq_rtt = -1;
 	ktime_t last_ackt = net_invalid_timestamp();
+	bool rtt_update;
 
 	while ((skb = tcp_write_queue_head(sk)) && skb != tcp_send_head(sk)) {
 		struct tcp_skb_cb *scb = TCP_SKB_CB(skb);
@@ -3063,14 +3064,13 @@ static int tcp_clean_rtx_queue(struct sock *sk, int prior_fackets,
 	if (skb && (TCP_SKB_CB(skb)->sacked & TCPCB_SACKED_ACKED))
 		flag |= FLAG_SACK_RENEGING;
 
-	if (tcp_ack_update_rtt(sk, flag, seq_rtt, sack_rtt) ||
-	    (flag & FLAG_ACKED))
-		tcp_rearm_rto(sk);
+	rtt_update = tcp_ack_update_rtt(sk, flag, seq_rtt, sack_rtt);
 
 	if (flag & FLAG_ACKED) {
 		const struct tcp_congestion_ops *ca_ops
 			= inet_csk(sk)->icsk_ca_ops;
 
+		tcp_rearm_rto(sk);
 		if (unlikely(icsk->icsk_mtup.probe_size &&
 			     !after(tp->mtu_probe.probe_seq_end, tp->snd_una))) {
 			tcp_mtup_probe_success(sk);
@@ -3109,6 +3109,13 @@ static int tcp_clean_rtx_queue(struct sock *sk, int prior_fackets,
 
 			ca_ops->pkts_acked(sk, pkts_acked, rtt_us);
 		}
+	} else if (skb && rtt_update && sack_rtt >= 0 &&
+		   sack_rtt > (s32)(now - TCP_SKB_CB(skb)->when)) {
+		/* Do not re-arm RTO if the sack RTT is measured from data sent
+		 * after when the head was last (re)transmitted. Otherwise the
+		 * timeout may continue to extend in loss recovery.
+		 */
+		tcp_rearm_rto(sk);
 	}
 
 #if FASTRETRANS_DEBUG > 0
Signature · 1 line
1.8.4.1
maintainer Acked-by on v1 1/1
#1 lore raw
Quoted 16 lines · Yuchung Cheng

Patch ed08495c3 "tcp: use RTT from SACK for RTO" always re-arms RTO upon obtaining a RTT sample from newly sacked data.

But technically RTO should only be re-armed when the data sent before the last (re)transmission of write queue head are (s)acked. Otherwise the RTO may continue to extend during loss recovery on data sent in the future.

Note that RTTs from ACK or timestamps do not have this problem, as the RTT source must be from data sent before.

The new RTO re-arm policy is
1) Always re-arm RTO if SND.UNA is advanced
2) Re-arm RTO if sack RTT is available, provided the sacked data was sent before the last time write_queue_head was sent.

Quoted 5 lines
---
 net/ipv4/tcp_input.c | 13 ++++++++++---
 1 file changed, 10 insertions(+), 3 deletions(-)

neal

Acked-by on v1 1/1
#2 lore raw
In reply to Yuchung Cheng
Quoted 16 lines · Yuchung Cheng

Patch ed08495c3 "tcp: use RTT from SACK for RTO" always re-arms RTO upon obtaining a RTT sample from newly sacked data.

But technically RTO should only be re-armed when the data sent before the last (re)transmission of write queue head are (s)acked. Otherwise the RTO may continue to extend during loss recovery on data sent in the future.

Note that RTTs from ACK or timestamps do not have this problem, as the RTT source must be from data sent before.

The new RTO re-arm policy is
1) Always re-arm RTO if SND.UNA is advanced
2) Re-arm RTO if sack RTT is available, provided the sacked data was sent before the last time write_queue_head was sent.

Quoted 5 lines
---
 net/ipv4/tcp_input.c | 13 ++++++++++---
 1 file changed, 10 insertions(+), 3 deletions(-)
maintainer
#3 lore raw
In reply to Yuchung Cheng

From: Yuchung Cheng <[email protected]>
Date: Thu, 24 Oct 2013 08:59:27 -0700

Quoted 11 lines

Patch ed08495c3 "tcp: use RTT from SACK for RTO" always re-arms RTO upon obtaining a RTT sample from newly sacked data.

But technically RTO should only be re-armed when the data sent before the last (re)transmission of write queue head are (s)acked. Otherwise the RTO may continue to extend during loss recovery on data sent in the future.

Note that RTTs from ACK or timestamps do not have this problem, as the RTT source must be from data sent before.

Quoted 7 lines

The new RTO re-arm policy is
1) Always re-arm RTO if SND.UNA is advanced
2) Re-arm RTO if sack RTT is available, provided the sacked data was sent before the last time write_queue_head was sent.

Also applied and queued up for -stable, thanks!