Skip to contentSkip to conversation

tcp: grow window for OOO packets only for SACK flows

PATCH In review Eric Dumazet 3 participants 3 messages Jun 16, 2020

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

PATCH 1/1 tcp: grow window for OOO packets only for SACK flows

Back in 2013, we made a change that broke fast retransmit for non SACK flows.

Indeed, for these flows, a sender needs to receive three duplicate ACK before starting fast retransmit. Sending ACK with different receive window do not count.

Even if enabling SACK is strongly recommended these days, there still are some cases where it has to be disabled.

Not increasing the window seems better than having to rely on RTO.

After the fix, following packetdrill test gives :

// Initialize connection

    0 socket(..., SOCK_STREAM, IPPROTO_TCP) = 3

+0 setsockopt(3, SOL_SOCKET, SO_REUSEADDR, [1], 4) = 0 +0 bind(3, ..., ...) = 0
+0 listen(3, 1) = 0

+0 < S 0:0(0) win 32792 <mss 1000,nop,wscale 7> +0 > S. 0:0(0) ack 1 <mss 1460,nop,wscale 8> +0 < . 1:1(0) ack 1 win 514

+0 accept(3, ..., ...) = 4

+0 < . 1:1001(1000) ack 1 win 514
// Quick ack
+0 > . 1:1(0) ack 1001 win 264

+0 < . 2001:3001(1000) ack 1 win 514
// DUPACK : Normally we should not change the window
+0 > . 1:1(0) ack 1001 win 264

+0 < . 3001:4001(1000) ack 1 win 514
// DUPACK : Normally we should not change the window
+0 > . 1:1(0) ack 1001 win 264

+0 < . 4001:5001(1000) ack 1 win 514
// DUPACK : Normally we should not change the window

    +0 > . 1:1(0) ack 1001 win 264

+0 < . 1001:2001(1000) ack 1 win 514
// Hole is repaired.
+0 > . 1:1(0) ack 5001 win 272

Fixes 4e4f1fc22681 ("tcp: properly increase rcv_ssthresh for ofo packets")
---
 net/ipv4/tcp_input.c | 12 ++++++++++--
 1 file changed, 10 insertions(+), 2 deletions(-)
net/ipv4/tcp_input.c +10 -2
index 83330a6cb24209cf6ac60d634245b7bc151ee6ac..12fda8f27b08bdf5c9f3bad422734f6b1965cef9 100644
@@ -4605,7 +4605,11 @@ static void tcp_data_queue_ofo(struct sock *sk, struct sk_buff *skb)
 	if (tcp_ooo_try_coalesce(sk, tp->ooo_last_skb,
 				 skb, &fragstolen)) {
 coalesce_done:
-		tcp_grow_window(sk, skb);
+		/* For non sack flows, do not grow window to force DUPACK
+		 * and trigger fast retransmit.
+		 */
+		if (tcp_is_sack(tp))
+			tcp_grow_window(sk, skb);
 		kfree_skb_partial(skb, fragstolen);
 		skb = NULL;
 		goto add_sack;
@@ -4689,7 +4693,11 @@ static void tcp_data_queue_ofo(struct sock *sk, struct sk_buff *skb)
 		tcp_sack_new_ofo_skb(sk, seq, end_seq);
 end:
 	if (skb) {
-		tcp_grow_window(sk, skb);
+		/* For non sack flows, do not grow window to force DUPACK
+		 * and trigger fast retransmit.
+		 */
+		if (tcp_is_sack(tp))
+			tcp_grow_window(sk, skb);
 		skb_condense(skb);
 		skb_set_owner_r(skb, sk);
 	}
Signature · 1 line
2.27.0.290.gba653c62da-goog
maintainer Acked-by on v1 1/1
#1 lore raw
Quoted 13 lines · Eric Dumazet

Back in 2013, we made a change that broke fast retransmit for non SACK flows.

Indeed, for these flows, a sender needs to receive three duplicate ACK before starting fast retransmit. Sending ACK with different receive window do not count.

Even if enabling SACK is strongly recommended these days, there still are some cases where it has to be disabled.

Not increasing the window seems better than having to rely on RTO.

...

Quoted 4 lines
Fixes 4e4f1fc22681 ("tcp: properly increase rcv_ssthresh for ofo packets")

---

Thanks for the fix, Eric!

neal

maintainer
#2 lore raw
In reply to Eric Dumazet

From: Eric Dumazet <[email protected]>
Date: Mon, 15 Jun 2020 20:37:07 -0700

Quoted 14 lines

Back in 2013, we made a change that broke fast retransmit for non SACK flows.

Indeed, for these flows, a sender needs to receive three duplicate ACK before starting fast retransmit. Sending ACK with different receive window do not count.

Even if enabling SACK is strongly recommended these days, there still are some cases where it has to be disabled.

Not increasing the window seems better than having to rely on RTO.

After the fix, following packetdrill test gives :

...

Quoted 3 lines
Fixes 4e4f1fc22681 ("tcp: properly increase rcv_ssthresh for ofo packets")

Applied and queued up for -stable, thanks Eric.