From 2b8ee332b9384596cefc91fead1c294fdd9252b4 Mon Sep 17 00:00:00 2001
From: Richard Scheffenegger <rscheff@FreeBSD.org>
Date: Wed, 31 Aug 2022 14:49:25 +0200
Subject: [PATCH] tcp: finish SACK loss recovery on sudden lack of SACK blocks

While a receiver should continue sending SACK blocks for the
duration of a SACK loss recovery, if for some reason the
TCP options no longer contain these SACK blocks, but we
already started maintaining the Scoreboard, keep on handling
incoming ACKs (without SACK) as belonging to the SACK recovery.

Reported by:		thj
Reviewed by:		tuexen, #transport
MFC after:		2 weeks
Sponsored by:		NetApp, Inc.
Differential Revision:	https://reviews.freebsd.org/D36046

(cherry picked from commit c21b7b55bea2cc2bf3b420c70a9018e703ed6f00)
---
 sys/netinet/tcp_input.c | 39 ++++++++++++++++++++++-----------------
 1 file changed, 22 insertions(+), 17 deletions(-)

diff --git a/sys/netinet/tcp_input.c b/sys/netinet/tcp_input.c
index 601b4a63716..b2a2fc878b8 100644
--- a/sys/netinet/tcp_input.c
+++ b/sys/netinet/tcp_input.c
@@ -287,6 +287,20 @@ kmod_tcpstat_add(int statnum, int val)
 	counter_u64_add(VNET(tcpstat)[statnum], val);
 }
 
+/*
+ * Make sure that we only start a SACK loss recovery when
+ * receiving a duplicate ACK with a SACK block, and also
+ * complete SACK loss recovery in case the other end
+ * reneges.
+ */
+static bool inline
+tcp_is_sack_recovery(struct tcpcb *tp, struct tcpopt *to)
+{
+	return ((tp->t_flags & TF_SACK_PERMIT) &&
+		((to->to_flags & TOF_SACK) ||
+		(!TAILQ_EMPTY(&tp->snd_holes))));
+}
+
 #ifdef TCP_HHOOK
 /*
  * Wrapper for the TCP established input helper hook.
@@ -2525,9 +2539,7 @@ tcp_do_segment(struct mbuf *m, struct tcphdr *th, struct socket *so,
 			TCPSTAT_INC(tcps_rcvacktoomuch);
 			goto dropafterack;
 		}
-		if ((tp->t_flags & TF_SACK_PERMIT) &&
-		    ((to.to_flags & TOF_SACK) ||
-		     !TAILQ_EMPTY(&tp->snd_holes)))
+		if (tcp_is_sack_recovery(tp, &to))
 			sack_changed = tcp_sack_doack(tp, &to, th->th_ack);
 		else
 			/*
@@ -2597,8 +2609,7 @@ tcp_do_segment(struct mbuf *m, struct tcphdr *th, struct socket *so,
 				 * duplicating packets or a possible DoS attack.
 				 */
 				if (th->th_ack != tp->snd_una ||
-				    ((tp->t_flags & TF_SACK_PERMIT) &&
-				    (to.to_flags & TOF_SACK) &&
+				    (tcp_is_sack_recovery(tp, &to) &&
 				    !sack_changed))
 					break;
 				else if (!tcp_timer_active(tp, TT_REXMT))
@@ -2610,8 +2621,7 @@ tcp_do_segment(struct mbuf *m, struct tcphdr *th, struct socket *so,
 					if (V_tcp_do_prr &&
 					    IN_FASTRECOVERY(tp->t_flags)) {
 						tcp_do_prr_ack(tp, th, &to);
-					} else if ((tp->t_flags & TF_SACK_PERMIT) &&
-					    (to.to_flags & TOF_SACK) &&
+					} else if (tcp_is_sack_recovery(tp, &to) &&
 					    IN_FASTRECOVERY(tp->t_flags)) {
 						int awnd;
 
@@ -2684,8 +2694,7 @@ tcp_do_segment(struct mbuf *m, struct tcphdr *th, struct socket *so,
 						 * snd_ssthresh is already updated by
 						 * cc_cong_signal.
 						 */
-						if ((tp->t_flags & TF_SACK_PERMIT) &&
-						    (to.to_flags & TOF_SACK)) {
+						if (tcp_is_sack_recovery(tp, &to)) {
 							tp->sackhint.prr_delivered =
 							    tp->sackhint.sacked_bytes;
 						} else {
@@ -2697,8 +2706,7 @@ tcp_do_segment(struct mbuf *m, struct tcphdr *th, struct socket *so,
 						tp->sackhint.recover_fs = max(1,
 						    tp->snd_nxt - tp->snd_una);
 					}
-					if ((tp->t_flags & TF_SACK_PERMIT) &&
-					    (to.to_flags & TOF_SACK)) {
+					if (tcp_is_sack_recovery(tp, &to)) {
 						TCPSTAT_INC(
 						    tcps_sack_recovery_episode);
 						tp->snd_recover = tp->snd_nxt;
@@ -2790,8 +2798,7 @@ tcp_do_segment(struct mbuf *m, struct tcphdr *th, struct socket *so,
 			 * from the left side. Such partial ACKs should not be
 			 * counted as dupacks here.
 			 */
-			if ((tp->t_flags & TF_SACK_PERMIT) &&
-			    (to.to_flags & TOF_SACK) &&
+			if (tcp_is_sack_recovery(tp, &to) &&
 			    sack_changed) {
 				tp->t_dupacks++;
 				/* limit overhead by setting maxseg last */
@@ -3973,8 +3980,7 @@ tcp_do_prr_ack(struct tcpcb *tp, struct tcphdr *th, struct tcpopt *to)
 	 * (del_data) and an estimate of how many bytes are in the
 	 * network.
 	 */
-	if (((tp->t_flags & TF_SACK_PERMIT) &&
-	    (to->to_flags & TOF_SACK)) ||
+	if (tcp_is_sack_recovery(tp, to) ||
 	    (IN_CONGRECOVERY(tp->t_flags) &&
 	     !IN_FASTRECOVERY(tp->t_flags))) {
 		del_data = tp->sackhint.delivered_data;
@@ -4018,8 +4024,7 @@ tcp_do_prr_ack(struct tcpcb *tp, struct tcphdr *th, struct tcpopt *to)
 	 * accordingly.
 	 */
 	if (IN_FASTRECOVERY(tp->t_flags)) {
-		if ((tp->t_flags & TF_SACK_PERMIT) &&
-		    (to->to_flags & TOF_SACK)) {
+		if (tcp_is_sack_recovery(tp, to)) {
 			tp->snd_cwnd = tp->snd_nxt - tp->snd_recover +
 					    tp->sackhint.sack_bytes_rexmit +
 					    (snd_cnt * maxseg);
-- 
2.45.0