tcp: retire FACK loss detection

FACK loss detection has been disabled by default and the
successor RACK subsumed FACK and can handle reordering better.
This patch removes FACK to simplify TCP loss recovery.

Signed-off-by: Yuchung Cheng <ycheng@google.com>
Reviewed-by: Eric Dumazet <edumazet@google.com>
Reviewed-by: Neal Cardwell <ncardwell@google.com>
Reviewed-by: Soheil Hassas Yeganeh <soheil@google.com>
Reviewed-by: Priyaranjan Jha <priyarjha@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>

6 files changed, 10 insertions, 60 deletions

diff --git a/net/ipv4/proc.c b/net/ipv4/proc.c
index 127153f1ed8a..9f37c4727861 100644
--- a/net/ipv4/proc.c
+++ b/net/ipv4/proc.c
@@ -212,7 +212,6 @@ static const struct snmp_mib snmp4_net_list[] = {
 	SNMP_MIB_ITEM("TCPRenoRecovery", LINUX_MIB_TCPRENORECOVERY),
 	SNMP_MIB_ITEM("TCPSackRecovery", LINUX_MIB_TCPSACKRECOVERY),
 	SNMP_MIB_ITEM("TCPSACKReneging", LINUX_MIB_TCPSACKRENEGING),
-	SNMP_MIB_ITEM("TCPFACKReorder", LINUX_MIB_TCPFACKREORDER),
 	SNMP_MIB_ITEM("TCPSACKReorder", LINUX_MIB_TCPSACKREORDER),
 	SNMP_MIB_ITEM("TCPRenoReorder", LINUX_MIB_TCPRENOREORDER),
 	SNMP_MIB_ITEM("TCPTSReorder", LINUX_MIB_TCPTSREORDER),
diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
index bc71a27d5ad9..337555076043 100644
--- a/net/ipv4/tcp.c
+++ b/net/ipv4/tcp.c
@@ -2509,8 +2509,6 @@ static int tcp_repair_options_est(struct sock *sk,
 				return -EINVAL;
 
 			tp->rx_opt.sack_ok |= TCP_SACK_SEEN;
-			if (sock_net(sk)->ipv4.sysctl_tcp_fack)
-				tcp_enable_fack(tp);
 			break;
 		case TCPOPT_TIMESTAMP:
 			if (opt.opt_val != 0)
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 9ceaa1fdc3ab..487e181cff86 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -842,18 +842,6 @@ __u32 tcp_init_cwnd(const struct tcp_sock *tp, const struct dst_entry *dst)
 	return min_t(__u32, cwnd, tp->snd_cwnd_clamp);
 }
 
-/*
- * Packet counting of FACK is based on in-order assumptions, therefore TCP
- * disables it when reordering is detected
- */
-void tcp_disable_fack(struct tcp_sock *tp)
-{
-	/* RFC3517 uses different metric in lost marker => reset on change */
-	if (tcp_is_fack(tp))
-		tp->lost_skb_hint = NULL;
-	tp->rx_opt.sack_ok &= ~TCP_FACK_ENABLED;
-}
-
 /* Take a notice that peer is sending D-SACKs */
 static void tcp_dsack_seen(struct tcp_sock *tp)
 {
@@ -881,7 +869,6 @@ static void tcp_update_reordering(struct sock *sk, const int metric,
 			 tp->sacked_out,
 			 tp->undo_marker ? tp->undo_retrans : 0);
 #endif
-		tcp_disable_fack(tp);
 	}
 
 	tp->rack.reord = 1;
@@ -891,8 +878,6 @@ static void tcp_update_reordering(struct sock *sk, const int metric,
 		mib_idx = LINUX_MIB_TCPTSREORDER;
 	else if (tcp_is_reno(tp))
 		mib_idx = LINUX_MIB_TCPRENOREORDER;
-	else if (tcp_is_fack(tp))
-		mib_idx = LINUX_MIB_TCPFACKREORDER;
 	else
 		mib_idx = LINUX_MIB_TCPSACKREORDER;
 
@@ -970,7 +955,6 @@ void tcp_skb_mark_lost_uncond_verify(struct tcp_sock *tp, struct sk_buff *skb)
  * 3. Loss detection event of two flavors:
  *	A. Scoreboard estimator decided the packet is lost.
  *	   A'. Reno "three dupacks" marks head of queue lost.
- *	   A''. Its FACK modification, head until snd.fack is lost.
  *	B. SACK arrives sacking SND.NXT at the moment, when the
  *	   segment was retransmitted.
  * 4. D-SACK added new rule: D-SACK changes any tag to S.
@@ -1248,7 +1232,7 @@ static u8 tcp_sacktag_one(struct sock *sk,
 		fack_count += pcount;
 
 		/* Lost marker hint past SACKed? Tweak RFC3517 cnt */
-		if (!tcp_is_fack(tp) && tp->lost_skb_hint &&
+		if (tp->lost_skb_hint &&
 		    before(start_seq, TCP_SKB_CB(tp->lost_skb_hint)->seq))
 			tp->lost_cnt_hint += pcount;
 
@@ -2051,10 +2035,6 @@ static inline int tcp_fackets_out(const struct tcp_sock *tp)
  * counter when SACK is enabled (without SACK, sacked_out is used for
  * that purpose).
  *
- * Instead, with FACK TCP uses fackets_out that includes both SACKed
- * segments up to the highest received SACK block so far and holes in
- * between them.
- *
  * With reordering, holes may still be in flight, so RFC3517 recovery
  * uses pure sacked_out (total number of SACKed segments) even though
  * it violates the RFC that uses duplicate ACKs, often these are equal
@@ -2064,10 +2044,10 @@ static inline int tcp_fackets_out(const struct tcp_sock *tp)
  */
 static inline int tcp_dupack_heuristics(const struct tcp_sock *tp)
 {
-	return tcp_is_fack(tp) ? tp->fackets_out : tp->sacked_out + 1;
+	return tp->sacked_out + 1;
 }
 
-/* Linux NewReno/SACK/FACK/ECN state machine.
+/* Linux NewReno/SACK/ECN state machine.
  * --------------------------------------
  *
  * "Open"	Normal state, no dubious events, fast path.
@@ -2132,16 +2112,6 @@ static inline int tcp_dupack_heuristics(const struct tcp_sock *tp)
  *		dynamically measured and adjusted. This is implemented in
  *		tcp_rack_mark_lost.
  *
- *		FACK (Disabled by default. Subsumbed by RACK):
- *		It is the simplest heuristics. As soon as we decided
- *		that something is lost, we decide that _all_ not SACKed
- *		packets until the most forward SACK are lost. I.e.
- *		lost_out = fackets_out - sacked_out and left_out = fackets_out.
- *		It is absolutely correct estimate, if network does not reorder
- *		packets. And it loses any connection to reality when reordering
- *		takes place. We use FACK by default until reordering
- *		is suspected on the path to this destination.
- *
  *		If the receiver does not support SACK:
  *
  *		NewReno (RFC6582): in Recovery we assume that one segment
@@ -2190,7 +2160,7 @@ static bool tcp_time_to_recover(struct sock *sk, int flag)
 }
 
 /* Detect loss in event "A" above by marking head of queue up as lost.
- * For FACK or non-SACK(Reno) senders, the first "packets" number of segments
+ * For non-SACK(Reno) senders, the first "packets" number of segments
  * are considered lost. For RFC3517 SACK, a segment is considered lost if it
  * has at least tp->reordering SACKed seqments above it; "packets" refers to
  * the maximum SACKed segments to pass before reaching this limit.
@@ -2226,12 +2196,12 @@ static void tcp_mark_head_lost(struct sock *sk, int packets, int mark_head)
 			break;
 
 		oldcnt = cnt;
-		if (tcp_is_fack(tp) || tcp_is_reno(tp) ||
+		if (tcp_is_reno(tp) ||
 		    (TCP_SKB_CB(skb)->sacked & TCPCB_SACKED_ACKED))
 			cnt += tcp_skb_pcount(skb);
 
 		if (cnt > packets) {
-			if ((tcp_is_sack(tp) && !tcp_is_fack(tp)) ||
+			if (tcp_is_sack(tp) ||
 			    (TCP_SKB_CB(skb)->sacked & TCPCB_SACKED_ACKED) ||
 			    (oldcnt >= packets))
 				break;
@@ -2262,11 +2232,6 @@ static void tcp_update_scoreboard(struct sock *sk, int fast_rexmit)
 
 	if (tcp_is_reno(tp)) {
 		tcp_mark_head_lost(sk, 1, 1);
-	} else if (tcp_is_fack(tp)) {
-		int lost = tp->fackets_out - tp->reordering;
-		if (lost <= 0)
-			lost = 1;
-		tcp_mark_head_lost(sk, lost, 0);
 	} else {
 		int sacked_upto = tp->sacked_out - tp->reordering;
 		if (sacked_upto >= 0)
@@ -3199,8 +3164,7 @@ static int tcp_clean_rtx_queue(struct sock *sk, int prior_fackets,
 			if (reord < prior_fackets && reord <= tp->fackets_out)
 				tcp_update_reordering(sk, tp->fackets_out - reord, 0);
 
-			delta = tcp_is_fack(tp) ? pkts_acked :
-						  prior_sacked - tp->sacked_out;
+			delta = prior_sacked - tp->sacked_out;
 			tp->lost_cnt_hint -= min(tp->lost_cnt_hint, delta);
 		}
 
@@ -5708,9 +5672,6 @@ static int tcp_rcv_synsent_state_process(struct sock *sk, struct sk_buff *skb,
 			tp->tcp_header_len = sizeof(struct tcphdr);
 		}
 
-		if (tcp_is_sack(tp) && sock_net(sk)->ipv4.sysctl_tcp_fack)
-			tcp_enable_fack(tp);
-
 		tcp_sync_mss(sk, icsk->icsk_pmtu_cookie);
 		tcp_initialize_rcv_mss(sk);
 
diff --git a/net/ipv4/tcp_metrics.c b/net/ipv4/tcp_metrics.c
index 9d5ddebfd831..7097f92d16e5 100644
--- a/net/ipv4/tcp_metrics.c
+++ b/net/ipv4/tcp_metrics.c
@@ -470,10 +470,8 @@ void tcp_init_metrics(struct sock *sk)
 		tp->snd_ssthresh = TCP_INFINITE_SSTHRESH;
 	}
 	val = tcp_metric_get(tm, TCP_METRIC_REORDERING);
-	if (val && tp->reordering != val) {
-		tcp_disable_fack(tp);
+	if (val && tp->reordering != val)
 		tp->reordering = val;
-	}
 
 	crtt = tcp_metric_get(tm, TCP_METRIC_RTT);
 	rcu_read_unlock();
diff --git a/net/ipv4/tcp_minisocks.c b/net/ipv4/tcp_minisocks.c
index 4bb86580decd..326c9282bf94 100644
--- a/net/ipv4/tcp_minisocks.c
+++ b/net/ipv4/tcp_minisocks.c
@@ -509,10 +509,7 @@ struct sock *tcp_create_openreq_child(const struct sock *sk,
 						       keepalive_time_when(newtp));
 
 		newtp->rx_opt.tstamp_ok = ireq->tstamp_ok;
-		if ((newtp->rx_opt.sack_ok = ireq->sack_ok) != 0) {
-			if (sock_net(sk)->ipv4.sysctl_tcp_fack)
-				tcp_enable_fack(newtp);
-		}
+		newtp->rx_opt.sack_ok = ireq->sack_ok;
 		newtp->window_clamp = req->rsk_window_clamp;
 		newtp->rcv_ssthresh = req->rsk_rcv_wnd;
 		newtp->rcv_wnd = req->rsk_rcv_wnd;
diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
index 9b98d35aa0d8..094c429b4401 100644
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -1257,7 +1257,7 @@ static void tcp_adjust_pcount(struct sock *sk, const struct sk_buff *skb, int de
 
 	if (tp->lost_skb_hint &&
 	    before(TCP_SKB_CB(skb)->seq, TCP_SKB_CB(tp->lost_skb_hint)->seq) &&
-	    (tcp_is_fack(tp) || (TCP_SKB_CB(skb)->sacked & TCPCB_SACKED_ACKED)))
+	    (TCP_SKB_CB(skb)->sacked & TCPCB_SACKED_ACKED))
 		tp->lost_cnt_hint -= decr;
 
 	tcp_verify_left_out(tp);
@@ -2961,9 +2961,6 @@ int tcp_retransmit_skb(struct sock *sk, struct sk_buff *skb, int segs)
  * retransmitted data is acknowledged.  It tries to continue
  * resending the rest of the retransmit queue, until either
  * we've sent it all or the congestion window limit is reached.
- * If doing SACK, the first ACK which comes back for a timeout
- * based retransmit packet might feed us FACK information again.
- * If so, we use it to avoid unnecessarily retransmissions.
  */
 void tcp_xmit_retransmit_queue(struct sock *sk)
 {