1 files changed, 81 insertions, 57 deletions

diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index bf4ced9273e8..518f04355fbf 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -261,7 +261,8 @@ static void tcp_ecn_accept_cwr(struct sock *sk, const struct sk_buff *skb)
 		 * cwnd may be very low (even just 1 packet), so we should ACK
 		 * immediately.
 		 */
-		inet_csk(sk)->icsk_ack.pending |= ICSK_ACK_NOW;
+		if (TCP_SKB_CB(skb)->seq != TCP_SKB_CB(skb)->end_seq)
+			inet_csk(sk)->icsk_ack.pending |= ICSK_ACK_NOW;
 	}
 }
 
@@ -437,7 +438,7 @@ static void tcp_grow_window(struct sock *sk, const struct sk_buff *skb)
 /* 3. Try to fixup all. It is made immediately after connection enters
  *    established state.
  */
-void tcp_init_buffer_space(struct sock *sk)
+static void tcp_init_buffer_space(struct sock *sk)
 {
 	int tcp_app_win = sock_net(sk)->ipv4.sysctl_tcp_app_win;
 	struct tcp_sock *tp = tcp_sk(sk);
@@ -2183,8 +2184,7 @@ 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 non-SACK(Reno) senders, the first "packets" number of segments
- * are considered lost. For RFC3517 SACK, a segment is considered lost if it
+ * 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.
  */
@@ -2192,10 +2192,9 @@ static void tcp_mark_head_lost(struct sock *sk, int packets, int mark_head)
 {
 	struct tcp_sock *tp = tcp_sk(sk);
 	struct sk_buff *skb;
-	int cnt, oldcnt, lost;
-	unsigned int mss;
+	int cnt;
 	/* Use SACK to deduce losses of new sequences sent during recovery */
-	const u32 loss_high = tcp_is_sack(tp) ?  tp->snd_nxt : tp->high_seq;
+	const u32 loss_high = tp->snd_nxt;
 
 	WARN_ON(packets > tp->packets_out);
 	skb = tp->lost_skb_hint;
@@ -2218,26 +2217,11 @@ static void tcp_mark_head_lost(struct sock *sk, int packets, int mark_head)
 		if (after(TCP_SKB_CB(skb)->end_seq, loss_high))
 			break;
 
-		oldcnt = cnt;
-		if (tcp_is_reno(tp) ||
-		    (TCP_SKB_CB(skb)->sacked & TCPCB_SACKED_ACKED))
+		if (TCP_SKB_CB(skb)->sacked & TCPCB_SACKED_ACKED)
 			cnt += tcp_skb_pcount(skb);
 
-		if (cnt > packets) {
-			if (tcp_is_sack(tp) ||
-			    (TCP_SKB_CB(skb)->sacked & TCPCB_SACKED_ACKED) ||
-			    (oldcnt >= packets))
-				break;
-
-			mss = tcp_skb_mss(skb);
-			/* If needed, chop off the prefix to mark as lost. */
-			lost = (packets - oldcnt) * mss;
-			if (lost < skb->len &&
-			    tcp_fragment(sk, TCP_FRAG_IN_RTX_QUEUE, skb,
-					 lost, mss, GFP_ATOMIC) < 0)
-				break;
-			cnt = packets;
-		}
+		if (cnt > packets)
+			break;
 
 		tcp_skb_mark_lost(tp, skb);
 
@@ -2849,8 +2833,7 @@ static void tcp_fastretrans_alert(struct sock *sk, const u32 prior_snd_una,
 			if (tcp_try_undo_partial(sk, prior_snd_una))
 				return;
 			/* Partial ACK arrived. Force fast retransmit. */
-			do_lost = tcp_is_reno(tp) ||
-				  tcp_force_fast_retransmit(sk);
+			do_lost = tcp_force_fast_retransmit(sk);
 		}
 		if (tcp_try_undo_dsack(sk)) {
 			tcp_try_keep_open(sk);
@@ -3014,7 +2997,7 @@ void tcp_rearm_rto(struct sock *sk)
 			rto = usecs_to_jiffies(max_t(int, delta_us, 1));
 		}
 		tcp_reset_xmit_timer(sk, ICSK_TIME_RETRANS, rto,
-				     TCP_RTO_MAX, tcp_rtx_queue_head(sk));
+				     TCP_RTO_MAX);
 	}
 }
 
@@ -3291,7 +3274,7 @@ static void tcp_ack_probe(struct sock *sk)
 		unsigned long when = tcp_probe0_when(sk, TCP_RTO_MAX);
 
 		tcp_reset_xmit_timer(sk, ICSK_TIME_PROBE0,
-				     when, TCP_RTO_MAX, NULL);
+				     when, TCP_RTO_MAX);
 	}
 }
 
@@ -3505,10 +3488,8 @@ static void tcp_replace_ts_recent(struct tcp_sock *tp, u32 seq)
 	}
 }
 
-/* This routine deals with acks during a TLP episode.
- * We mark the end of a TLP episode on receiving TLP dupack or when
- * ack is after tlp_high_seq.
- * Ref: loss detection algorithm in draft-dukkipati-tcpm-tcp-loss-probe.
+/* This routine deals with acks during a TLP episode and ends an episode by
+ * resetting tlp_high_seq. Ref: TLP algorithm in draft-ietf-tcpm-rack
  */
 static void tcp_process_tlp_ack(struct sock *sk, u32 ack, int flag)
 {
@@ -3517,7 +3498,10 @@ static void tcp_process_tlp_ack(struct sock *sk, u32 ack, int flag)
 	if (before(ack, tp->tlp_high_seq))
 		return;
 
-	if (flag & FLAG_DSACKING_ACK) {
+	if (!tp->tlp_retrans) {
+		/* TLP of new data has been acknowledged */
+		tp->tlp_high_seq = 0;
+	} else if (flag & FLAG_DSACKING_ACK) {
 		/* This DSACK means original and TLP probe arrived; no loss */
 		tp->tlp_high_seq = 0;
 	} else if (after(ack, tp->tlp_high_seq)) {
@@ -3683,6 +3667,15 @@ static int tcp_ack(struct sock *sk, const struct sk_buff *skb, int flag)
 		tcp_in_ack_event(sk, ack_ev_flags);
 	}
 
+	/* This is a deviation from RFC3168 since it states that:
+	 * "When the TCP data sender is ready to set the CWR bit after reducing
+	 * the congestion window, it SHOULD set the CWR bit only on the first
+	 * new data packet that it transmits."
+	 * We accept CWR on pure ACKs to be more robust
+	 * with widely-deployed TCP implementations that do this.
+	 */
+	tcp_ecn_accept_cwr(sk, skb);
+
 	/* We passed data and got it acked, remove any soft error
 	 * log. Something worked...
 	 */
@@ -3926,10 +3919,6 @@ void tcp_parse_options(const struct net *net,
 				 */
 				break;
 #endif
-			case TCPOPT_MPTCP:
-				mptcp_parse_option(skb, ptr, opsize, opt_rx);
-				break;
-
 			case TCPOPT_FASTOPEN:
 				tcp_parse_fastopen_option(
 					opsize - TCPOLEN_FASTOPEN_BASE,
@@ -4327,6 +4316,33 @@ static void tcp_sack_maybe_coalesce(struct tcp_sock *tp)
 	}
 }
 
+static void tcp_sack_compress_send_ack(struct sock *sk)
+{
+	struct tcp_sock *tp = tcp_sk(sk);
+
+	if (!tp->compressed_ack)
+		return;
+
+	if (hrtimer_try_to_cancel(&tp->compressed_ack_timer) == 1)
+		__sock_put(sk);
+
+	/* Since we have to send one ack finally,
+	 * substract one from tp->compressed_ack to keep
+	 * LINUX_MIB_TCPACKCOMPRESSED accurate.
+	 */
+	NET_ADD_STATS(sock_net(sk), LINUX_MIB_TCPACKCOMPRESSED,
+		      tp->compressed_ack - 1);
+
+	tp->compressed_ack = 0;
+	tcp_send_ack(sk);
+}
+
+/* Reasonable amount of sack blocks included in TCP SACK option
+ * The max is 4, but this becomes 3 if TCP timestamps are there.
+ * Given that SACK packets might be lost, be conservative and use 2.
+ */
+#define TCP_SACK_BLOCKS_EXPECTED 2
+
 static void tcp_sack_new_ofo_skb(struct sock *sk, u32 seq, u32 end_seq)
 {
 	struct tcp_sock *tp = tcp_sk(sk);
@@ -4339,6 +4355,8 @@ static void tcp_sack_new_ofo_skb(struct sock *sk, u32 seq, u32 end_seq)
 
 	for (this_sack = 0; this_sack < cur_sacks; this_sack++, sp++) {
 		if (tcp_sack_extend(sp, seq, end_seq)) {
+			if (this_sack >= TCP_SACK_BLOCKS_EXPECTED)
+				tcp_sack_compress_send_ack(sk);
 			/* Rotate this_sack to the first one. */
 			for (; this_sack > 0; this_sack--, sp--)
 				swap(*sp, *(sp - 1));
@@ -4348,6 +4366,9 @@ static void tcp_sack_new_ofo_skb(struct sock *sk, u32 seq, u32 end_seq)
 		}
 	}
 
+	if (this_sack >= TCP_SACK_BLOCKS_EXPECTED)
+		tcp_sack_compress_send_ack(sk);
+
 	/* Could not find an adjacent existing SACK, build a new one,
 	 * put it at the front, and shift everyone else down.  We
 	 * always know there is at least one SACK present already here.
@@ -4355,8 +4376,6 @@ static void tcp_sack_new_ofo_skb(struct sock *sk, u32 seq, u32 end_seq)
 	 * If the sack array is full, forget about the last one.
 	 */
 	if (this_sack >= TCP_NUM_SACKS) {
-		if (tp->compressed_ack > TCP_FASTRETRANS_THRESH)
-			tcp_send_ack(sk);
 		this_sack--;
 		tp->rx_opt.num_sacks--;
 		sp--;
@@ -4564,6 +4583,7 @@ static void tcp_data_queue_ofo(struct sock *sk, struct sk_buff *skb)
 
 	if (unlikely(tcp_try_rmem_schedule(sk, skb, skb->truesize))) {
 		NET_INC_STATS(sock_net(sk), LINUX_MIB_TCPOFODROP);
+		sk->sk_data_ready(sk);
 		tcp_drop(sk, skb);
 		return;
 	}
@@ -4597,7 +4617,11 @@ 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;
@@ -4681,7 +4705,11 @@ add_sack:
 		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);
 	}
@@ -4761,7 +4789,8 @@ void tcp_data_ready(struct sock *sk)
 	const struct tcp_sock *tp = tcp_sk(sk);
 	int avail = tp->rcv_nxt - tp->copied_seq;
 
-	if (avail < sk->sk_rcvlowat && !sock_flag(sk, SOCK_DONE))
+	if (avail < sk->sk_rcvlowat && !tcp_rmem_pressure(sk) &&
+	    !sock_flag(sk, SOCK_DONE))
 		return;
 
 	sk->sk_data_ready(sk);
@@ -4783,8 +4812,6 @@ static void tcp_data_queue(struct sock *sk, struct sk_buff *skb)
 	skb_dst_drop(skb);
 	__skb_pull(skb, tcp_hdr(skb)->doff * 4);
 
-	tcp_ecn_accept_cwr(sk, skb);
-
 	tp->rx_opt.dsack = 0;
 
 	/*  Queue data for delivery to the user.
@@ -4803,6 +4830,7 @@ queue_and_out:
 			sk_forced_mem_schedule(sk, skb->truesize);
 		else if (tcp_try_rmem_schedule(sk, skb, skb->truesize)) {
 			NET_INC_STATS(sock_net(sk), LINUX_MIB_TCPRCVQDROP);
+			sk->sk_data_ready(sk);
 			goto drop;
 		}
 
@@ -5275,15 +5303,13 @@ send_now:
 
 	if (tp->compressed_ack_rcv_nxt != tp->rcv_nxt) {
 		tp->compressed_ack_rcv_nxt = tp->rcv_nxt;
-		if (tp->compressed_ack > TCP_FASTRETRANS_THRESH)
-			NET_ADD_STATS(sock_net(sk), LINUX_MIB_TCPACKCOMPRESSED,
-				      tp->compressed_ack - TCP_FASTRETRANS_THRESH);
-		tp->compressed_ack = 0;
+		tp->dup_ack_counter = 0;
 	}
-
-	if (++tp->compressed_ack <= TCP_FASTRETRANS_THRESH)
+	if (tp->dup_ack_counter < TCP_FASTRETRANS_THRESH) {
+		tp->dup_ack_counter++;
 		goto send_now;
-
+	}
+	tp->compressed_ack++;
 	if (hrtimer_is_queued(&tp->compressed_ack_timer))
 		return;
 
@@ -5296,8 +5322,9 @@ send_now:
 	delay = min_t(unsigned long, sock_net(sk)->ipv4.sysctl_tcp_comp_sack_delay_ns,
 		      rtt * (NSEC_PER_USEC >> 3)/20);
 	sock_hold(sk);
-	hrtimer_start(&tp->compressed_ack_timer, ns_to_ktime(delay),
-		      HRTIMER_MODE_REL_PINNED_SOFT);
+	hrtimer_start_range_ns(&tp->compressed_ack_timer, ns_to_ktime(delay),
+			       sock_net(sk)->ipv4.sysctl_tcp_comp_sack_slack_ns,
+			       HRTIMER_MODE_REL_PINNED_SOFT);
 }
 
 static inline void tcp_ack_snd_check(struct sock *sk)
@@ -5990,9 +6017,6 @@ static int tcp_rcv_synsent_state_process(struct sock *sk, struct sk_buff *skb,
 		tcp_sync_mss(sk, icsk->icsk_pmtu_cookie);
 		tcp_initialize_rcv_mss(sk);
 
-		if (sk_is_mptcp(sk))
-			mptcp_rcv_synsent(sk);
-
 		/* Remember, tcp_poll() does not lock socket!
 		 * Change state from SYN-SENT only after copied_seq
 		 * is initialized. */