1 files changed, 69 insertions, 31 deletions

diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 21ed010d7551..755b44c7506a 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -260,7 +260,8 @@ 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;
 	}
 }
 
@@ -359,7 +360,8 @@ static void tcp_sndbuf_expand(struct sock *sk)
 	sndmem *= nr_segs * per_mss;
 
 	if (sk->sk_sndbuf < sndmem)
-		sk->sk_sndbuf = min(sndmem, sock_net(sk)->ipv4.sysctl_tcp_wmem[2]);
+		WRITE_ONCE(sk->sk_sndbuf,
+			   min(sndmem, sock_net(sk)->ipv4.sysctl_tcp_wmem[2]));
 }
 
 /* 2. Tuning advertised window (window_clamp, rcv_ssthresh)
@@ -483,8 +485,9 @@ static void tcp_clamp_window(struct sock *sk)
 	    !(sk->sk_userlocks & SOCK_RCVBUF_LOCK) &&
 	    !tcp_under_memory_pressure(sk) &&
 	    sk_memory_allocated(sk) < sk_prot_mem_limits(sk, 0)) {
-		sk->sk_rcvbuf = min(atomic_read(&sk->sk_rmem_alloc),
-				    net->ipv4.sysctl_tcp_rmem[2]);
+		WRITE_ONCE(sk->sk_rcvbuf,
+			   min(atomic_read(&sk->sk_rmem_alloc),
+			       net->ipv4.sysctl_tcp_rmem[2]));
 	}
 	if (atomic_read(&sk->sk_rmem_alloc) > sk->sk_rcvbuf)
 		tp->rcv_ssthresh = min(tp->window_clamp, 2U * tp->advmss);
@@ -648,7 +651,7 @@ void tcp_rcv_space_adjust(struct sock *sk)
 		rcvbuf = min_t(u64, rcvwin * rcvmem,
 			       sock_net(sk)->ipv4.sysctl_tcp_rmem[2]);
 		if (rcvbuf > sk->sk_rcvbuf) {
-			sk->sk_rcvbuf = rcvbuf;
+			WRITE_ONCE(sk->sk_rcvbuf, rcvbuf);
 
 			/* Make the window clamp follow along.  */
 			tp->window_clamp = tcp_win_from_space(sk, rcvbuf);
@@ -909,9 +912,10 @@ static void tcp_check_sack_reordering(struct sock *sk, const u32 low_seq,
 /* This must be called before lost_out is incremented */
 static void tcp_verify_retransmit_hint(struct tcp_sock *tp, struct sk_buff *skb)
 {
-	if (!tp->retransmit_skb_hint ||
-	    before(TCP_SKB_CB(skb)->seq,
-		   TCP_SKB_CB(tp->retransmit_skb_hint)->seq))
+	if ((!tp->retransmit_skb_hint && tp->retrans_out >= tp->lost_out) ||
+	    (tp->retransmit_skb_hint &&
+	     before(TCP_SKB_CB(skb)->seq,
+		    TCP_SKB_CB(tp->retransmit_skb_hint)->seq)))
 		tp->retransmit_skb_hint = skb;
 }
 
@@ -1721,8 +1725,11 @@ tcp_sacktag_write_queue(struct sock *sk, const struct sk_buff *ack_skb,
 		}
 
 		/* Ignore very old stuff early */
-		if (!after(sp[used_sacks].end_seq, prior_snd_una))
+		if (!after(sp[used_sacks].end_seq, prior_snd_una)) {
+			if (i == 0)
+				first_sack_index = -1;
 			continue;
+		}
 
 		used_sacks++;
 	}
@@ -2662,7 +2669,7 @@ static void tcp_process_loss(struct sock *sk, int flag, int num_dupack,
 	struct tcp_sock *tp = tcp_sk(sk);
 	bool recovered = !before(tp->snd_una, tp->high_seq);
 
-	if ((flag & FLAG_SND_UNA_ADVANCED || tp->fastopen_rsk) &&
+	if ((flag & FLAG_SND_UNA_ADVANCED || rcu_access_pointer(tp->fastopen_rsk)) &&
 	    tcp_try_undo_loss(sk, false))
 		return;
 
@@ -2933,6 +2940,8 @@ static bool tcp_ack_update_rtt(struct sock *sk, const int flag,
 		u32 delta = tcp_time_stamp(tp) - tp->rx_opt.rcv_tsecr;
 
 		if (likely(delta < INT_MAX / (USEC_PER_SEC / TCP_TS_HZ))) {
+			if (!delta)
+				delta = 1;
 			seq_rtt_us = delta * (USEC_PER_SEC / TCP_TS_HZ);
 			ca_rtt_us = seq_rtt_us;
 		}
@@ -2986,7 +2995,7 @@ void tcp_rearm_rto(struct sock *sk)
 	/* If the retrans timer is currently being used by Fast Open
 	 * for SYN-ACK retrans purpose, stay put.
 	 */
-	if (tp->fastopen_rsk)
+	if (rcu_access_pointer(tp->fastopen_rsk))
 		return;
 
 	if (!tp->packets_out) {
@@ -3154,6 +3163,7 @@ static int tcp_clean_rtx_queue(struct sock *sk, u32 prior_fack,
 			tp->retransmit_skb_hint = NULL;
 		if (unlikely(skb == tp->lost_skb_hint))
 			tp->lost_skb_hint = NULL;
+		tcp_highest_sack_replace(sk, skb, next);
 		tcp_rtx_queue_unlink_and_free(skb, sk);
 	}
 
@@ -3358,7 +3368,7 @@ static void tcp_rcv_nxt_update(struct tcp_sock *tp, u32 seq)
 
 	sock_owned_by_me((struct sock *)tp);
 	tp->bytes_received += delta;
-	tp->rcv_nxt = seq;
+	WRITE_ONCE(tp->rcv_nxt, seq);
 }
 
 /* Update our send window.
@@ -3493,10 +3503,8 @@ 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)
 {
@@ -3505,7 +3513,10 @@ 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)) {
@@ -3671,6 +3682,15 @@ 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...
 	 */
@@ -4500,6 +4520,7 @@ 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;
 	}
@@ -4532,7 +4553,11 @@ 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;
@@ -4616,7 +4641,11 @@ 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);
 	}
@@ -4696,7 +4725,8 @@ 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);
@@ -4715,8 +4745,6 @@ 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.
@@ -4735,6 +4763,7 @@ 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;
 		}
 
@@ -5308,7 +5337,7 @@ static void tcp_check_urg(struct sock *sk, const struct tcphdr *th)
 	}
 
 	tp->urg_data = TCP_URG_NOTYET;
-	tp->urg_seq = ptr;
+	WRITE_ONCE(tp->urg_seq, ptr);
 
 	/* Disable header prediction. */
 	tp->pred_flags = 0;
@@ -5884,7 +5913,7 @@ static int tcp_rcv_synsent_state_process(struct sock *sk, struct sk_buff *skb,
 		/* Ok.. it's good. Set up sequence numbers and
 		 * move to established.
 		 */
-		tp->rcv_nxt = TCP_SKB_CB(skb)->seq + 1;
+		WRITE_ONCE(tp->rcv_nxt, TCP_SKB_CB(skb)->seq + 1);
 		tp->rcv_wup = TCP_SKB_CB(skb)->seq + 1;
 
 		/* RFC1323: The window in SYN & SYN/ACK segments is
@@ -5913,7 +5942,7 @@ static int tcp_rcv_synsent_state_process(struct sock *sk, struct sk_buff *skb,
 		/* Remember, tcp_poll() does not lock socket!
 		 * Change state from SYN-SENT only after copied_seq
 		 * is initialized. */
-		tp->copied_seq = tp->rcv_nxt;
+		WRITE_ONCE(tp->copied_seq, tp->rcv_nxt);
 
 		smc_check_reset_syn(tp);
 
@@ -5987,8 +6016,8 @@ discard:
 			tp->tcp_header_len = sizeof(struct tcphdr);
 		}
 
-		tp->rcv_nxt = TCP_SKB_CB(skb)->seq + 1;
-		tp->copied_seq = tp->rcv_nxt;
+		WRITE_ONCE(tp->rcv_nxt, TCP_SKB_CB(skb)->seq + 1);
+		WRITE_ONCE(tp->copied_seq, tp->rcv_nxt);
 		tp->rcv_wup = TCP_SKB_CB(skb)->seq + 1;
 
 		/* RFC1323: The window in SYN & SYN/ACK segments is
@@ -6039,7 +6068,13 @@ reset_and_undo:
 
 static void tcp_rcv_synrecv_state_fastopen(struct sock *sk)
 {
-	tcp_try_undo_loss(sk, false);
+	struct request_sock *req;
+
+	/* If we are still handling the SYNACK RTO, see if timestamp ECR allows
+	 * undo. If peer SACKs triggered fast recovery, we can't undo here.
+	 */
+	if (inet_csk(sk)->icsk_ca_state == TCP_CA_Loss)
+		tcp_try_undo_loss(sk, false);
 
 	/* Reset rtx states to prevent spurious retransmits_timed_out() */
 	tcp_sk(sk)->retrans_stamp = 0;
@@ -6048,7 +6083,9 @@ static void tcp_rcv_synrecv_state_fastopen(struct sock *sk)
 	/* Once we leave TCP_SYN_RECV or TCP_FIN_WAIT_1,
 	 * we no longer need req so release it.
 	 */
-	reqsk_fastopen_remove(sk, tcp_sk(sk)->fastopen_rsk, false);
+	req = rcu_dereference_protected(tcp_sk(sk)->fastopen_rsk,
+					lockdep_sock_is_held(sk));
+	reqsk_fastopen_remove(sk, req, false);
 
 	/* Re-arm the timer because data may have been sent out.
 	 * This is similar to the regular data transmission case
@@ -6123,7 +6160,8 @@ int tcp_rcv_state_process(struct sock *sk, struct sk_buff *skb)
 
 	tcp_mstamp_refresh(tp);
 	tp->rx_opt.saw_tstamp = 0;
-	req = tp->fastopen_rsk;
+	req = rcu_dereference_protected(tp->fastopen_rsk,
+					lockdep_sock_is_held(sk));
 	if (req) {
 		bool req_stolen;
 
@@ -6163,7 +6201,7 @@ int tcp_rcv_state_process(struct sock *sk, struct sk_buff *skb)
 			tcp_try_undo_spurious_syn(sk);
 			tp->retrans_stamp = 0;
 			tcp_init_transfer(sk, BPF_SOCK_OPS_PASSIVE_ESTABLISHED_CB);
-			tp->copied_seq = tp->rcv_nxt;
+			WRITE_ONCE(tp->copied_seq, tp->rcv_nxt);
 		}
 		smp_mb();
 		tcp_set_state(sk, TCP_ESTABLISHED);