From 3c7151275c0c9a80c3375f9874b1c7129a105eea Mon Sep 17 00:00:00 2001
From: jbaron@akamai.com
Date: Mon, 20 Apr 2015 20:05:07 +0000
Subject: tcp: add memory barriers to write space paths

Ensure that we either see that the buffer has write space
in tcp_poll() or that we perform a wakeup from the input
side. Did not run into any actual problem here, but thought
that we should make things explicit.

Signed-off-by: Jason Baron <jbaron@akamai.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
 net/ipv4/tcp_input.c | 2 ++
 1 file changed, 2 insertions(+)

(limited to 'net/ipv4/tcp_input.c')

diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index a7ef679dd3ea..3a4d9b34bed4 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -4845,6 +4845,8 @@ static void tcp_check_space(struct sock *sk)
 {
 	if (sock_flag(sk, SOCK_QUEUE_SHRUNK)) {
 		sock_reset_flag(sk, SOCK_QUEUE_SHRUNK);
+		/* pairs with tcp_poll() */
+		smp_mb__after_atomic();
 		if (sk->sk_socket &&
 		    test_bit(SOCK_NOSPACE, &sk->sk_socket->flags))
 			tcp_new_space(sk);
-- 
cgit v1.2.3-55-g7522


From 0df48c26d8418c5c9fba63fac15b660d70ca2f1c Mon Sep 17 00:00:00 2001
From: Eric Dumazet
Date: Tue, 28 Apr 2015 15:28:17 -0700
Subject: tcp: add tcpi_bytes_acked to tcp_info

This patch tracks total number of bytes acked for a TCP socket.
This is the sum of all changes done to tp->snd_una, and allows
for precise tracking of delivered data.

RFC4898 named this : tcpEStatsAppHCThruOctetsAcked

This is a 64bit field, and can be fetched both from TCP_INFO
getsockopt() if one has a handle on a TCP socket, or from inet_diag
netlink facility (iproute2/ss patch will follow)

Note that tp->bytes_acked was placed near tp->snd_una for
best data locality and minimal performance impact.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Yuchung Cheng <ycheng@google.com>
Cc: Matt Mathis <mattmathis@google.com>
Cc: Eric Salo <salo@google.com>
Cc: Martin Lau <kafai@fb.com>
Cc: Chris Rapier <rapier@psc.edu>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
 include/linux/tcp.h      |  4 ++++
 include/net/tcp.h        |  2 +-
 include/uapi/linux/tcp.h |  1 +
 net/ipv4/tcp.c           |  6 +++++-
 net/ipv4/tcp_input.c     | 13 +++++++++++--
 5 files changed, 22 insertions(+), 4 deletions(-)

(limited to 'net/ipv4/tcp_input.c')

diff --git a/include/linux/tcp.h b/include/linux/tcp.h
index 0caa3a2d4106..0f73b43171da 100644
--- a/include/linux/tcp.h
+++ b/include/linux/tcp.h
@@ -150,6 +150,10 @@ struct tcp_sock {
 	u32	rcv_wup;	/* rcv_nxt on last window update sent	*/
  	u32	snd_nxt;	/* Next sequence we send		*/
 
+	u64	bytes_acked;	/* RFC4898 tcpEStatsAppHCThruOctetsAcked
+				 * sum(delta(snd_una)), or how many bytes
+				 * were acked.
+				 */
  	u32	snd_una;	/* First byte we want an ack for	*/
  	u32	snd_sml;	/* Last byte of the most recently transmitted small packet */
 	u32	rcv_tstamp;	/* timestamp of last received ACK (for keepalives) */
diff --git a/include/net/tcp.h b/include/net/tcp.h
index 051dc5c2802d..dd7b4ea6a10c 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -576,7 +576,7 @@ static inline int tcp_bound_to_half_wnd(struct tcp_sock *tp, int pktsize)
 }
 
 /* tcp.c */
-void tcp_get_info(const struct sock *, struct tcp_info *);
+void tcp_get_info(struct sock *, struct tcp_info *);
 
 /* Read 'sendfile()'-style from a TCP socket */
 typedef int (*sk_read_actor_t)(read_descriptor_t *, struct sk_buff *,
diff --git a/include/uapi/linux/tcp.h b/include/uapi/linux/tcp.h
index 3b9718328d8b..6666e98a0af9 100644
--- a/include/uapi/linux/tcp.h
+++ b/include/uapi/linux/tcp.h
@@ -189,6 +189,7 @@ struct tcp_info {
 
 	__u64	tcpi_pacing_rate;
 	__u64	tcpi_max_pacing_rate;
+	__u64	tcpi_bytes_acked; /* RFC4898 tcpEStatsAppHCThruOctetsAcked */
 };
 
 /* for TCP_MD5SIG socket option */
diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
index 8c5cd9efebbc..4bf0e8ca7b5b 100644
--- a/net/ipv4/tcp.c
+++ b/net/ipv4/tcp.c
@@ -2592,7 +2592,7 @@ EXPORT_SYMBOL(compat_tcp_setsockopt);
 #endif
 
 /* Return information about state of tcp endpoint in API format. */
-void tcp_get_info(const struct sock *sk, struct tcp_info *info)
+void tcp_get_info(struct sock *sk, struct tcp_info *info)
 {
 	const struct tcp_sock *tp = tcp_sk(sk);
 	const struct inet_connection_sock *icsk = inet_csk(sk);
@@ -2663,6 +2663,10 @@ void tcp_get_info(const struct sock *sk, struct tcp_info *info)
 
 	rate = READ_ONCE(sk->sk_max_pacing_rate);
 	info->tcpi_max_pacing_rate = rate != ~0U ? rate : ~0ULL;
+
+	spin_lock_bh(&sk->sk_lock.slock);
+	info->tcpi_bytes_acked = tp->bytes_acked;
+	spin_unlock_bh(&sk->sk_lock.slock);
 }
 EXPORT_SYMBOL_GPL(tcp_get_info);
 
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 3a4d9b34bed4..378d3f4d4dc3 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -3280,6 +3280,15 @@ static inline bool tcp_may_update_window(const struct tcp_sock *tp,
 		(ack_seq == tp->snd_wl1 && nwin > tp->snd_wnd);
 }
 
+/* If we update tp->snd_una, also update tp->bytes_acked */
+static void tcp_snd_una_update(struct tcp_sock *tp, u32 ack)
+{
+	u32 delta = ack - tp->snd_una;
+
+	tp->bytes_acked += delta;
+	tp->snd_una = ack;
+}
+
 /* Update our send window.
  *
  * Window update algorithm, described in RFC793/RFC1122 (used in linux-2.2
@@ -3315,7 +3324,7 @@ static int tcp_ack_update_window(struct sock *sk, const struct sk_buff *skb, u32
 		}
 	}
 
-	tp->snd_una = ack;
+	tcp_snd_una_update(tp, ack);
 
 	return flag;
 }
@@ -3497,7 +3506,7 @@ static int tcp_ack(struct sock *sk, const struct sk_buff *skb, int flag)
 		 * Note, we use the fact that SND.UNA>=SND.WL2.
 		 */
 		tcp_update_wl(tp, ack_seq);
-		tp->snd_una = ack;
+		tcp_snd_una_update(tp, ack);
 		flag |= FLAG_WIN_UPDATE;
 
 		tcp_in_ack_event(sk, CA_ACK_WIN_UPDATE);
-- 
cgit v1.2.3-55-g7522


From bdd1f9edacb5f5835d1e6276571bbbe5b88ded48 Mon Sep 17 00:00:00 2001
From: Eric Dumazet
Date: Tue, 28 Apr 2015 15:28:18 -0700
Subject: tcp: add tcpi_bytes_received to tcp_info

This patch tracks total number of payload bytes received on a TCP socket.
This is the sum of all changes done to tp->rcv_nxt

RFC4898 named this : tcpEStatsAppHCThruOctetsReceived

This is a 64bit field, and can be fetched both from TCP_INFO
getsockopt() if one has a handle on a TCP socket, or from inet_diag
netlink facility (iproute2/ss patch will follow)

Note that tp->bytes_received was placed near tp->rcv_nxt for
best data locality and minimal performance impact.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Yuchung Cheng <ycheng@google.com>
Cc: Matt Mathis <mattmathis@google.com>
Cc: Eric Salo <salo@google.com>
Cc: Martin Lau <kafai@fb.com>
Cc: Chris Rapier <rapier@psc.edu>
Acked-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
 include/linux/tcp.h      |  4 ++++
 include/uapi/linux/tcp.h |  1 +
 net/ipv4/tcp.c           |  1 +
 net/ipv4/tcp_fastopen.c  |  1 +
 net/ipv4/tcp_input.c     | 17 +++++++++++++----
 5 files changed, 20 insertions(+), 4 deletions(-)

(limited to 'net/ipv4/tcp_input.c')

diff --git a/include/linux/tcp.h b/include/linux/tcp.h
index 0f73b43171da..3b2911502a8c 100644
--- a/include/linux/tcp.h
+++ b/include/linux/tcp.h
@@ -145,6 +145,10 @@ struct tcp_sock {
  *	read the code and the spec side by side (and laugh ...)
  *	See RFC793 and RFC1122. The RFC writes these in capitals.
  */
+	u64	bytes_received;	/* RFC4898 tcpEStatsAppHCThruOctetsReceived
+				 * sum(delta(rcv_nxt)), or how many bytes
+				 * were acked.
+				 */
  	u32	rcv_nxt;	/* What we want to receive next 	*/
 	u32	copied_seq;	/* Head of yet unread data		*/
 	u32	rcv_wup;	/* rcv_nxt on last window update sent	*/
diff --git a/include/uapi/linux/tcp.h b/include/uapi/linux/tcp.h
index 6666e98a0af9..a48f93f3207b 100644
--- a/include/uapi/linux/tcp.h
+++ b/include/uapi/linux/tcp.h
@@ -190,6 +190,7 @@ struct tcp_info {
 	__u64	tcpi_pacing_rate;
 	__u64	tcpi_max_pacing_rate;
 	__u64	tcpi_bytes_acked; /* RFC4898 tcpEStatsAppHCThruOctetsAcked */
+	__u64	tcpi_bytes_received; /* RFC4898 tcpEStatsAppHCThruOctetsReceived */
 };
 
 /* for TCP_MD5SIG socket option */
diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
index 4bf0e8ca7b5b..99fcc0b22c92 100644
--- a/net/ipv4/tcp.c
+++ b/net/ipv4/tcp.c
@@ -2666,6 +2666,7 @@ void tcp_get_info(struct sock *sk, struct tcp_info *info)
 
 	spin_lock_bh(&sk->sk_lock.slock);
 	info->tcpi_bytes_acked = tp->bytes_acked;
+	info->tcpi_bytes_received = tp->bytes_received;
 	spin_unlock_bh(&sk->sk_lock.slock);
 }
 EXPORT_SYMBOL_GPL(tcp_get_info);
diff --git a/net/ipv4/tcp_fastopen.c b/net/ipv4/tcp_fastopen.c
index e3d87aca6be8..3c673d5e6cff 100644
--- a/net/ipv4/tcp_fastopen.c
+++ b/net/ipv4/tcp_fastopen.c
@@ -206,6 +206,7 @@ static bool tcp_fastopen_create_child(struct sock *sk,
 			skb_set_owner_r(skb2, child);
 			__skb_queue_tail(&child->sk_receive_queue, skb2);
 			tp->syn_data_acked = 1;
+			tp->bytes_received = end_seq - TCP_SKB_CB(skb)->seq - 1;
 		} else {
 			end_seq = TCP_SKB_CB(skb)->seq + 1;
 		}
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 378d3f4d4dc3..7e6962bcfc30 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -3289,6 +3289,15 @@ static void tcp_snd_una_update(struct tcp_sock *tp, u32 ack)
 	tp->snd_una = ack;
 }
 
+/* If we update tp->rcv_nxt, also update tp->bytes_received */
+static void tcp_rcv_nxt_update(struct tcp_sock *tp, u32 seq)
+{
+	u32 delta = seq - tp->rcv_nxt;
+
+	tp->bytes_received += delta;
+	tp->rcv_nxt = seq;
+}
+
 /* Update our send window.
  *
  * Window update algorithm, described in RFC793/RFC1122 (used in linux-2.2
@@ -4245,7 +4254,7 @@ static void tcp_ofo_queue(struct sock *sk)
 
 		tail = skb_peek_tail(&sk->sk_receive_queue);
 		eaten = tail && tcp_try_coalesce(sk, tail, skb, &fragstolen);
-		tp->rcv_nxt = TCP_SKB_CB(skb)->end_seq;
+		tcp_rcv_nxt_update(tp, TCP_SKB_CB(skb)->end_seq);
 		if (!eaten)
 			__skb_queue_tail(&sk->sk_receive_queue, skb);
 		if (TCP_SKB_CB(skb)->tcp_flags & TCPHDR_FIN)
@@ -4413,7 +4422,7 @@ static int __must_check tcp_queue_rcv(struct sock *sk, struct sk_buff *skb, int
 	__skb_pull(skb, hdrlen);
 	eaten = (tail &&
 		 tcp_try_coalesce(sk, tail, skb, fragstolen)) ? 1 : 0;
-	tcp_sk(sk)->rcv_nxt = TCP_SKB_CB(skb)->end_seq;
+	tcp_rcv_nxt_update(tcp_sk(sk), TCP_SKB_CB(skb)->end_seq);
 	if (!eaten) {
 		__skb_queue_tail(&sk->sk_receive_queue, skb);
 		skb_set_owner_r(skb, sk);
@@ -4506,7 +4515,7 @@ queue_and_out:
 
 			eaten = tcp_queue_rcv(sk, skb, 0, &fragstolen);
 		}
-		tp->rcv_nxt = TCP_SKB_CB(skb)->end_seq;
+		tcp_rcv_nxt_update(tp, TCP_SKB_CB(skb)->end_seq);
 		if (skb->len)
 			tcp_event_data_recv(sk, skb);
 		if (TCP_SKB_CB(skb)->tcp_flags & TCPHDR_FIN)
@@ -5254,7 +5263,7 @@ void tcp_rcv_established(struct sock *sk, struct sk_buff *skb,
 					tcp_rcv_rtt_measure_ts(sk, skb);
 
 					__skb_pull(skb, tcp_header_len);
-					tp->rcv_nxt = TCP_SKB_CB(skb)->end_seq;
+					tcp_rcv_nxt_update(tp, TCP_SKB_CB(skb)->end_seq);
 					NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPHPHITSTOUSER);
 					eaten = 1;
 				}
-- 
cgit v1.2.3-55-g7522


From 9dac8835440622c2f84591673969d510ce198c11 Mon Sep 17 00:00:00 2001
From: Yuchung Cheng
Date: Wed, 29 Apr 2015 11:28:30 -0700
Subject: tcp: update reordering first before detecting loss

tcp_mark_lost_retrans is not used when FACK is disabled. Since
tcp_update_reordering may disable FACK, it should be called first
before tcp_mark_lost_retrans.

Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Nandita Dukkipati <nanditad@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
 net/ipv4/tcp_input.c | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

(limited to 'net/ipv4/tcp_input.c')

diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 7e6962bcfc30..bc790ea9960f 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -1820,14 +1820,12 @@ advance_sp:
 	for (j = 0; j < used_sacks; j++)
 		tp->recv_sack_cache[i++] = sp[j];
 
-	tcp_mark_lost_retrans(sk);
-
-	tcp_verify_left_out(tp);
-
 	if ((state.reord < tp->fackets_out) &&
 	    ((inet_csk(sk)->icsk_ca_state != TCP_CA_Loss) || tp->undo_marker))
 		tcp_update_reordering(sk, tp->fackets_out - state.reord, 0);
 
+	tcp_mark_lost_retrans(sk);
+	tcp_verify_left_out(tp);
 out:
 
 #if FASTRETRANS_DEBUG > 0
-- 
cgit v1.2.3-55-g7522


From 196da974758550a3933c8b0244ef98148df10552 Mon Sep 17 00:00:00 2001
From: Kenneth Klette Jonassen
Date: Fri, 1 May 2015 01:10:57 +0200
Subject: tcp: move struct tcp_sacktag_state to tcp_ack()

Later patch passes two values set in tcp_sacktag_one() to
tcp_clean_rtx_queue(). Prepare passing them via struct tcp_sacktag_state.

Acked-by: Yuchung Cheng <ycheng@google.com>
Cc: Eric Dumazet <edumazet@google.com>
Signed-off-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
 net/ipv4/tcp_input.c | 45 ++++++++++++++++++++++-----------------------
 1 file changed, 22 insertions(+), 23 deletions(-)

(limited to 'net/ipv4/tcp_input.c')

diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index bc790ea9960f..9902cf1abdf4 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -1634,7 +1634,7 @@ static int tcp_sack_cache_ok(const struct tcp_sock *tp, const struct tcp_sack_bl
 
 static int
 tcp_sacktag_write_queue(struct sock *sk, const struct sk_buff *ack_skb,
-			u32 prior_snd_una, long *sack_rtt_us)
+			u32 prior_snd_una, struct tcp_sacktag_state *state)
 {
 	struct tcp_sock *tp = tcp_sk(sk);
 	const unsigned char *ptr = (skb_transport_header(ack_skb) +
@@ -1642,7 +1642,6 @@ tcp_sacktag_write_queue(struct sock *sk, const struct sk_buff *ack_skb,
 	struct tcp_sack_block_wire *sp_wire = (struct tcp_sack_block_wire *)(ptr+2);
 	struct tcp_sack_block sp[TCP_NUM_SACKS];
 	struct tcp_sack_block *cache;
-	struct tcp_sacktag_state state;
 	struct sk_buff *skb;
 	int num_sacks = min(TCP_NUM_SACKS, (ptr[1] - TCPOLEN_SACK_BASE) >> 3);
 	int used_sacks;
@@ -1650,9 +1649,8 @@ tcp_sacktag_write_queue(struct sock *sk, const struct sk_buff *ack_skb,
 	int i, j;
 	int first_sack_index;
 
-	state.flag = 0;
-	state.reord = tp->packets_out;
-	state.rtt_us = -1L;
+	state->flag = 0;
+	state->reord = tp->packets_out;
 
 	if (!tp->sacked_out) {
 		if (WARN_ON(tp->fackets_out))
@@ -1663,7 +1661,7 @@ tcp_sacktag_write_queue(struct sock *sk, const struct sk_buff *ack_skb,
 	found_dup_sack = tcp_check_dsack(sk, ack_skb, sp_wire,
 					 num_sacks, prior_snd_una);
 	if (found_dup_sack)
-		state.flag |= FLAG_DSACKING_ACK;
+		state->flag |= FLAG_DSACKING_ACK;
 
 	/* Eliminate too old ACKs, but take into
 	 * account more or less fresh ones, they can
@@ -1728,7 +1726,7 @@ tcp_sacktag_write_queue(struct sock *sk, const struct sk_buff *ack_skb,
 	}
 
 	skb = tcp_write_queue_head(sk);
-	state.fack_count = 0;
+	state->fack_count = 0;
 	i = 0;
 
 	if (!tp->sacked_out) {
@@ -1762,10 +1760,10 @@ tcp_sacktag_write_queue(struct sock *sk, const struct sk_buff *ack_skb,
 
 			/* Head todo? */
 			if (before(start_seq, cache->start_seq)) {
-				skb = tcp_sacktag_skip(skb, sk, &state,
+				skb = tcp_sacktag_skip(skb, sk, state,
 						       start_seq);
 				skb = tcp_sacktag_walk(skb, sk, next_dup,
-						       &state,
+						       state,
 						       start_seq,
 						       cache->start_seq,
 						       dup_sack);
@@ -1776,7 +1774,7 @@ tcp_sacktag_write_queue(struct sock *sk, const struct sk_buff *ack_skb,
 				goto advance_sp;
 
 			skb = tcp_maybe_skipping_dsack(skb, sk, next_dup,
-						       &state,
+						       state,
 						       cache->end_seq);
 
 			/* ...tail remains todo... */
@@ -1785,12 +1783,12 @@ tcp_sacktag_write_queue(struct sock *sk, const struct sk_buff *ack_skb,
 				skb = tcp_highest_sack(sk);
 				if (!skb)
 					break;
-				state.fack_count = tp->fackets_out;
+				state->fack_count = tp->fackets_out;
 				cache++;
 				goto walk;
 			}
 
-			skb = tcp_sacktag_skip(skb, sk, &state, cache->end_seq);
+			skb = tcp_sacktag_skip(skb, sk, state, cache->end_seq);
 			/* Check overlap against next cached too (past this one already) */
 			cache++;
 			continue;
@@ -1800,12 +1798,12 @@ tcp_sacktag_write_queue(struct sock *sk, const struct sk_buff *ack_skb,
 			skb = tcp_highest_sack(sk);
 			if (!skb)
 				break;
-			state.fack_count = tp->fackets_out;
+			state->fack_count = tp->fackets_out;
 		}
-		skb = tcp_sacktag_skip(skb, sk, &state, start_seq);
+		skb = tcp_sacktag_skip(skb, sk, state, start_seq);
 
 walk:
-		skb = tcp_sacktag_walk(skb, sk, next_dup, &state,
+		skb = tcp_sacktag_walk(skb, sk, next_dup, state,
 				       start_seq, end_seq, dup_sack);
 
 advance_sp:
@@ -1820,9 +1818,9 @@ advance_sp:
 	for (j = 0; j < used_sacks; j++)
 		tp->recv_sack_cache[i++] = sp[j];
 
-	if ((state.reord < tp->fackets_out) &&
+	if ((state->reord < tp->fackets_out) &&
 	    ((inet_csk(sk)->icsk_ca_state != TCP_CA_Loss) || tp->undo_marker))
-		tcp_update_reordering(sk, tp->fackets_out - state.reord, 0);
+		tcp_update_reordering(sk, tp->fackets_out - state->reord, 0);
 
 	tcp_mark_lost_retrans(sk);
 	tcp_verify_left_out(tp);
@@ -1834,8 +1832,7 @@ out:
 	WARN_ON((int)tp->retrans_out < 0);
 	WARN_ON((int)tcp_packets_in_flight(tp) < 0);
 #endif
-	*sack_rtt_us = state.rtt_us;
-	return state.flag;
+	return state->flag;
 }
 
 /* Limits sacked_out so that sum with lost_out isn't ever larger than
@@ -3459,6 +3456,7 @@ static int tcp_ack(struct sock *sk, const struct sk_buff *skb, int flag)
 {
 	struct inet_connection_sock *icsk = inet_csk(sk);
 	struct tcp_sock *tp = tcp_sk(sk);
+	struct tcp_sacktag_state sack_state;
 	u32 prior_snd_una = tp->snd_una;
 	u32 ack_seq = TCP_SKB_CB(skb)->seq;
 	u32 ack = TCP_SKB_CB(skb)->ack_seq;
@@ -3467,7 +3465,8 @@ static int tcp_ack(struct sock *sk, const struct sk_buff *skb, int flag)
 	int prior_packets = tp->packets_out;
 	const int prior_unsacked = tp->packets_out - tp->sacked_out;
 	int acked = 0; /* Number of packets newly acked */
-	long sack_rtt_us = -1L;
+
+	sack_state.rtt_us = -1L;
 
 	/* We very likely will need to access write queue head. */
 	prefetchw(sk->sk_write_queue.next);
@@ -3531,7 +3530,7 @@ static int tcp_ack(struct sock *sk, const struct sk_buff *skb, int flag)
 
 		if (TCP_SKB_CB(skb)->sacked)
 			flag |= tcp_sacktag_write_queue(sk, skb, prior_snd_una,
-							&sack_rtt_us);
+							&sack_state);
 
 		if (tcp_ecn_rcv_ecn_echo(tp, tcp_hdr(skb))) {
 			flag |= FLAG_ECE;
@@ -3556,7 +3555,7 @@ static int tcp_ack(struct sock *sk, const struct sk_buff *skb, int flag)
 	/* See if we can take anything off of the retransmit queue. */
 	acked = tp->packets_out;
 	flag |= tcp_clean_rtx_queue(sk, prior_fackets, prior_snd_una,
-				    sack_rtt_us);
+				    sack_state.rtt_us);
 	acked -= tp->packets_out;
 
 	/* Advance cwnd if state allows */
@@ -3608,7 +3607,7 @@ old_ack:
 	 */
 	if (TCP_SKB_CB(skb)->sacked) {
 		flag |= tcp_sacktag_write_queue(sk, skb, prior_snd_una,
-						&sack_rtt_us);
+						&sack_state);
 		tcp_fastretrans_alert(sk, acked, prior_unsacked,
 				      is_dupack, flag);
 	}
-- 
cgit v1.2.3-55-g7522


From 31231a8a873026410eab438c5757430546a517d1 Mon Sep 17 00:00:00 2001
From: Kenneth Klette Jonassen
Date: Fri, 1 May 2015 01:10:58 +0200
Subject: tcp: improve RTT from SACK for CC

tcp_sacktag_one() always picks the earliest sequence SACKed for RTT.
This might not make sense for congestion control in cases where:

  1. ACKs are lost, i.e. a SACK following a lost SACK covers both
     new and old segments at the receiver.
  2. The receiver disregards the RFC 5681 recommendation to immediately
     ACK out-of-order segments.

Give congestion control a RTT for the latest segment SACKed, which is the
most accurate RTT estimate, but preserve the conservative RTT for RTO.

Removes the call to skb_mstamp_get() in tcp_sacktag_one().

Cc: Yuchung Cheng <ycheng@google.com>
Cc: Eric Dumazet <edumazet@google.com>
Signed-off-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no>
Acked-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
 net/ipv4/tcp_input.c | 40 ++++++++++++++++++++++------------------
 1 file changed, 22 insertions(+), 18 deletions(-)

(limited to 'net/ipv4/tcp_input.c')

diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 9902cf1abdf4..f563d2a88809 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -1130,7 +1130,12 @@ static bool tcp_check_dsack(struct sock *sk, const struct sk_buff *ack_skb,
 struct tcp_sacktag_state {
 	int	reord;
 	int	fack_count;
-	long	rtt_us; /* RTT measured by SACKing never-retransmitted data */
+	/* Timestamps for earliest and latest never-retransmitted segment
+	 * that was SACKed. RTO needs the earliest RTT to stay conservative,
+	 * but congestion control should still get an accurate delay signal.
+	 */
+	struct skb_mstamp first_sackt;
+	struct skb_mstamp last_sackt;
 	int	flag;
 };
 
@@ -1233,14 +1238,9 @@ static u8 tcp_sacktag_one(struct sock *sk,
 							   state->reord);
 				if (!after(end_seq, tp->high_seq))
 					state->flag |= FLAG_ORIG_SACK_ACKED;
-				/* Pick the earliest sequence sacked for RTT */
-				if (state->rtt_us < 0) {
-					struct skb_mstamp now;
-
-					skb_mstamp_get(&now);
-					state->rtt_us = skb_mstamp_us_delta(&now,
-								xmit_time);
-				}
+				if (state->first_sackt.v64 == 0)
+					state->first_sackt = *xmit_time;
+				state->last_sackt = *xmit_time;
 			}
 
 			if (sacked & TCPCB_LOST) {
@@ -3049,7 +3049,8 @@ static void tcp_ack_tstamp(struct sock *sk, struct sk_buff *skb,
  * arrived at the other end.
  */
 static int tcp_clean_rtx_queue(struct sock *sk, int prior_fackets,
-			       u32 prior_snd_una, long sack_rtt_us)
+			       u32 prior_snd_una,
+			       struct tcp_sacktag_state *sack)
 {
 	const struct inet_connection_sock *icsk = inet_csk(sk);
 	struct skb_mstamp first_ackt, last_ackt, now;
@@ -3057,8 +3058,9 @@ static int tcp_clean_rtx_queue(struct sock *sk, int prior_fackets,
 	u32 prior_sacked = tp->sacked_out;
 	u32 reord = tp->packets_out;
 	bool fully_acked = true;
-	long ca_seq_rtt_us = -1L;
+	long sack_rtt_us = -1L;
 	long seq_rtt_us = -1L;
+	long ca_rtt_us = -1L;
 	struct sk_buff *skb;
 	u32 pkts_acked = 0;
 	bool rtt_update;
@@ -3147,7 +3149,11 @@ static int tcp_clean_rtx_queue(struct sock *sk, int prior_fackets,
 	skb_mstamp_get(&now);
 	if (likely(first_ackt.v64)) {
 		seq_rtt_us = skb_mstamp_us_delta(&now, &first_ackt);
-		ca_seq_rtt_us = skb_mstamp_us_delta(&now, &last_ackt);
+		ca_rtt_us = skb_mstamp_us_delta(&now, &last_ackt);
+	}
+	if (sack->first_sackt.v64) {
+		sack_rtt_us = skb_mstamp_us_delta(&now, &sack->first_sackt);
+		ca_rtt_us = skb_mstamp_us_delta(&now, &sack->last_sackt);
 	}
 
 	rtt_update = tcp_ack_update_rtt(sk, flag, seq_rtt_us, sack_rtt_us);
@@ -3178,10 +3184,8 @@ static int tcp_clean_rtx_queue(struct sock *sk, int prior_fackets,
 
 		tp->fackets_out -= min(pkts_acked, tp->fackets_out);
 
-		if (ca_ops->pkts_acked) {
-			long rtt_us = min_t(ulong, ca_seq_rtt_us, sack_rtt_us);
-			ca_ops->pkts_acked(sk, pkts_acked, rtt_us);
-		}
+		if (ca_ops->pkts_acked)
+			ca_ops->pkts_acked(sk, pkts_acked, ca_rtt_us);
 
 	} else if (skb && rtt_update && sack_rtt_us >= 0 &&
 		   sack_rtt_us > skb_mstamp_us_delta(&now, &skb->skb_mstamp)) {
@@ -3466,7 +3470,7 @@ static int tcp_ack(struct sock *sk, const struct sk_buff *skb, int flag)
 	const int prior_unsacked = tp->packets_out - tp->sacked_out;
 	int acked = 0; /* Number of packets newly acked */
 
-	sack_state.rtt_us = -1L;
+	sack_state.first_sackt.v64 = 0;
 
 	/* We very likely will need to access write queue head. */
 	prefetchw(sk->sk_write_queue.next);
@@ -3555,7 +3559,7 @@ static int tcp_ack(struct sock *sk, const struct sk_buff *skb, int flag)
 	/* See if we can take anything off of the retransmit queue. */
 	acked = tp->packets_out;
 	flag |= tcp_clean_rtx_queue(sk, prior_fackets, prior_snd_una,
-				    sack_state.rtt_us);
+				    &sack_state);
 	acked -= tp->packets_out;
 
 	/* Advance cwnd if state allows */
-- 
cgit v1.2.3-55-g7522


From 138998fdd12e7362756e158d00856a2aabd5f0c1 Mon Sep 17 00:00:00 2001
From: Kenneth Klette Jonassen
Date: Fri, 1 May 2015 01:10:59 +0200
Subject: tcp: invoke pkts_acked hook on every ACK

Invoking pkts_acked is currently conditioned on FLAG_ACKED:
receiving a cumulative ACK of new data, or ACK with SYN flag set.

Remove this condition so that CC may get RTT measurements from all SACKs.

Cc: Yuchung Cheng <ycheng@google.com>
Cc: Eric Dumazet <edumazet@google.com>
Cc: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Kenneth Klette Jonassen <kennetkl@ifi.uio.no>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
 net/ipv4/tcp_input.c | 9 +++------
 1 file changed, 3 insertions(+), 6 deletions(-)

(limited to 'net/ipv4/tcp_input.c')

diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index f563d2a88809..09bdc4abfcbb 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -3159,9 +3159,6 @@ static int tcp_clean_rtx_queue(struct sock *sk, int prior_fackets,
 	rtt_update = tcp_ack_update_rtt(sk, flag, seq_rtt_us, sack_rtt_us);
 
 	if (flag & FLAG_ACKED) {
-		const struct tcp_congestion_ops *ca_ops
-			= inet_csk(sk)->icsk_ca_ops;
-
 		tcp_rearm_rto(sk);
 		if (unlikely(icsk->icsk_mtup.probe_size &&
 			     !after(tp->mtu_probe.probe_seq_end, tp->snd_una))) {
@@ -3184,9 +3181,6 @@ static int tcp_clean_rtx_queue(struct sock *sk, int prior_fackets,
 
 		tp->fackets_out -= min(pkts_acked, tp->fackets_out);
 
-		if (ca_ops->pkts_acked)
-			ca_ops->pkts_acked(sk, pkts_acked, ca_rtt_us);
-
 	} else if (skb && rtt_update && sack_rtt_us >= 0 &&
 		   sack_rtt_us > skb_mstamp_us_delta(&now, &skb->skb_mstamp)) {
 		/* Do not re-arm RTO if the sack RTT is measured from data sent
@@ -3196,6 +3190,9 @@ static int tcp_clean_rtx_queue(struct sock *sk, int prior_fackets,
 		tcp_rearm_rto(sk);
 	}
 
+	if (icsk->icsk_ca_ops->pkts_acked)
+		icsk->icsk_ca_ops->pkts_acked(sk, pkts_acked, ca_rtt_us);
+
 #if FASTRETRANS_DEBUG > 0
 	WARN_ON((int)tp->sacked_out < 0);
 	WARN_ON((int)tp->lost_out < 0);
-- 
cgit v1.2.3-55-g7522


From cd8ae85299d54155702a56811b2e035e63064d3d Mon Sep 17 00:00:00 2001
From: Eric Dumazet
Date: Sun, 3 May 2015 21:34:46 -0700
Subject: tcp: provide SYN headers for passive connections

This patch allows a server application to get the TCP SYN headers for
its passive connections.  This is useful if the server is doing
fingerprinting of clients based on SYN packet contents.

Two socket options are added: TCP_SAVE_SYN and TCP_SAVED_SYN.

The first is used on a socket to enable saving the SYN headers
for child connections. This can be set before or after the listen()
call.

The latter is used to retrieve the SYN headers for passive connections,
if the parent listener has enabled TCP_SAVE_SYN.

TCP_SAVED_SYN is read once, it frees the saved SYN headers.

The data returned in TCP_SAVED_SYN are network (IPv4/IPv6) and TCP
headers.

Original patch was written by Tom Herbert, I changed it to not hold
a full skb (and associated dst and conntracking reference).

We have used such patch for about 3 years at Google.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Tested-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
 include/linux/tcp.h        |  8 ++++++++
 include/net/request_sock.h |  4 +++-
 include/uapi/linux/tcp.h   |  2 ++
 net/ipv4/tcp.c             | 35 +++++++++++++++++++++++++++++++++++
 net/ipv4/tcp_input.c       | 18 ++++++++++++++++++
 net/ipv4/tcp_ipv4.c        |  1 +
 net/ipv4/tcp_minisocks.c   |  3 +++
 7 files changed, 70 insertions(+), 1 deletion(-)

(limited to 'net/ipv4/tcp_input.c')

diff --git a/include/linux/tcp.h b/include/linux/tcp.h
index 3b2911502a8c..e6fb5df22db1 100644
--- a/include/linux/tcp.h
+++ b/include/linux/tcp.h
@@ -199,6 +199,7 @@ struct tcp_sock {
 		syn_fastopen:1,	/* SYN includes Fast Open option */
 		syn_fastopen_exp:1,/* SYN includes Fast Open exp. option */
 		syn_data_acked:1,/* data in SYN is acked by SYN-ACK */
+		save_syn:1,	/* Save headers of SYN packet */
 		is_cwnd_limited:1;/* forward progress limited by snd_cwnd? */
 	u32	tlp_high_seq;	/* snd_nxt at the time of TLP retransmit. */
 
@@ -326,6 +327,7 @@ struct tcp_sock {
 	 * socket. Used to retransmit SYNACKs etc.
 	 */
 	struct request_sock *fastopen_rsk;
+	u32	*saved_syn;
 };
 
 enum tsq_flags {
@@ -393,4 +395,10 @@ static inline int fastopen_init_queue(struct sock *sk, int backlog)
 	return 0;
 }
 
+static inline void tcp_saved_syn_free(struct tcp_sock *tp)
+{
+	kfree(tp->saved_syn);
+	tp->saved_syn = NULL;
+}
+
 #endif	/* _LINUX_TCP_H */
diff --git a/include/net/request_sock.h b/include/net/request_sock.h
index 9f4265ce8892..87935cad2f7b 100644
--- a/include/net/request_sock.h
+++ b/include/net/request_sock.h
@@ -64,6 +64,7 @@ struct request_sock {
 	struct timer_list		rsk_timer;
 	const struct request_sock_ops	*rsk_ops;
 	struct sock			*sk;
+	u32				*saved_syn;
 	u32				secid;
 	u32				peer_secid;
 };
@@ -77,7 +78,7 @@ reqsk_alloc(const struct request_sock_ops *ops, struct sock *sk_listener)
 		req->rsk_ops = ops;
 		sock_hold(sk_listener);
 		req->rsk_listener = sk_listener;
-
+		req->saved_syn = NULL;
 		/* Following is temporary. It is coupled with debugging
 		 * helpers in reqsk_put() & reqsk_free()
 		 */
@@ -104,6 +105,7 @@ static inline void reqsk_free(struct request_sock *req)
 	req->rsk_ops->destructor(req);
 	if (req->rsk_listener)
 		sock_put(req->rsk_listener);
+	kfree(req->saved_syn);
 	kmem_cache_free(req->rsk_ops->slab, req);
 }
 
diff --git a/include/uapi/linux/tcp.h b/include/uapi/linux/tcp.h
index faa72f4fa547..51ebedba577f 100644
--- a/include/uapi/linux/tcp.h
+++ b/include/uapi/linux/tcp.h
@@ -113,6 +113,8 @@ enum {
 #define TCP_TIMESTAMP		24
 #define TCP_NOTSENT_LOWAT	25	/* limit number of unsent bytes in write queue */
 #define TCP_CC_INFO		26	/* Get Congestion Control (optional) info */
+#define TCP_SAVE_SYN		27	/* Record SYN headers for new connections */
+#define TCP_SAVED_SYN		28	/* Get SYN headers recorded for connection */
 
 struct tcp_repair_opt {
 	__u32	opt_code;
diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
index 46efa03d2b11..ecccfdc50d76 100644
--- a/net/ipv4/tcp.c
+++ b/net/ipv4/tcp.c
@@ -2482,6 +2482,13 @@ static int do_tcp_setsockopt(struct sock *sk, int level,
 			icsk->icsk_syn_retries = val;
 		break;
 
+	case TCP_SAVE_SYN:
+		if (val < 0 || val > 1)
+			err = -EINVAL;
+		else
+			tp->save_syn = val;
+		break;
+
 	case TCP_LINGER2:
 		if (val < 0)
 			tp->linger2 = -1;
@@ -2818,6 +2825,34 @@ static int do_tcp_getsockopt(struct sock *sk, int level,
 	case TCP_NOTSENT_LOWAT:
 		val = tp->notsent_lowat;
 		break;
+	case TCP_SAVE_SYN:
+		val = tp->save_syn;
+		break;
+	case TCP_SAVED_SYN: {
+		if (get_user(len, optlen))
+			return -EFAULT;
+
+		lock_sock(sk);
+		if (tp->saved_syn) {
+			len = min_t(unsigned int, tp->saved_syn[0], len);
+			if (put_user(len, optlen)) {
+				release_sock(sk);
+				return -EFAULT;
+			}
+			if (copy_to_user(optval, tp->saved_syn + 1, len)) {
+				release_sock(sk);
+				return -EFAULT;
+			}
+			tcp_saved_syn_free(tp);
+			release_sock(sk);
+		} else {
+			release_sock(sk);
+			len = 0;
+			if (put_user(len, optlen))
+				return -EFAULT;
+		}
+		return 0;
+	}
 	default:
 		return -ENOPROTOOPT;
 	}
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 09bdc4abfcbb..df2ca615cd0c 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -6060,6 +6060,23 @@ static bool tcp_syn_flood_action(struct sock *sk,
 	return want_cookie;
 }
 
+static void tcp_reqsk_record_syn(const struct sock *sk,
+				 struct request_sock *req,
+				 const struct sk_buff *skb)
+{
+	if (tcp_sk(sk)->save_syn) {
+		u32 len = skb_network_header_len(skb) + tcp_hdrlen(skb);
+		u32 *copy;
+
+		copy = kmalloc(len + sizeof(u32), GFP_ATOMIC);
+		if (copy) {
+			copy[0] = len;
+			memcpy(&copy[1], skb_network_header(skb), len);
+			req->saved_syn = copy;
+		}
+	}
+}
+
 int tcp_conn_request(struct request_sock_ops *rsk_ops,
 		     const struct tcp_request_sock_ops *af_ops,
 		     struct sock *sk, struct sk_buff *skb)
@@ -6192,6 +6209,7 @@ int tcp_conn_request(struct request_sock_ops *rsk_ops,
 		tcp_rsk(req)->tfo_listener = false;
 		af_ops->queue_hash_add(sk, req, TCP_TIMEOUT_INIT);
 	}
+	tcp_reqsk_record_syn(sk, req, skb);
 
 	return 0;
 
diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c
index fc1c658ec6c1..91cb4768a860 100644
--- a/net/ipv4/tcp_ipv4.c
+++ b/net/ipv4/tcp_ipv4.c
@@ -1802,6 +1802,7 @@ void tcp_v4_destroy_sock(struct sock *sk)
 
 	/* If socket is aborted during connect operation */
 	tcp_free_fastopen_req(tp);
+	tcp_saved_syn_free(tp);
 
 	sk_sockets_allocated_dec(sk);
 	sock_release_memcg(sk);
diff --git a/net/ipv4/tcp_minisocks.c b/net/ipv4/tcp_minisocks.c
index e5d7649136fc..ebe2ab2596ed 100644
--- a/net/ipv4/tcp_minisocks.c
+++ b/net/ipv4/tcp_minisocks.c
@@ -536,6 +536,9 @@ struct sock *tcp_create_openreq_child(struct sock *sk, struct request_sock *req,
 		newtp->fastopen_rsk = NULL;
 		newtp->syn_data_acked = 0;
 
+		newtp->saved_syn = req->saved_syn;
+		req->saved_syn = NULL;
+
 		TCP_INC_STATS_BH(sock_net(sk), TCP_MIB_PASSIVEOPENS);
 	}
 	return newsk;
-- 
cgit v1.2.3-55-g7522


From 21c8fe9915276d923f8c1e43434fd6d37a3b9aef Mon Sep 17 00:00:00 2001
From: Eric Dumazet
Date: Wed, 6 May 2015 14:26:24 -0700
Subject: tcp: adjust window probe timers to safer values

With the advent of small rto timers in datacenter TCP,
(ip route ... rto_min x), the following can happen :

1) Qdisc is full, transmit fails.

   TCP sets a timer based on icsk_rto to retry the transmit, without
   exponential backoff.
   With low icsk_rto, and lot of sockets, all cpus are servicing timer
   interrupts like crazy.
   Intent of the code was to retry with a timer between 200 (TCP_RTO_MIN)
   and 500ms (TCP_RESOURCE_PROBE_INTERVAL)

2) Receivers can send zero windows if they don't drain their receive queue.

   TCP sends zero window probes, based on icsk_rto current value, with
   exponential backoff.
   With /proc/sys/net/ipv4/tcp_retries2 being 15 (or even smaller in
   some cases), sender can abort in less than one or two minutes !
   If receiver stops the sender, it obviously doesn't care of very tight
   rto. Probability of dropping the ACK reopening the window is not
   worth the risk.

Lets change the base timer to be at least 200ms (TCP_RTO_MIN) for these
events (but not normal RTO based retransmits)

A followup patch adds a new SNMP counter, as it would have helped a lot
diagnosing this issue.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: Yuchung Cheng <ycheng@google.com>
Acked-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
 include/net/tcp.h     | 27 ++++++++++++++++++++++-----
 net/ipv4/tcp_input.c  |  2 +-
 net/ipv4/tcp_output.c |  2 +-
 3 files changed, 24 insertions(+), 7 deletions(-)

(limited to 'net/ipv4/tcp_input.c')

diff --git a/include/net/tcp.h b/include/net/tcp.h
index 6d204f3f9df8..7a2248a35b13 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -1043,14 +1043,31 @@ static inline bool tcp_is_cwnd_limited(const struct sock *sk)
 	return tp->is_cwnd_limited;
 }
 
-static inline void tcp_check_probe_timer(struct sock *sk)
+/* Something is really bad, we could not queue an additional packet,
+ * because qdisc is full or receiver sent a 0 window.
+ * We do not want to add fuel to the fire, or abort too early,
+ * so make sure the timer we arm now is at least 200ms in the future,
+ * regardless of current icsk_rto value (as it could be ~2ms)
+ */
+static inline unsigned long tcp_probe0_base(const struct sock *sk)
 {
-	const struct tcp_sock *tp = tcp_sk(sk);
-	const struct inet_connection_sock *icsk = inet_csk(sk);
+	return max_t(unsigned long, inet_csk(sk)->icsk_rto, TCP_RTO_MIN);
+}
 
-	if (!tp->packets_out && !icsk->icsk_pending)
+/* Variant of inet_csk_rto_backoff() used for zero window probes */
+static inline unsigned long tcp_probe0_when(const struct sock *sk,
+					    unsigned long max_when)
+{
+	u64 when = (u64)tcp_probe0_base(sk) << inet_csk(sk)->icsk_backoff;
+
+	return (unsigned long)min_t(u64, when, max_when);
+}
+
+static inline void tcp_check_probe_timer(struct sock *sk)
+{
+	if (!tcp_sk(sk)->packets_out && !inet_csk(sk)->icsk_pending)
 		inet_csk_reset_xmit_timer(sk, ICSK_TIME_PROBE0,
-					  icsk->icsk_rto, TCP_RTO_MAX);
+					  tcp_probe0_base(sk), TCP_RTO_MAX);
 }
 
 static inline void tcp_init_wl(struct tcp_sock *tp, u32 seq)
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index df2ca615cd0c..cf8b20ff6658 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -3233,7 +3233,7 @@ static void tcp_ack_probe(struct sock *sk)
 		 * This function is not for random using!
 		 */
 	} else {
-		unsigned long when = inet_csk_rto_backoff(icsk, TCP_RTO_MAX);
+		unsigned long when = tcp_probe0_when(sk, TCP_RTO_MAX);
 
 		inet_csk_reset_xmit_timer(sk, ICSK_TIME_PROBE0,
 					  when, TCP_RTO_MAX);
diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
index a369e8a70b2c..b76c719e1979 100644
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -3490,7 +3490,7 @@ void tcp_send_probe0(struct sock *sk)
 		probe_max = TCP_RESOURCE_PROBE_INTERVAL;
 	}
 	inet_csk_reset_xmit_timer(sk, ICSK_TIME_PROBE0,
-				  inet_csk_rto_backoff(icsk, probe_max),
+				  tcp_probe0_when(sk, probe_max),
 				  TCP_RTO_MAX);
 }
 
-- 
cgit v1.2.3-55-g7522


From b8da51ebb1aa93908350f95efae73aecbc2e266c Mon Sep 17 00:00:00 2001
From: Eric Dumazet
Date: Fri, 15 May 2015 12:39:27 -0700
Subject: tcp: introduce tcp_under_memory_pressure()

Introduce an optimized version of sk_under_memory_pressure()
for TCP. Our intent is to use it in fast paths.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
 include/net/tcp.h     | 8 ++++++++
 net/ipv4/tcp_input.c  | 8 ++++----
 net/ipv4/tcp_output.c | 4 ++--
 net/ipv4/tcp_timer.c  | 2 +-
 4 files changed, 15 insertions(+), 7 deletions(-)

(limited to 'net/ipv4/tcp_input.c')

diff --git a/include/net/tcp.h b/include/net/tcp.h
index 841691a296dc..0d85223efa4c 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -286,6 +286,14 @@ extern atomic_long_t tcp_memory_allocated;
 extern struct percpu_counter tcp_sockets_allocated;
 extern int tcp_memory_pressure;
 
+/* optimized version of sk_under_memory_pressure() for TCP sockets */
+static inline bool tcp_under_memory_pressure(const struct sock *sk)
+{
+	if (mem_cgroup_sockets_enabled && sk->sk_cgrp)
+		return !!sk->sk_cgrp->memory_pressure;
+
+	return tcp_memory_pressure;
+}
 /*
  * The next routines deal with comparing 32 bit unsigned ints
  * and worry about wraparound (automatic with unsigned arithmetic).
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index cf8b20ff6658..093779f7e893 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -359,7 +359,7 @@ static void tcp_grow_window(struct sock *sk, const struct sk_buff *skb)
 	/* Check #1 */
 	if (tp->rcv_ssthresh < tp->window_clamp &&
 	    (int)tp->rcv_ssthresh < tcp_space(sk) &&
-	    !sk_under_memory_pressure(sk)) {
+	    !tcp_under_memory_pressure(sk)) {
 		int incr;
 
 		/* Check #2. Increase window, if skb with such overhead
@@ -446,7 +446,7 @@ static void tcp_clamp_window(struct sock *sk)
 
 	if (sk->sk_rcvbuf < sysctl_tcp_rmem[2] &&
 	    !(sk->sk_userlocks & SOCK_RCVBUF_LOCK) &&
-	    !sk_under_memory_pressure(sk) &&
+	    !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),
 				    sysctl_tcp_rmem[2]);
@@ -4781,7 +4781,7 @@ static int tcp_prune_queue(struct sock *sk)
 
 	if (atomic_read(&sk->sk_rmem_alloc) >= sk->sk_rcvbuf)
 		tcp_clamp_window(sk);
-	else if (sk_under_memory_pressure(sk))
+	else if (tcp_under_memory_pressure(sk))
 		tp->rcv_ssthresh = min(tp->rcv_ssthresh, 4U * tp->advmss);
 
 	tcp_collapse_ofo_queue(sk);
@@ -4825,7 +4825,7 @@ static bool tcp_should_expand_sndbuf(const struct sock *sk)
 		return false;
 
 	/* If we are under global TCP memory pressure, do not expand.  */
-	if (sk_under_memory_pressure(sk))
+	if (tcp_under_memory_pressure(sk))
 		return false;
 
 	/* If we are under soft global TCP memory pressure, do not expand.  */
diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
index bac1a950d087..08c2cc40b26d 100644
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -2392,7 +2392,7 @@ u32 __tcp_select_window(struct sock *sk)
 	if (free_space < (full_space >> 1)) {
 		icsk->icsk_ack.quick = 0;
 
-		if (sk_under_memory_pressure(sk))
+		if (tcp_under_memory_pressure(sk))
 			tp->rcv_ssthresh = min(tp->rcv_ssthresh,
 					       4U * tp->advmss);
 
@@ -2843,7 +2843,7 @@ void tcp_send_fin(struct sock *sk)
 	 * Note: in the latter case, FIN packet will be sent after a timeout,
 	 * as TCP stack thinks it has already been transmitted.
 	 */
-	if (tskb && (tcp_send_head(sk) || sk_under_memory_pressure(sk))) {
+	if (tskb && (tcp_send_head(sk) || tcp_under_memory_pressure(sk))) {
 coalesce:
 		TCP_SKB_CB(tskb)->tcp_flags |= TCPHDR_FIN;
 		TCP_SKB_CB(tskb)->end_seq++;
diff --git a/net/ipv4/tcp_timer.c b/net/ipv4/tcp_timer.c
index 65bf670e8714..5b752f58a900 100644
--- a/net/ipv4/tcp_timer.c
+++ b/net/ipv4/tcp_timer.c
@@ -247,7 +247,7 @@ void tcp_delack_timer_handler(struct sock *sk)
 	}
 
 out:
-	if (sk_under_memory_pressure(sk))
+	if (tcp_under_memory_pressure(sk))
 		sk_mem_reclaim(sk);
 }
 
-- 
cgit v1.2.3-55-g7522


From 76dfa6082032b5c179864816fa508879421678eb Mon Sep 17 00:00:00 2001
From: Eric Dumazet
Date: Fri, 15 May 2015 12:39:29 -0700
Subject: tcp: allow one skb to be received per socket under memory pressure

While testing tight tcp_mem settings, I found tcp sessions could be
stuck because we do not allow even one skb to be received on them.

By allowing one skb to be received, we introduce fairness and
eventuallu force memory hogs to release their allocation.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
 net/ipv4/tcp_input.c | 10 ++++++----
 1 file changed, 6 insertions(+), 4 deletions(-)

(limited to 'net/ipv4/tcp_input.c')

diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 093779f7e893..40c435997e54 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -4507,10 +4507,12 @@ static void tcp_data_queue(struct sock *sk, struct sk_buff *skb)
 
 		if (eaten <= 0) {
 queue_and_out:
-			if (eaten < 0 &&
-			    tcp_try_rmem_schedule(sk, skb, skb->truesize))
-				goto drop;
-
+			if (eaten < 0) {
+				if (skb_queue_len(&sk->sk_receive_queue) == 0)
+					sk_forced_mem_schedule(sk, skb->truesize);
+				else if (tcp_try_rmem_schedule(sk, skb, skb->truesize))
+					goto drop;
+			}
 			eaten = tcp_queue_rcv(sk, skb, 0, &fragstolen);
 		}
 		tcp_rcv_nxt_update(tp, TCP_SKB_CB(skb)->end_seq);
-- 
cgit v1.2.3-55-g7522


From da34ac7626b571d262f92b93f11eb32dd58d9c4e Mon Sep 17 00:00:00 2001
From: Yuchung Cheng
Date: Mon, 18 May 2015 12:31:44 -0700
Subject: tcp: only undo on partial ACKs in CA_Loss

Undo based on TCP timestamps should only happen on ACKs that advance
SND_UNA, according to the Eifel algorithm in RFC 3522:

Section 3.2:

  (4) If the value of the Timestamp Echo Reply field of the
      acceptable ACK's Timestamps option is smaller than the
      value of RetransmitTS, then proceed to step (5),

Section Terminology:
   We use the term 'acceptable ACK' as defined in [RFC793].  That is an
   ACK that acknowledges previously unacknowledged data.

This is because upon receiving an out-of-order packet, the receiver
returns the last timestamp that advances RCV_NXT, not the current
timestamp of the packet in the DUPACK. Without checking the flag,
the DUPACK will cause tcp_packet_delayed() to return true and
tcp_try_undo_loss() will revert cwnd reduction.

Note that we check the condition in CA_Recovery already by only
calling tcp_try_undo_partial() if FLAG_SND_UNA_ADVANCED is set or
tcp_try_undo_recovery() if snd_una crosses high_seq.

Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
 net/ipv4/tcp_input.c | 9 ++++++---
 1 file changed, 6 insertions(+), 3 deletions(-)

(limited to 'net/ipv4/tcp_input.c')

diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index bc790ea9960f..9faf775a8c4a 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -2698,11 +2698,16 @@ static void tcp_process_loss(struct sock *sk, int flag, bool is_dupack)
 	struct tcp_sock *tp = tcp_sk(sk);
 	bool recovered = !before(tp->snd_una, tp->high_seq);
 
+	if ((flag & FLAG_SND_UNA_ADVANCED) &&
+	    tcp_try_undo_loss(sk, false))
+		return;
+
 	if (tp->frto) { /* F-RTO RFC5682 sec 3.1 (sack enhanced version). */
 		/* Step 3.b. A timeout is spurious if not all data are
 		 * lost, i.e., never-retransmitted data are (s)acked.
 		 */
-		if (tcp_try_undo_loss(sk, flag & FLAG_ORIG_SACK_ACKED))
+		if ((flag & FLAG_ORIG_SACK_ACKED) &&
+		    tcp_try_undo_loss(sk, true))
 			return;
 
 		if (after(tp->snd_nxt, tp->high_seq) &&
@@ -2732,8 +2737,6 @@ static void tcp_process_loss(struct sock *sk, int flag, bool is_dupack)
 		else if (flag & FLAG_SND_UNA_ADVANCED)
 			tcp_reset_reno_sack(tp);
 	}
-	if (tcp_try_undo_loss(sk, false))
-		return;
 	tcp_xmit_retransmit_queue(sk);
 }
 
-- 
cgit v1.2.3-55-g7522


From b7b0ed910cd8450db6d98cd4361c644bb1c88412 Mon Sep 17 00:00:00 2001
From: Yuchung Cheng
Date: Mon, 18 May 2015 12:31:45 -0700
Subject: tcp: don't over-send F-RTO probes

After sending the new data packets to probe (step 2), F-RTO may
incorrectly send more probes if the next ACK advances SND_UNA and
does not sack new packet. However F-RTO RFC 5682 probes at most
once. This bug may cause sender to always send new data instead of
repairing holes, inducing longer HoL blocking on the receiver for
the application.

Signed-off-by: Yuchung Cheng <ycheng@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
 net/ipv4/tcp_input.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

(limited to 'net/ipv4/tcp_input.c')

diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 9faf775a8c4a..243d674b3ef5 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -2710,9 +2710,9 @@ static void tcp_process_loss(struct sock *sk, int flag, bool is_dupack)
 		    tcp_try_undo_loss(sk, true))
 			return;
 
-		if (after(tp->snd_nxt, tp->high_seq) &&
-		    (flag & FLAG_DATA_SACKED || is_dupack)) {
-			tp->frto = 0; /* Loss was real: 2nd part of step 3.a */
+		if (after(tp->snd_nxt, tp->high_seq)) {
+			if (flag & FLAG_DATA_SACKED || is_dupack)
+				tp->frto = 0; /* Step 3.a. loss was real */
 		} else if (flag & FLAG_SND_UNA_ADVANCED && !recovered) {
 			tp->high_seq = tp->snd_nxt;
 			__tcp_push_pending_frames(sk, tcp_current_mss(sk),
-- 
cgit v1.2.3-55-g7522


From d654976cbf852ee20612ee10dbe57cdacda9f452 Mon Sep 17 00:00:00 2001
From: Eric Dumazet
Date: Thu, 21 May 2015 21:51:19 -0700
Subject: tcp: fix a potential deadlock in tcp_get_info()

Taking socket spinlock in tcp_get_info() can deadlock, as
inet_diag_dump_icsk() holds the &hashinfo->ehash_locks[i],
while packet processing can use the reverse locking order.

We could avoid this locking for TCP_LISTEN states, but lockdep would
certainly get confused as all TCP sockets share same lockdep classes.

[  523.722504] ======================================================
[  523.728706] [ INFO: possible circular locking dependency detected ]
[  523.734990] 4.1.0-dbg-DEV #1676 Not tainted
[  523.739202] -------------------------------------------------------
[  523.745474] ss/18032 is trying to acquire lock:
[  523.750002]  (slock-AF_INET){+.-...}, at: [<ffffffff81669d44>] tcp_get_info+0x2c4/0x360
[  523.758129]
[  523.758129] but task is already holding lock:
[  523.763968]  (&(&hashinfo->ehash_locks[i])->rlock){+.-...}, at: [<ffffffff816bcb75>] inet_diag_dump_icsk+0x1d5/0x6c0
[  523.774661]
[  523.774661] which lock already depends on the new lock.
[  523.774661]
[  523.782850]
[  523.782850] the existing dependency chain (in reverse order) is:
[  523.790326]
-> #1 (&(&hashinfo->ehash_locks[i])->rlock){+.-...}:
[  523.796599]        [<ffffffff811126bb>] lock_acquire+0xbb/0x270
[  523.802565]        [<ffffffff816f5868>] _raw_spin_lock+0x38/0x50
[  523.808628]        [<ffffffff81665af8>] __inet_hash_nolisten+0x78/0x110
[  523.815273]        [<ffffffff816819db>] tcp_v4_syn_recv_sock+0x24b/0x350
[  523.822067]        [<ffffffff81684d41>] tcp_check_req+0x3c1/0x500
[  523.828199]        [<ffffffff81682d09>] tcp_v4_do_rcv+0x239/0x3d0
[  523.834331]        [<ffffffff816842fe>] tcp_v4_rcv+0xa8e/0xc10
[  523.840202]        [<ffffffff81658fa3>] ip_local_deliver_finish+0x133/0x3e0
[  523.847214]        [<ffffffff81659a9a>] ip_local_deliver+0xaa/0xc0
[  523.853440]        [<ffffffff816593b8>] ip_rcv_finish+0x168/0x5c0
[  523.859624]        [<ffffffff81659db7>] ip_rcv+0x307/0x420

Lets use u64_sync infrastructure instead. As a bonus, 64bit
arches get optimized, as these are nop for them.

Fixes: 0df48c26d841 ("tcp: add tcpi_bytes_acked to tcp_info")
Signed-off-by: Eric Dumazet <edumazet@google.com>
Signed-off-by: David S. Miller <davem@davemloft.net>
---
 include/linux/tcp.h     |  2 ++
 net/ipv4/tcp.c          | 11 +++++++----
 net/ipv4/tcp_fastopen.c |  4 ++++
 net/ipv4/tcp_input.c    |  4 ++++
 4 files changed, 17 insertions(+), 4 deletions(-)

(limited to 'net/ipv4/tcp_input.c')

diff --git a/include/linux/tcp.h b/include/linux/tcp.h
index 3b2911502a8c..e8bbf403618f 100644
--- a/include/linux/tcp.h
+++ b/include/linux/tcp.h
@@ -158,6 +158,8 @@ struct tcp_sock {
 				 * sum(delta(snd_una)), or how many bytes
 				 * were acked.
 				 */
+	struct u64_stats_sync syncp; /* protects 64bit vars (cf tcp_get_info()) */
+
  	u32	snd_una;	/* First byte we want an ack for	*/
  	u32	snd_sml;	/* Last byte of the most recently transmitted small packet */
 	u32	rcv_tstamp;	/* timestamp of last received ACK (for keepalives) */
diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
index 46efa03d2b11..f1377f2a0472 100644
--- a/net/ipv4/tcp.c
+++ b/net/ipv4/tcp.c
@@ -402,6 +402,7 @@ void tcp_init_sock(struct sock *sk)
 	tp->snd_ssthresh = TCP_INFINITE_SSTHRESH;
 	tp->snd_cwnd_clamp = ~0;
 	tp->mss_cache = TCP_MSS_DEFAULT;
+	u64_stats_init(&tp->syncp);
 
 	tp->reordering = sysctl_tcp_reordering;
 	tcp_enable_early_retrans(tp);
@@ -2598,6 +2599,7 @@ void tcp_get_info(struct sock *sk, struct tcp_info *info)
 	const struct tcp_sock *tp = tcp_sk(sk);
 	const struct inet_connection_sock *icsk = inet_csk(sk);
 	u32 now = tcp_time_stamp;
+	unsigned int start;
 	u32 rate;
 
 	memset(info, 0, sizeof(*info));
@@ -2665,10 +2667,11 @@ void tcp_get_info(struct sock *sk, struct tcp_info *info)
 	rate = READ_ONCE(sk->sk_max_pacing_rate);
 	info->tcpi_max_pacing_rate = rate != ~0U ? rate : ~0ULL;
 
-	spin_lock_bh(&sk->sk_lock.slock);
-	info->tcpi_bytes_acked = tp->bytes_acked;
-	info->tcpi_bytes_received = tp->bytes_received;
-	spin_unlock_bh(&sk->sk_lock.slock);
+	do {
+		start = u64_stats_fetch_begin_irq(&tp->syncp);
+		info->tcpi_bytes_acked = tp->bytes_acked;
+		info->tcpi_bytes_received = tp->bytes_received;
+	} while (u64_stats_fetch_retry_irq(&tp->syncp, start));
 }
 EXPORT_SYMBOL_GPL(tcp_get_info);
 
diff --git a/net/ipv4/tcp_fastopen.c b/net/ipv4/tcp_fastopen.c
index 3c673d5e6cff..46b087a27503 100644
--- a/net/ipv4/tcp_fastopen.c
+++ b/net/ipv4/tcp_fastopen.c
@@ -206,6 +206,10 @@ static bool tcp_fastopen_create_child(struct sock *sk,
 			skb_set_owner_r(skb2, child);
 			__skb_queue_tail(&child->sk_receive_queue, skb2);
 			tp->syn_data_acked = 1;
+
+			/* u64_stats_update_begin(&tp->syncp) not needed here,
+			 * as we certainly are not changing upper 32bit value (0)
+			 */
 			tp->bytes_received = end_seq - TCP_SKB_CB(skb)->seq - 1;
 		} else {
 			end_seq = TCP_SKB_CB(skb)->seq + 1;
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 243d674b3ef5..c9ab964189a0 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -3286,7 +3286,9 @@ static void tcp_snd_una_update(struct tcp_sock *tp, u32 ack)
 {
 	u32 delta = ack - tp->snd_una;
 
+	u64_stats_update_begin(&tp->syncp);
 	tp->bytes_acked += delta;
+	u64_stats_update_end(&tp->syncp);
 	tp->snd_una = ack;
 }
 
@@ -3295,7 +3297,9 @@ static void tcp_rcv_nxt_update(struct tcp_sock *tp, u32 seq)
 {
 	u32 delta = seq - tp->rcv_nxt;
 
+	u64_stats_update_begin(&tp->syncp);
 	tp->bytes_received += delta;
+	u64_stats_update_end(&tp->syncp);
 	tp->rcv_nxt = seq;
 }
 
-- 
cgit v1.2.3-55-g7522