[PATCH net-next 0/2] tcp: tcp_info extensions

[PATCH net-next 0/2] tcp: tcp_info extensions

[Eric Dumazet]

As discussed during Chris Rapier presentation in Ottawa / netdev0.1,
we add to tcp_info the first two fields are highly wanted.

Each field is added into a single patch for easy code review.

(Corresponding iproute2/ss patches will be sent)

Next fields will follow once consensus is reached.

Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: Matt Mathis <mattmathis@google.com>
Cc: Eric Salo <salo@google.com>
Cc: Yuchung Cheng <ycheng@google.com>
Cc: Martin Lau <kafai@fb.com>
Cc: Chris Rapier <rapier@psc.edu>

Eric Dumazet (2):
  tcp: add tcpi_bytes_acked to tcp_info
  tcp: add tcpi_bytes_received to tcp_info

 include/linux/tcp.h      |  8 ++++++++
 include/net/tcp.h        |  2 +-
 include/uapi/linux/tcp.h |  2 ++
 net/ipv4/tcp.c           |  7 ++++++-
 net/ipv4/tcp_input.c     | 30 ++++++++++++++++++++++++------
 5 files changed, 41 insertions(+), 8 deletions(-)

-- 
2.2.0.rc0.207.ga3a616c

[PATCH net-next 1/2] tcp: add tcpi_bytes_acked to tcp_info

[Eric Dumazet]

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>
Cc: Matt Mathis <mattmathis@google.com>
Cc: Eric Salo <salo@google.com>
Cc: Yuchung Cheng <ycheng@google.com>
Cc: Martin Lau <kafai@fb.com>
Cc: Chris Rapier <rapier@psc.edu>
---
 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(-)

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);
-- 
2.2.0.rc0.207.ga3a616c

[PATCH net-next 2/2] tcp: add tcpi_bytes_received to tcp_info

[Eric Dumazet]

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: Matt Mathis <mattmathis@google.com>
Cc: Eric Salo <salo@google.com>
Cc: Yuchung Cheng <ycheng@google.com>
Cc: Martin Lau <kafai@fb.com>
Cc: Chris Rapier <rapier@psc.edu>
---
 include/linux/tcp.h      |  4 ++++
 include/uapi/linux/tcp.h |  1 +
 net/ipv4/tcp.c           |  1 +
 net/ipv4/tcp_input.c     | 17 +++++++++++++----
 4 files changed, 19 insertions(+), 4 deletions(-)

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_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;
 				}
-- 
2.2.0.rc0.207.ga3a616c

Re: [PATCH net-next 1/2] tcp: add tcpi_bytes_acked to tcp_info

[Yuchung Cheng]

On Tue, Apr 28, 2015 at 10:32 AM, Eric Dumazet <edumazet@google.com> wrote:
>
> 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>
> Cc: Matt Mathis <mattmathis@google.com>
> Cc: Eric Salo <salo@google.com>
> Cc: Yuchung Cheng <ycheng@google.com>
> Cc: Martin Lau <kafai@fb.com>
> Cc: Chris Rapier <rapier@psc.edu>
Acked-by: Yuchung Cheng <ycheng@google.com>

> ---
>  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(-)
>
> 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);
> --
> 2.2.0.rc0.207.ga3a616c
>

Re: [PATCH net-next 2/2] tcp: add tcpi_bytes_received to tcp_info

[Yuchung Cheng]

On Tue, Apr 28, 2015 at 10:32 AM, Eric Dumazet <edumazet@google.com> wrote:
> 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: Matt Mathis <mattmathis@google.com>
> Cc: Eric Salo <salo@google.com>
> Cc: Yuchung Cheng <ycheng@google.com>
> Cc: Martin Lau <kafai@fb.com>
> Cc: Chris Rapier <rapier@psc.edu>
> ---
>  include/linux/tcp.h      |  4 ++++
>  include/uapi/linux/tcp.h |  1 +
>  net/ipv4/tcp.c           |  1 +
>  net/ipv4/tcp_input.c     | 17 +++++++++++++----
>  4 files changed, 19 insertions(+), 4 deletions(-)
>
> 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_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;
> +}
Account for received syn-data when rcv_nxt is updated in tcp_fastopen.c?

> +
>  /* 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;
>                                 }
> --
> 2.2.0.rc0.207.ga3a616c
>

Re: [PATCH net-next 2/2] tcp: add tcpi_bytes_received to tcp_info

[Eric Dumazet]

On Tue, 2015-04-28 at 14:07 -0700, Yuchung Cheng wrote:

> Account for received syn-data when rcv_nxt is updated in tcp_fastopen.c?

Right, I'll send a v2 with the following FO part :

Note that I do not call the helper to keep it static and inline
candidate.

Thanks !

diff --git a/net/ipv4/tcp_fastopen.c b/net/ipv4/tcp_fastopen.c
index e3d87aca6be8..b1b110d07816 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;
                }