From mboxrd@z Thu Jan 1 00:00:00 1970 From: Yuchung Cheng Subject: Re: [PATCH net-next] tcp: add tcpi_segs_in and tcpi_segs_out to tcp_info Date: Thu, 21 May 2015 12:41:21 -0700 Message-ID: References: <1430260098-14127-1-git-send-email-edumazet@google.com> <20150512130855.GA29412@localhost.localdomain> <1431462713.566.95.camel@edumazet-glaptop2.roam.corp.google.com> <1432164941.4060.57.camel@edumazet-glaptop2.roam.corp.google.com> Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Cc: Marcelo Ricardo Leitner , Eric Dumazet , "David S. Miller" , netdev , Matt Mathis , Craig Gallek , Martin Lau , Chris Rapier To: Eric Dumazet Return-path: Received: from mail-oi0-f43.google.com ([209.85.218.43]:36168 "EHLO mail-oi0-f43.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754076AbbEUTmD (ORCPT ); Thu, 21 May 2015 15:42:03 -0400 Received: by oihb142 with SMTP id b142so6573634oih.3 for ; Thu, 21 May 2015 12:42:01 -0700 (PDT) In-Reply-To: <1432164941.4060.57.camel@edumazet-glaptop2.roam.corp.google.com> Sender: netdev-owner@vger.kernel.org List-ID: On Wed, May 20, 2015 at 4:35 PM, Eric Dumazet wrote: > > From: Marcelo Ricardo Leitner > > This patch tracks the total number of inbound and outbound segments on a > TCP socket. One may use this number to have an idea on connection > quality when compared against the retransmissions. > > RFC4898 named these : tcpEStatsPerfSegsIn and tcpEStatsPerfSegsOut > > These are a 32bit field each 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->segs_out was placed near tp->snd_nxt for good data > locality and minimal performance impact, while tp->segs_in was placed > near tp->bytes_received for the same reason. > > Join work with Eric Dumazet. > > Note that received SYN are accounted on the listener, but sent SYNACK > are not accounted. > > Signed-off-by: Marcelo Ricardo Leitner > Signed-off-by: Eric Dumazet > --- > include/linux/tcp.h | 7 ++++++- > include/uapi/linux/tcp.h | 4 +++- > net/ipv4/tcp.c | 2 ++ > net/ipv4/tcp_ipv4.c | 1 + > net/ipv4/tcp_minisocks.c | 1 + > net/ipv4/tcp_output.c | 1 + > net/ipv6/tcp_ipv6.c | 1 + > 7 files changed, 15 insertions(+), 2 deletions(-) > > diff --git a/include/linux/tcp.h b/include/linux/tcp.h > index e6fb5df22db1fb3a2a902581d958e6d4881b399b..f0212026c77fc1d74db96c0312fe9892f56c2a64 100644 > --- a/include/linux/tcp.h > +++ b/include/linux/tcp.h > @@ -149,11 +149,16 @@ struct tcp_sock { > * sum(delta(rcv_nxt)), or how many bytes > * were acked. > */ > + u32 segs_in; /* RFC4898 tcpEStatsPerfSegsIn > + * total number of segments in. > + */ > 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 */ > u32 snd_nxt; /* Next sequence we send */ > - > + u32 segs_out; /* RFC4898 tcpEStatsPerfSegsOut > + * The total number of segments sent. > + */ > u64 bytes_acked; /* RFC4898 tcpEStatsAppHCThruOctetsAcked > * sum(delta(snd_una)), or how many bytes > * were acked. > diff --git a/include/uapi/linux/tcp.h b/include/uapi/linux/tcp.h > index 51ebedba577f36e75b9aefd1cdf9e191f47f734f..65a77b071e22bec39225799e808b44b35bb1910c 100644 > --- a/include/uapi/linux/tcp.h > +++ b/include/uapi/linux/tcp.h > @@ -192,8 +192,10 @@ struct tcp_info { > > __u64 tcpi_pacing_rate; > __u64 tcpi_max_pacing_rate; > - __u64 tcpi_bytes_acked; /* RFC4898 tcpEStatsAppHCThruOctetsAcked */ > + __u64 tcpi_bytes_acked; /* RFC4898 tcpEStatsAppHCThruOctetsAcked */ > __u64 tcpi_bytes_received; /* RFC4898 tcpEStatsAppHCThruOctetsReceived */ > + __u32 tcpi_segs_out; /* RFC4898 tcpEStatsPerfSegsOut */ > + __u32 tcpi_segs_in; /* RFC4898 tcpEStatsPerfSegsIn */ > }; > > /* for TCP_MD5SIG socket option */ > diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c > index bb9bb844204f9f0cf9197fe323f287dce5e5bbd9..f283aba62cf313651f54f0e63448e89e4bafa689 100644 > --- a/net/ipv4/tcp.c > +++ b/net/ipv4/tcp.c > @@ -2686,6 +2686,8 @@ 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; > + info->tcpi_segs_out = tp->segs_out; > + info->tcpi_segs_in = tp->segs_in; > spin_unlock_bh(&sk->sk_lock.slock); > } > EXPORT_SYMBOL_GPL(tcp_get_info); > diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c > index 0cc4b5a630cd49c60ea0767debf4ee171f41ad3e..feb875769b8d57dcdb85e12b782f3f5e0fb6193a 100644 > --- a/net/ipv4/tcp_ipv4.c > +++ b/net/ipv4/tcp_ipv4.c > @@ -1626,6 +1626,7 @@ process: > skb->dev = NULL; > > bh_lock_sock_nested(sk); > + tcp_sk(sk)->segs_in += max_t(u16, 1, skb_shinfo(skb)->gso_segs); > ret = 0; > if (!sock_owned_by_user(sk)) { > if (!tcp_prequeue(sk, skb)) > diff --git a/net/ipv4/tcp_minisocks.c b/net/ipv4/tcp_minisocks.c > index ebe2ab2596ed3c2fbd27ecb3bf6f6b66c9e06e08..b62d15c8694679ed8688ab628f7edeb0b065dfe8 100644 > --- a/net/ipv4/tcp_minisocks.c > +++ b/net/ipv4/tcp_minisocks.c > @@ -448,6 +448,7 @@ struct sock *tcp_create_openreq_child(struct sock *sk, struct request_sock *req, > > newtp->rcv_wup = newtp->copied_seq = > newtp->rcv_nxt = treq->rcv_isn + 1; > + newtp->segs_in = 0; It'd be nice to count SYN and SYNACKs for apps tracking the handshake stats. For syn-cookies we can't do much. But for the rest we can account req->num_retrans for SYN-ACKs sent, and perhaps track SYN received in request sock? > > newtp->snd_sml = newtp->snd_una = > newtp->snd_nxt = newtp->snd_up = treq->snt_isn + 1; > diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c > index e29d43b5a0bb46305b0ab4af29bd05a61abd522d..e19594ac540a5c6091c43afc30c7510287260968 100644 > --- a/net/ipv4/tcp_output.c > +++ b/net/ipv4/tcp_output.c > @@ -1027,6 +1027,7 @@ static int tcp_transmit_skb(struct sock *sk, struct sk_buff *skb, int clone_it, > TCP_ADD_STATS(sock_net(sk), TCP_MIB_OUTSEGS, > tcp_skb_pcount(skb)); > > + tp->segs_out += tcp_skb_pcount(skb); > /* OK, its time to fill skb_shinfo(skb)->gso_segs */ > skb_shinfo(skb)->gso_segs = tcp_skb_pcount(skb); > > diff --git a/net/ipv6/tcp_ipv6.c b/net/ipv6/tcp_ipv6.c > index b6575d6655681e8e84993a5db929c7309d47d4d3..beac6bf840b9a9d1e2f281d2b1c71b5a3414b824 100644 > --- a/net/ipv6/tcp_ipv6.c > +++ b/net/ipv6/tcp_ipv6.c > @@ -1421,6 +1421,7 @@ process: > skb->dev = NULL; > > bh_lock_sock_nested(sk); > + tcp_sk(sk)->segs_in += max_t(u16, 1, skb_shinfo(skb)->gso_segs); some helper like e.g. tcp_event_pkt_recv() for v4/v6? i am curious why use max instead of the ternary op? > ret = 0; > if (!sock_owned_by_user(sk)) { > if (!tcp_prequeue(sk, skb)) > >