From: David Morley <morleyd.kernel@gmail•com>
To: David Miller <davem@davemloft•net>,
Eric Dumazet <edumazet@google•com>,
Jakub Kicinski <kuba@kernel•org>
Cc: netdev@vger•kernel.org, David Morley <morleyd@google•com>,
Neal Cardwell <ncardwell@google•com>,
Yuchung Cheng <ycheng@google•com>
Subject: [PATCH net-next v3 1/2] tcp: record last received ipv6 flowlabel
Date: Fri, 6 Oct 2023 01:18:40 +0000 [thread overview]
Message-ID: <20231006011841.3558307-2-morleyd.kernel@gmail.com> (raw)
In-Reply-To: <20231006011841.3558307-1-morleyd.kernel@gmail.com>
From: David Morley <morleyd@google•com>
In order to better estimate whether a data packet has been
retransmitted or is the result of a TLP, we save the last received
ipv6 flowlabel.
To make space for this field we resize the "ato" field in
inet_connection_sock as the current value of TCP_DELACK_MAX can be
fully contained in 8 bits and add a compile_time_assert ensuring this
field is the required size.
v2: addressed kernel bot feedback about dccp_delack_timer()
v3: addressed build error introduced by commit bbf80d713fe7 ("tcp:
derive delack_max from rto_min")
Signed-off-by: David Morley <morleyd@google•com>
Signed-off-by: Neal Cardwell <ncardwell@google•com>
Signed-off-by: Yuchung Cheng <ycheng@google•com>
Tested-by: David Morley <morleyd@google•com>
Reviewed-by: Eric Dumazet <edumazet@google•com>
---
include/net/inet_connection_sock.h | 5 ++++-
include/net/tcp.h | 2 ++
net/dccp/timer.c | 4 ++--
net/ipv4/tcp.c | 4 ++--
net/ipv4/tcp_input.c | 15 +++++++++++++++
net/ipv4/tcp_timer.c | 2 +-
6 files changed, 26 insertions(+), 6 deletions(-)
diff --git a/include/net/inet_connection_sock.h b/include/net/inet_connection_sock.h
index 5d2fcc137b88..d6d9d1c1985a 100644
--- a/include/net/inet_connection_sock.h
+++ b/include/net/inet_connection_sock.h
@@ -114,7 +114,10 @@ struct inet_connection_sock {
__u8 quick; /* Scheduled number of quick acks */
__u8 pingpong; /* The session is interactive */
__u8 retry; /* Number of attempts */
- __u32 ato; /* Predicted tick of soft clock */
+ #define ATO_BITS 8
+ __u32 ato:ATO_BITS, /* Predicted tick of soft clock */
+ lrcv_flowlabel:20, /* last received ipv6 flowlabel */
+ unused:4;
unsigned long timeout; /* Currently scheduled timeout */
__u32 lrcvtime; /* timestamp of last received data packet */
__u16 last_seg_size; /* Size of last incoming segment */
diff --git a/include/net/tcp.h b/include/net/tcp.h
index af9cb37fbe53..2083f99f9e96 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -131,6 +131,8 @@ void tcp_time_wait(struct sock *sk, int state, int timeo);
#define TCP_FIN_TIMEOUT_MAX (120 * HZ) /* max TCP_LINGER2 value (two minutes) */
#define TCP_DELACK_MAX ((unsigned)(HZ/5)) /* maximal time to delay before sending an ACK */
+static_assert((1 << ATO_BITS) > TCP_DELACK_MAX);
+
#if HZ >= 100
#define TCP_DELACK_MIN ((unsigned)(HZ/25)) /* minimal time to delay before sending an ACK */
#define TCP_ATO_MIN ((unsigned)(HZ/25))
diff --git a/net/dccp/timer.c b/net/dccp/timer.c
index b3255e87cc7e..a4cfb47b60e5 100644
--- a/net/dccp/timer.c
+++ b/net/dccp/timer.c
@@ -196,8 +196,8 @@ static void dccp_delack_timer(struct timer_list *t)
if (inet_csk_ack_scheduled(sk)) {
if (!inet_csk_in_pingpong_mode(sk)) {
/* Delayed ACK missed: inflate ATO. */
- icsk->icsk_ack.ato = min(icsk->icsk_ack.ato << 1,
- icsk->icsk_rto);
+ icsk->icsk_ack.ato = min_t(u32, icsk->icsk_ack.ato << 1,
+ icsk->icsk_rto);
} else {
/* Delayed ACK missed: leave pingpong mode and
* deflate ATO.
diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
index e54f91eb943b..706b8b271ff6 100644
--- a/net/ipv4/tcp.c
+++ b/net/ipv4/tcp.c
@@ -3762,8 +3762,8 @@ void tcp_get_info(struct sock *sk, struct tcp_info *info)
info->tcpi_options |= TCPI_OPT_SYN_DATA;
info->tcpi_rto = jiffies_to_usecs(icsk->icsk_rto);
- info->tcpi_ato = jiffies_to_usecs(min(icsk->icsk_ack.ato,
- tcp_delack_max(sk)));
+ info->tcpi_ato = jiffies_to_usecs(min_t(u32, icsk->icsk_ack.ato,
+ tcp_delack_max(sk)));
info->tcpi_snd_mss = tp->mss_cache;
info->tcpi_rcv_mss = icsk->icsk_ack.rcv_mss;
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 22c2a7c2e65e..c7c15d4b95e5 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -765,6 +765,16 @@ void tcp_rcv_space_adjust(struct sock *sk)
tp->rcvq_space.time = tp->tcp_mstamp;
}
+static void tcp_save_lrcv_flowlabel(struct sock *sk, const struct sk_buff *skb)
+{
+#if IS_ENABLED(CONFIG_IPV6)
+ struct inet_connection_sock *icsk = inet_csk(sk);
+
+ if (skb->protocol == htons(ETH_P_IPV6))
+ icsk->icsk_ack.lrcv_flowlabel = ntohl(ip6_flowlabel(ipv6_hdr(skb)));
+#endif
+}
+
/* There is something which you must keep in mind when you analyze the
* behavior of the tp->ato delayed ack timeout interval. When a
* connection starts up, we want to ack as quickly as possible. The
@@ -813,6 +823,7 @@ static void tcp_event_data_recv(struct sock *sk, struct sk_buff *skb)
}
}
icsk->icsk_ack.lrcvtime = now;
+ tcp_save_lrcv_flowlabel(sk, skb);
tcp_ecn_check_ce(sk, skb);
@@ -4506,6 +4517,9 @@ static void tcp_rcv_spurious_retrans(struct sock *sk, const struct sk_buff *skb)
if (TCP_SKB_CB(skb)->seq == tcp_sk(sk)->duplicate_sack[0].start_seq &&
sk_rethink_txhash(sk))
NET_INC_STATS(sock_net(sk), LINUX_MIB_TCPDUPLICATEDATAREHASH);
+
+ /* Save last flowlabel after a spurious retrans. */
+ tcp_save_lrcv_flowlabel(sk, skb);
}
static void tcp_send_dupack(struct sock *sk, const struct sk_buff *skb)
@@ -4822,6 +4836,7 @@ static void tcp_data_queue_ofo(struct sock *sk, struct sk_buff *skb)
u32 seq, end_seq;
bool fragstolen;
+ tcp_save_lrcv_flowlabel(sk, skb);
tcp_ecn_check_ce(sk, skb);
if (unlikely(tcp_try_rmem_schedule(sk, skb, skb->truesize))) {
diff --git a/net/ipv4/tcp_timer.c b/net/ipv4/tcp_timer.c
index 3f61c6a70a1f..0862b73dd3b5 100644
--- a/net/ipv4/tcp_timer.c
+++ b/net/ipv4/tcp_timer.c
@@ -322,7 +322,7 @@ void tcp_delack_timer_handler(struct sock *sk)
if (inet_csk_ack_scheduled(sk)) {
if (!inet_csk_in_pingpong_mode(sk)) {
/* Delayed ACK missed: inflate ATO. */
- icsk->icsk_ack.ato = min(icsk->icsk_ack.ato << 1, icsk->icsk_rto);
+ icsk->icsk_ack.ato = min_t(u32, icsk->icsk_ack.ato << 1, icsk->icsk_rto);
} else {
/* Delayed ACK missed: leave pingpong mode and
* deflate ATO.
--
2.42.0.609.gbb76f46606-goog
next prev parent reply other threads:[~2023-10-06 1:18 UTC|newest]
Thread overview: 4+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-10-06 1:18 [PATCH net-next v3 0/2] tcp: save flowlabel and use for receiver repathing David Morley
2023-10-06 1:18 ` David Morley [this message]
2023-10-06 1:18 ` [PATCH net-next v3 2/2] tcp: change data receiver flowlabel after one dup David Morley
2023-10-10 8:30 ` [PATCH net-next v3 0/2] tcp: save flowlabel and use for receiver repathing patchwork-bot+netdevbpf
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=20231006011841.3558307-2-morleyd.kernel@gmail.com \
--to=morleyd.kernel@gmail$(echo .)com \
--cc=davem@davemloft$(echo .)net \
--cc=edumazet@google$(echo .)com \
--cc=kuba@kernel$(echo .)org \
--cc=morleyd@google$(echo .)com \
--cc=ncardwell@google$(echo .)com \
--cc=netdev@vger$(echo .)kernel.org \
--cc=ycheng@google$(echo .)com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox