diff options
| author | Eric Dumazet <edumazet@google.com> | 2023-12-07 18:13:42 +0000 |
|---|---|---|
| committer | Jakub Kicinski <kuba@kernel.org> | 2023-12-08 17:15:51 -0800 |
| commit | 9c25aae0132b8babbe71ebb2160186d69231d9b3 (patch) | |
| tree | 86bd1fd6f173616afa85e8322459c36f0aa4f99a /net/ipv4/tcp_input.c | |
| parent | a45f1e46274256852990c479fbb1198a7d84529b (diff) | |
tcp: fix tcp_disordered_ack() vs usec TS resolution
After commit 939463016b7a ("tcp: change data receiver flowlabel after one dup")
we noticed an increase of TCPACKSkippedPAWS events.
Neal Cardwell tracked the issue to tcp_disordered_ack() assumption
about remote peer TS clock.
RFC 1323 & 7323 are suggesting the following:
"timestamp clock frequency in the range 1 ms to 1 sec per tick
between 1ms and 1sec."
This has to be adjusted for 1 MHz clock frequency.
This hints at reorders of SACK packets on send side,
this might deserve a future patch.
(skb->ooo_okay is always set for pure ACK packets)
Fixes: 614e8316aa4c ("tcp: add support for usec resolution in TCP TS values")
Co-developed-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Neal Cardwell <ncardwell@google.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
Cc: David Morley <morleyd@google.com>
Link: https://lore.kernel.org/r/20231207181342.525181-1-edumazet@google.com
Signed-off-by: Jakub Kicinski <kuba@kernel.org>
Diffstat (limited to 'net/ipv4/tcp_input.c')
| -rw-r--r-- | net/ipv4/tcp_input.c | 22 |
1 files changed, 20 insertions, 2 deletions
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 90de838a2745..701cb87043f2 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -4368,6 +4368,23 @@ EXPORT_SYMBOL(tcp_do_parse_auth_options); * up to bandwidth of 18Gigabit/sec. 8) ] */ +/* Estimates max number of increments of remote peer TSval in + * a replay window (based on our current RTO estimation). + */ +static u32 tcp_tsval_replay(const struct sock *sk) +{ + /* If we use usec TS resolution, + * then expect the remote peer to use the same resolution. + */ + if (tcp_sk(sk)->tcp_usec_ts) + return inet_csk(sk)->icsk_rto * (USEC_PER_SEC / HZ); + + /* RFC 7323 recommends a TSval clock between 1ms and 1sec. + * We know that some OS (including old linux) can use 1200 Hz. + */ + return inet_csk(sk)->icsk_rto * 1200 / HZ; +} + static int tcp_disordered_ack(const struct sock *sk, const struct sk_buff *skb) { const struct tcp_sock *tp = tcp_sk(sk); @@ -4375,7 +4392,7 @@ static int tcp_disordered_ack(const struct sock *sk, const struct sk_buff *skb) u32 seq = TCP_SKB_CB(skb)->seq; u32 ack = TCP_SKB_CB(skb)->ack_seq; - return (/* 1. Pure ACK with correct sequence number. */ + return /* 1. Pure ACK with correct sequence number. */ (th->ack && seq == TCP_SKB_CB(skb)->end_seq && seq == tp->rcv_nxt) && /* 2. ... and duplicate ACK. */ @@ -4385,7 +4402,8 @@ static int tcp_disordered_ack(const struct sock *sk, const struct sk_buff *skb) !tcp_may_update_window(tp, ack, seq, ntohs(th->window) << tp->rx_opt.snd_wscale) && /* 4. ... and sits in replay window. */ - (s32)(tp->rx_opt.ts_recent - tp->rx_opt.rcv_tsval) <= (inet_csk(sk)->icsk_rto * 1024) / HZ); + (s32)(tp->rx_opt.ts_recent - tp->rx_opt.rcv_tsval) <= + tcp_tsval_replay(sk); } static inline bool tcp_paws_discard(const struct sock *sk, |