Commit 9c25aae0 authored by Eric Dumazet's avatar Eric Dumazet Committed by Jakub Kicinski
Browse files

tcp: fix tcp_disordered_ack() vs usec TS resolution 



After commit 93946301 ("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: 614e8316 ("tcp: add support for usec resolution in TCP TS values")
Co-developed-by: default avatarNeal Cardwell <ncardwell@google.com>
Signed-off-by: default avatarNeal Cardwell <ncardwell@google.com>
Signed-off-by: default avatarEric 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: default avatarJakub Kicinski <kuba@kernel.org>
parent a45f1e46

net/ipv4/tcp_input.c

+20 −2
Original line number Diff line number Diff line
@@ -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,