* [PATCH net-next 0/2] tcp: reduce quickack pressure for ECN
@ 2018-05-21 22:08 Eric Dumazet
2018-05-21 22:08 ` [PATCH net-next 1/2] tcp: add max_quickacks param to tcp_incr_quickack and tcp_enter_quickack_mode Eric Dumazet
` (2 more replies)
0 siblings, 3 replies; 11+ messages in thread
From: Eric Dumazet @ 2018-05-21 22:08 UTC (permalink / raw)
To: David S . Miller
Cc: netdev, Van Jacobson, Neal Cardwell, Yuchung Cheng,
Soheil Hassas Yeganeh, Eric Dumazet, Eric Dumazet
Small patch series changing TCP behavior vs quickack and ECN
First patch is a refactoring, adding parameter to tcp_incr_quickack()
and tcp_enter_quickack_mode() helpers.
Second patch implements the change, lowering number of ACK packets
sent after an ECN event.
Eric Dumazet (2):
tcp: add max_quickacks param to tcp_incr_quickack and
tcp_enter_quickack_mode
tcp: do not aggressively quick ack after ECN events
net/ipv4/tcp_input.c | 24 +++++++++++++-----------
1 file changed, 13 insertions(+), 11 deletions(-)
--
2.17.0.441.gb46fe60e1d-goog
^ permalink raw reply [flat|nested] 11+ messages in thread* [PATCH net-next 1/2] tcp: add max_quickacks param to tcp_incr_quickack and tcp_enter_quickack_mode 2018-05-21 22:08 [PATCH net-next 0/2] tcp: reduce quickack pressure for ECN Eric Dumazet @ 2018-05-21 22:08 ` Eric Dumazet 2018-05-22 14:03 ` Soheil Hassas Yeganeh ` (2 more replies) 2018-05-21 22:08 ` [PATCH net-next 2/2] tcp: do not aggressively quick ack after ECN events Eric Dumazet 2018-05-22 19:43 ` [PATCH net-next 0/2] tcp: reduce quickack pressure for ECN David Miller 2 siblings, 3 replies; 11+ messages in thread From: Eric Dumazet @ 2018-05-21 22:08 UTC (permalink / raw) To: David S . Miller Cc: netdev, Van Jacobson, Neal Cardwell, Yuchung Cheng, Soheil Hassas Yeganeh, Eric Dumazet, Eric Dumazet We want to add finer control of the number of ACK packets sent after ECN events. This patch is not changing current behavior, it only enables following change. Signed-off-by: Eric Dumazet <edumazet@google.com> --- net/ipv4/tcp_input.c | 24 +++++++++++++----------- 1 file changed, 13 insertions(+), 11 deletions(-) diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index aebb29ab2fdf2ceaa182cd11928f145a886149ff..2e970e9f4e09d966b703af2d14d521a4328eba7e 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -203,21 +203,23 @@ static void tcp_measure_rcv_mss(struct sock *sk, const struct sk_buff *skb) } } -static void tcp_incr_quickack(struct sock *sk) +static void tcp_incr_quickack(struct sock *sk, unsigned int max_quickacks) { struct inet_connection_sock *icsk = inet_csk(sk); unsigned int quickacks = tcp_sk(sk)->rcv_wnd / (2 * icsk->icsk_ack.rcv_mss); if (quickacks == 0) quickacks = 2; + quickacks = min(quickacks, max_quickacks); if (quickacks > icsk->icsk_ack.quick) - icsk->icsk_ack.quick = min(quickacks, TCP_MAX_QUICKACKS); + icsk->icsk_ack.quick = quickacks; } -static void tcp_enter_quickack_mode(struct sock *sk) +static void tcp_enter_quickack_mode(struct sock *sk, unsigned int max_quickacks) { struct inet_connection_sock *icsk = inet_csk(sk); - tcp_incr_quickack(sk); + + tcp_incr_quickack(sk, max_quickacks); icsk->icsk_ack.pingpong = 0; icsk->icsk_ack.ato = TCP_ATO_MIN; } @@ -261,7 +263,7 @@ static void __tcp_ecn_check_ce(struct tcp_sock *tp, const struct sk_buff *skb) * it is probably a retransmit. */ if (tp->ecn_flags & TCP_ECN_SEEN) - tcp_enter_quickack_mode((struct sock *)tp); + tcp_enter_quickack_mode((struct sock *)tp, TCP_MAX_QUICKACKS); break; case INET_ECN_CE: if (tcp_ca_needs_ecn((struct sock *)tp)) @@ -269,7 +271,7 @@ static void __tcp_ecn_check_ce(struct tcp_sock *tp, const struct sk_buff *skb) if (!(tp->ecn_flags & TCP_ECN_DEMAND_CWR)) { /* Better not delay acks, sender can have a very low cwnd */ - tcp_enter_quickack_mode((struct sock *)tp); + tcp_enter_quickack_mode((struct sock *)tp, TCP_MAX_QUICKACKS); tp->ecn_flags |= TCP_ECN_DEMAND_CWR; } tp->ecn_flags |= TCP_ECN_SEEN; @@ -686,7 +688,7 @@ static void tcp_event_data_recv(struct sock *sk, struct sk_buff *skb) /* The _first_ data packet received, initialize * delayed ACK engine. */ - tcp_incr_quickack(sk); + tcp_incr_quickack(sk, TCP_MAX_QUICKACKS); icsk->icsk_ack.ato = TCP_ATO_MIN; } else { int m = now - icsk->icsk_ack.lrcvtime; @@ -702,7 +704,7 @@ static void tcp_event_data_recv(struct sock *sk, struct sk_buff *skb) /* Too long gap. Apparently sender failed to * restart window, so that we send ACKs quickly. */ - tcp_incr_quickack(sk); + tcp_incr_quickack(sk, TCP_MAX_QUICKACKS); sk_mem_reclaim(sk); } } @@ -4179,7 +4181,7 @@ static void tcp_send_dupack(struct sock *sk, const struct sk_buff *skb) if (TCP_SKB_CB(skb)->end_seq != TCP_SKB_CB(skb)->seq && before(TCP_SKB_CB(skb)->seq, tp->rcv_nxt)) { NET_INC_STATS(sock_net(sk), LINUX_MIB_DELAYEDACKLOST); - tcp_enter_quickack_mode(sk); + tcp_enter_quickack_mode(sk, TCP_MAX_QUICKACKS); if (tcp_is_sack(tp) && sock_net(sk)->ipv4.sysctl_tcp_dsack) { u32 end_seq = TCP_SKB_CB(skb)->end_seq; @@ -4706,7 +4708,7 @@ static void tcp_data_queue(struct sock *sk, struct sk_buff *skb) tcp_dsack_set(sk, TCP_SKB_CB(skb)->seq, TCP_SKB_CB(skb)->end_seq); out_of_window: - tcp_enter_quickack_mode(sk); + tcp_enter_quickack_mode(sk, TCP_MAX_QUICKACKS); inet_csk_schedule_ack(sk); drop: tcp_drop(sk, skb); @@ -5790,7 +5792,7 @@ static int tcp_rcv_synsent_state_process(struct sock *sk, struct sk_buff *skb, * to stand against the temptation 8) --ANK */ inet_csk_schedule_ack(sk); - tcp_enter_quickack_mode(sk); + tcp_enter_quickack_mode(sk, TCP_MAX_QUICKACKS); inet_csk_reset_xmit_timer(sk, ICSK_TIME_DACK, TCP_DELACK_MAX, TCP_RTO_MAX); -- 2.17.0.441.gb46fe60e1d-goog ^ permalink raw reply related [flat|nested] 11+ messages in thread
* Re: [PATCH net-next 1/2] tcp: add max_quickacks param to tcp_incr_quickack and tcp_enter_quickack_mode 2018-05-21 22:08 ` [PATCH net-next 1/2] tcp: add max_quickacks param to tcp_incr_quickack and tcp_enter_quickack_mode Eric Dumazet @ 2018-05-22 14:03 ` Soheil Hassas Yeganeh 2018-05-22 17:27 ` Neal Cardwell 2018-05-23 0:30 ` kbuild test robot 2 siblings, 0 replies; 11+ messages in thread From: Soheil Hassas Yeganeh @ 2018-05-22 14:03 UTC (permalink / raw) To: Eric Dumazet Cc: David S . Miller, netdev, Van Jacobson, Neal Cardwell, Yuchung Cheng, Eric Dumazet On Mon, May 21, 2018 at 6:08 PM, Eric Dumazet <edumazet@google.com> wrote: > We want to add finer control of the number of ACK packets sent after > ECN events. > > This patch is not changing current behavior, it only enables following > change. > > Signed-off-by: Eric Dumazet <edumazet@google.com> Acked-by: Soheil Hassas Yeganeh <soheil@google.com> > --- > net/ipv4/tcp_input.c | 24 +++++++++++++----------- > 1 file changed, 13 insertions(+), 11 deletions(-) > > diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c > index aebb29ab2fdf2ceaa182cd11928f145a886149ff..2e970e9f4e09d966b703af2d14d521a4328eba7e 100644 > --- a/net/ipv4/tcp_input.c > +++ b/net/ipv4/tcp_input.c > @@ -203,21 +203,23 @@ static void tcp_measure_rcv_mss(struct sock *sk, const struct sk_buff *skb) > } > } > > -static void tcp_incr_quickack(struct sock *sk) > +static void tcp_incr_quickack(struct sock *sk, unsigned int max_quickacks) > { > struct inet_connection_sock *icsk = inet_csk(sk); > unsigned int quickacks = tcp_sk(sk)->rcv_wnd / (2 * icsk->icsk_ack.rcv_mss); > > if (quickacks == 0) > quickacks = 2; > + quickacks = min(quickacks, max_quickacks); > if (quickacks > icsk->icsk_ack.quick) > - icsk->icsk_ack.quick = min(quickacks, TCP_MAX_QUICKACKS); > + icsk->icsk_ack.quick = quickacks; > } > > -static void tcp_enter_quickack_mode(struct sock *sk) > +static void tcp_enter_quickack_mode(struct sock *sk, unsigned int max_quickacks) > { > struct inet_connection_sock *icsk = inet_csk(sk); > - tcp_incr_quickack(sk); > + > + tcp_incr_quickack(sk, max_quickacks); > icsk->icsk_ack.pingpong = 0; > icsk->icsk_ack.ato = TCP_ATO_MIN; > } > @@ -261,7 +263,7 @@ static void __tcp_ecn_check_ce(struct tcp_sock *tp, const struct sk_buff *skb) > * it is probably a retransmit. > */ > if (tp->ecn_flags & TCP_ECN_SEEN) > - tcp_enter_quickack_mode((struct sock *)tp); > + tcp_enter_quickack_mode((struct sock *)tp, TCP_MAX_QUICKACKS); > break; > case INET_ECN_CE: > if (tcp_ca_needs_ecn((struct sock *)tp)) > @@ -269,7 +271,7 @@ static void __tcp_ecn_check_ce(struct tcp_sock *tp, const struct sk_buff *skb) > > if (!(tp->ecn_flags & TCP_ECN_DEMAND_CWR)) { > /* Better not delay acks, sender can have a very low cwnd */ > - tcp_enter_quickack_mode((struct sock *)tp); > + tcp_enter_quickack_mode((struct sock *)tp, TCP_MAX_QUICKACKS); > tp->ecn_flags |= TCP_ECN_DEMAND_CWR; > } > tp->ecn_flags |= TCP_ECN_SEEN; > @@ -686,7 +688,7 @@ static void tcp_event_data_recv(struct sock *sk, struct sk_buff *skb) > /* The _first_ data packet received, initialize > * delayed ACK engine. > */ > - tcp_incr_quickack(sk); > + tcp_incr_quickack(sk, TCP_MAX_QUICKACKS); > icsk->icsk_ack.ato = TCP_ATO_MIN; > } else { > int m = now - icsk->icsk_ack.lrcvtime; > @@ -702,7 +704,7 @@ static void tcp_event_data_recv(struct sock *sk, struct sk_buff *skb) > /* Too long gap. Apparently sender failed to > * restart window, so that we send ACKs quickly. > */ > - tcp_incr_quickack(sk); > + tcp_incr_quickack(sk, TCP_MAX_QUICKACKS); > sk_mem_reclaim(sk); > } > } > @@ -4179,7 +4181,7 @@ static void tcp_send_dupack(struct sock *sk, const struct sk_buff *skb) > if (TCP_SKB_CB(skb)->end_seq != TCP_SKB_CB(skb)->seq && > before(TCP_SKB_CB(skb)->seq, tp->rcv_nxt)) { > NET_INC_STATS(sock_net(sk), LINUX_MIB_DELAYEDACKLOST); > - tcp_enter_quickack_mode(sk); > + tcp_enter_quickack_mode(sk, TCP_MAX_QUICKACKS); > > if (tcp_is_sack(tp) && sock_net(sk)->ipv4.sysctl_tcp_dsack) { > u32 end_seq = TCP_SKB_CB(skb)->end_seq; > @@ -4706,7 +4708,7 @@ static void tcp_data_queue(struct sock *sk, struct sk_buff *skb) > tcp_dsack_set(sk, TCP_SKB_CB(skb)->seq, TCP_SKB_CB(skb)->end_seq); > > out_of_window: > - tcp_enter_quickack_mode(sk); > + tcp_enter_quickack_mode(sk, TCP_MAX_QUICKACKS); > inet_csk_schedule_ack(sk); > drop: > tcp_drop(sk, skb); > @@ -5790,7 +5792,7 @@ static int tcp_rcv_synsent_state_process(struct sock *sk, struct sk_buff *skb, > * to stand against the temptation 8) --ANK > */ > inet_csk_schedule_ack(sk); > - tcp_enter_quickack_mode(sk); > + tcp_enter_quickack_mode(sk, TCP_MAX_QUICKACKS); > inet_csk_reset_xmit_timer(sk, ICSK_TIME_DACK, > TCP_DELACK_MAX, TCP_RTO_MAX); > > -- > 2.17.0.441.gb46fe60e1d-goog > ^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: [PATCH net-next 1/2] tcp: add max_quickacks param to tcp_incr_quickack and tcp_enter_quickack_mode 2018-05-21 22:08 ` [PATCH net-next 1/2] tcp: add max_quickacks param to tcp_incr_quickack and tcp_enter_quickack_mode Eric Dumazet 2018-05-22 14:03 ` Soheil Hassas Yeganeh @ 2018-05-22 17:27 ` Neal Cardwell 2018-05-23 0:30 ` kbuild test robot 2 siblings, 0 replies; 11+ messages in thread From: Neal Cardwell @ 2018-05-22 17:27 UTC (permalink / raw) To: Eric Dumazet Cc: David Miller, Netdev, Van Jacobson, Yuchung Cheng, Soheil Hassas Yeganeh, Eric Dumazet On Mon, May 21, 2018 at 6:09 PM Eric Dumazet <edumazet@google.com> wrote: > We want to add finer control of the number of ACK packets sent after > ECN events. > This patch is not changing current behavior, it only enables following > change. > Signed-off-by: Eric Dumazet <edumazet@google.com> > --- Acked-by: Neal Cardwell <ncardwell@google.com> Thanks! neal ^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: [PATCH net-next 1/2] tcp: add max_quickacks param to tcp_incr_quickack and tcp_enter_quickack_mode 2018-05-21 22:08 ` [PATCH net-next 1/2] tcp: add max_quickacks param to tcp_incr_quickack and tcp_enter_quickack_mode Eric Dumazet 2018-05-22 14:03 ` Soheil Hassas Yeganeh 2018-05-22 17:27 ` Neal Cardwell @ 2018-05-23 0:30 ` kbuild test robot 2018-05-23 1:37 ` Neal Cardwell 2 siblings, 1 reply; 11+ messages in thread From: kbuild test robot @ 2018-05-23 0:30 UTC (permalink / raw) To: Eric Dumazet Cc: kbuild-all, David S . Miller, netdev, Van Jacobson, Neal Cardwell, Yuchung Cheng, Soheil Hassas Yeganeh, Eric Dumazet, Eric Dumazet [-- Attachment #1: Type: text/plain, Size: 9440 bytes --] Hi Eric, Thank you for the patch! Yet something to improve: [auto build test ERROR on net/master] [also build test ERROR on v4.17-rc6 next-20180517] [cannot apply to net-next/master] [if your patch is applied to the wrong git tree, please drop us a note to help improve the system] url: https://github.com/0day-ci/linux/commits/Eric-Dumazet/tcp-add-max_quickacks-param-to-tcp_incr_quickack-and-tcp_enter_quickack_mode/20180523-075103 config: i386-randconfig-x012-201820 (attached as .config) compiler: gcc-7 (Debian 7.3.0-16) 7.3.0 reproduce: # save the attached .config to linux build tree make ARCH=i386 All errors (new ones prefixed by >>): net//ipv4/tcp_input.c: In function 'tcp_data_queue': >> net//ipv4/tcp_input.c:4656:2: error: too few arguments to function 'tcp_enter_quickack_mode' tcp_enter_quickack_mode(sk); ^~~~~~~~~~~~~~~~~~~~~~~ net//ipv4/tcp_input.c:199:13: note: declared here static void tcp_enter_quickack_mode(struct sock *sk, unsigned int max_quickacks) ^~~~~~~~~~~~~~~~~~~~~~~ vim +/tcp_enter_quickack_mode +4656 net//ipv4/tcp_input.c 292e8d8c8 Pavel Emelyanov 2012-05-10 4577 ^1da177e4 Linus Torvalds 2005-04-16 4578 static void tcp_data_queue(struct sock *sk, struct sk_buff *skb) ^1da177e4 Linus Torvalds 2005-04-16 4579 { ^1da177e4 Linus Torvalds 2005-04-16 4580 struct tcp_sock *tp = tcp_sk(sk); 5357f0bd4 Eric Dumazet 2017-08-01 4581 bool fragstolen; 5357f0bd4 Eric Dumazet 2017-08-01 4582 int eaten; ^1da177e4 Linus Torvalds 2005-04-16 4583 532182cd6 Eric Dumazet 2016-04-01 4584 if (TCP_SKB_CB(skb)->seq == TCP_SKB_CB(skb)->end_seq) { 532182cd6 Eric Dumazet 2016-04-01 4585 __kfree_skb(skb); 532182cd6 Eric Dumazet 2016-04-01 4586 return; 532182cd6 Eric Dumazet 2016-04-01 4587 } f84af32cb Eric Dumazet 2010-04-28 4588 skb_dst_drop(skb); 155c6e1ad Peter Pan(潘卫平 2014-09-24 4589) __skb_pull(skb, tcp_hdr(skb)->doff * 4); ^1da177e4 Linus Torvalds 2005-04-16 4590 735d38311 Florian Westphal 2014-09-29 4591 tcp_ecn_accept_cwr(tp, skb); ^1da177e4 Linus Torvalds 2005-04-16 4592 ^1da177e4 Linus Torvalds 2005-04-16 4593 tp->rx_opt.dsack = 0; ^1da177e4 Linus Torvalds 2005-04-16 4594 ^1da177e4 Linus Torvalds 2005-04-16 4595 /* Queue data for delivery to the user. ^1da177e4 Linus Torvalds 2005-04-16 4596 * Packets in sequence go to the receive queue. ^1da177e4 Linus Torvalds 2005-04-16 4597 * Out of sequence packets to the out_of_order_queue. ^1da177e4 Linus Torvalds 2005-04-16 4598 */ ^1da177e4 Linus Torvalds 2005-04-16 4599 if (TCP_SKB_CB(skb)->seq == tp->rcv_nxt) { ^1da177e4 Linus Torvalds 2005-04-16 4600 if (tcp_receive_window(tp) == 0) ^1da177e4 Linus Torvalds 2005-04-16 4601 goto out_of_window; ^1da177e4 Linus Torvalds 2005-04-16 4602 ^1da177e4 Linus Torvalds 2005-04-16 4603 /* Ok. In sequence. In window. */ ^1da177e4 Linus Torvalds 2005-04-16 4604 queue_and_out: 76dfa6082 Eric Dumazet 2015-05-15 4605 if (skb_queue_len(&sk->sk_receive_queue) == 0) 76dfa6082 Eric Dumazet 2015-05-15 4606 sk_forced_mem_schedule(sk, skb->truesize); 76dfa6082 Eric Dumazet 2015-05-15 4607 else if (tcp_try_rmem_schedule(sk, skb, skb->truesize)) ^1da177e4 Linus Torvalds 2005-04-16 4608 goto drop; 5357f0bd4 Eric Dumazet 2017-08-01 4609 b081f85c2 Eric Dumazet 2012-05-02 4610 eaten = tcp_queue_rcv(sk, skb, 0, &fragstolen); bdd1f9eda Eric Dumazet 2015-04-28 4611 tcp_rcv_nxt_update(tp, TCP_SKB_CB(skb)->end_seq); ^1da177e4 Linus Torvalds 2005-04-16 4612 if (skb->len) 9e412ba76 Ilpo Järvinen 2007-04-20 4613 tcp_event_data_recv(sk, skb); 155c6e1ad Peter Pan(潘卫平 2014-09-24 4614) if (TCP_SKB_CB(skb)->tcp_flags & TCPHDR_FIN) 20c4cb792 Eric Dumazet 2011-10-20 4615 tcp_fin(sk); ^1da177e4 Linus Torvalds 2005-04-16 4616 9f5afeae5 Yaogong Wang 2016-09-07 4617 if (!RB_EMPTY_ROOT(&tp->out_of_order_queue)) { ^1da177e4 Linus Torvalds 2005-04-16 4618 tcp_ofo_queue(sk); ^1da177e4 Linus Torvalds 2005-04-16 4619 ^1da177e4 Linus Torvalds 2005-04-16 4620 /* RFC2581. 4.2. SHOULD send immediate ACK, when ^1da177e4 Linus Torvalds 2005-04-16 4621 * gap in queue is filled. ^1da177e4 Linus Torvalds 2005-04-16 4622 */ 9f5afeae5 Yaogong Wang 2016-09-07 4623 if (RB_EMPTY_ROOT(&tp->out_of_order_queue)) 463c84b97 Arnaldo Carvalho de Melo 2005-08-09 4624 inet_csk(sk)->icsk_ack.pingpong = 0; ^1da177e4 Linus Torvalds 2005-04-16 4625 } ^1da177e4 Linus Torvalds 2005-04-16 4626 ^1da177e4 Linus Torvalds 2005-04-16 4627 if (tp->rx_opt.num_sacks) ^1da177e4 Linus Torvalds 2005-04-16 4628 tcp_sack_remove(tp); ^1da177e4 Linus Torvalds 2005-04-16 4629 31770e34e Florian Westphal 2017-08-30 4630 tcp_fast_path_check(sk); 31770e34e Florian Westphal 2017-08-30 4631 923dd347b Eric Dumazet 2012-05-02 4632 if (eaten > 0) 923dd347b Eric Dumazet 2012-05-02 4633 kfree_skb_partial(skb, fragstolen); 1d57f1953 Eric Dumazet 2012-09-17 4634 if (!sock_flag(sk, SOCK_DEAD)) 676d23690 David S. Miller 2014-04-11 4635 sk->sk_data_ready(sk); ^1da177e4 Linus Torvalds 2005-04-16 4636 return; ^1da177e4 Linus Torvalds 2005-04-16 4637 } ^1da177e4 Linus Torvalds 2005-04-16 4638 ^1da177e4 Linus Torvalds 2005-04-16 4639 if (!after(TCP_SKB_CB(skb)->end_seq, tp->rcv_nxt)) { ^1da177e4 Linus Torvalds 2005-04-16 4640 /* A retransmit, 2nd most common case. Force an immediate ack. */ c10d9310e Eric Dumazet 2016-04-29 4641 NET_INC_STATS(sock_net(sk), LINUX_MIB_DELAYEDACKLOST); 1ed834655 Pavel Emelyanov 2008-07-16 4642 tcp_dsack_set(sk, TCP_SKB_CB(skb)->seq, TCP_SKB_CB(skb)->end_seq); ^1da177e4 Linus Torvalds 2005-04-16 4643 ^1da177e4 Linus Torvalds 2005-04-16 4644 out_of_window: 265e9de2c Eric Dumazet 2018-05-21 4645 tcp_enter_quickack_mode(sk, TCP_MAX_QUICKACKS); 463c84b97 Arnaldo Carvalho de Melo 2005-08-09 4646 inet_csk_schedule_ack(sk); ^1da177e4 Linus Torvalds 2005-04-16 4647 drop: 532182cd6 Eric Dumazet 2016-04-01 4648 tcp_drop(sk, skb); ^1da177e4 Linus Torvalds 2005-04-16 4649 return; ^1da177e4 Linus Torvalds 2005-04-16 4650 } ^1da177e4 Linus Torvalds 2005-04-16 4651 ^1da177e4 Linus Torvalds 2005-04-16 4652 /* Out of window. F.e. zero window probe. */ ^1da177e4 Linus Torvalds 2005-04-16 4653 if (!before(TCP_SKB_CB(skb)->seq, tp->rcv_nxt + tcp_receive_window(tp))) ^1da177e4 Linus Torvalds 2005-04-16 4654 goto out_of_window; ^1da177e4 Linus Torvalds 2005-04-16 4655 463c84b97 Arnaldo Carvalho de Melo 2005-08-09 @4656 tcp_enter_quickack_mode(sk); ^1da177e4 Linus Torvalds 2005-04-16 4657 ^1da177e4 Linus Torvalds 2005-04-16 4658 if (before(TCP_SKB_CB(skb)->seq, tp->rcv_nxt)) { ^1da177e4 Linus Torvalds 2005-04-16 4659 /* Partial packet, seq < rcv_next < end_seq */ ^1da177e4 Linus Torvalds 2005-04-16 4660 SOCK_DEBUG(sk, "partial packet: rcv_next %X seq %X - %X\n", ^1da177e4 Linus Torvalds 2005-04-16 4661 tp->rcv_nxt, TCP_SKB_CB(skb)->seq, ^1da177e4 Linus Torvalds 2005-04-16 4662 TCP_SKB_CB(skb)->end_seq); ^1da177e4 Linus Torvalds 2005-04-16 4663 1ed834655 Pavel Emelyanov 2008-07-16 4664 tcp_dsack_set(sk, TCP_SKB_CB(skb)->seq, tp->rcv_nxt); ^1da177e4 Linus Torvalds 2005-04-16 4665 ^1da177e4 Linus Torvalds 2005-04-16 4666 /* If window is closed, drop tail of packet. But after ^1da177e4 Linus Torvalds 2005-04-16 4667 * remembering D-SACK for its head made in previous line. ^1da177e4 Linus Torvalds 2005-04-16 4668 */ ^1da177e4 Linus Torvalds 2005-04-16 4669 if (!tcp_receive_window(tp)) ^1da177e4 Linus Torvalds 2005-04-16 4670 goto out_of_window; ^1da177e4 Linus Torvalds 2005-04-16 4671 goto queue_and_out; ^1da177e4 Linus Torvalds 2005-04-16 4672 } ^1da177e4 Linus Torvalds 2005-04-16 4673 e86b29196 Eric Dumazet 2012-03-18 4674 tcp_data_queue_ofo(sk, skb); ^1da177e4 Linus Torvalds 2005-04-16 4675 } ^1da177e4 Linus Torvalds 2005-04-16 4676 :::::: The code at line 4656 was first introduced by commit :::::: 463c84b97f24010a67cd871746d6a7e4c925a5f9 [NET]: Introduce inet_connection_sock :::::: TO: Arnaldo Carvalho de Melo <acme@ghostprotocols.net> :::::: CC: David S. Miller <davem@sunset.davemloft.net> --- 0-DAY kernel test infrastructure Open Source Technology Center https://lists.01.org/pipermail/kbuild-all Intel Corporation [-- Attachment #2: .config.gz --] [-- Type: application/gzip, Size: 33476 bytes --] ^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: [PATCH net-next 1/2] tcp: add max_quickacks param to tcp_incr_quickack and tcp_enter_quickack_mode 2018-05-23 0:30 ` kbuild test robot @ 2018-05-23 1:37 ` Neal Cardwell 0 siblings, 0 replies; 11+ messages in thread From: Neal Cardwell @ 2018-05-23 1:37 UTC (permalink / raw) To: kbuild test robot Cc: Eric Dumazet, kbuild-all, David Miller, Netdev, Van Jacobson, Yuchung Cheng, Soheil Hassas Yeganeh, Eric Dumazet On Tue, May 22, 2018 at 8:31 PM kbuild test robot <lkp@intel.com> wrote: > Hi Eric, > Thank you for the patch! Yet something to improve: > [auto build test ERROR on net/master] > [also build test ERROR on v4.17-rc6 next-20180517] > [cannot apply to net-next/master] > [if your patch is applied to the wrong git tree, please drop us a note to help improve the system] > url: https://github.com/0day-ci/linux/commits/Eric-Dumazet/tcp-add-max_quickacks-param-to-tcp_incr_quickack-and-tcp_enter_quickack_mode/20180523-075103 > config: i386-randconfig-x012-201820 (attached as .config) > compiler: gcc-7 (Debian 7.3.0-16) 7.3.0 > reproduce: > # save the attached .config to linux build tree > make ARCH=i386 > All errors (new ones prefixed by >>): > net//ipv4/tcp_input.c: In function 'tcp_data_queue': > >> net//ipv4/tcp_input.c:4656:2: error: too few arguments to function 'tcp_enter_quickack_mode' > tcp_enter_quickack_mode(sk); > ^~~~~~~~~~~~~~~~~~~~~~~ > net//ipv4/tcp_input.c:199:13: note: declared here > static void tcp_enter_quickack_mode(struct sock *sk, unsigned int max_quickacks) > ^~~~~~~~~~~~~~~~~~~~~~~ ... For the record, this is an error in the tool, rather than the patch. The tool seems to be using a stale net-next tree for building this patch. The compile error is here in line 4656: > ^1da177e4 Linus Torvalds 2005-04-16 4652 /* Out of window. F.e. zero window probe. */ > ^1da177e4 Linus Torvalds 2005-04-16 4653 if (!before(TCP_SKB_CB(skb)->seq, tp->rcv_nxt + tcp_receive_window(tp))) > ^1da177e4 Linus Torvalds 2005-04-16 4654 goto out_of_window; > ^1da177e4 Linus Torvalds 2005-04-16 4655 > 463c84b97 Arnaldo Carvalho de Melo 2005-08-09 @4656 tcp_enter_quickack_mode(sk); > ^1da177e4 Linus Torvalds 2005-04-16 4657 > ^1da177e4 Linus Torvalds 2005-04-16 4658 if (before(TCP_SKB_CB(skb)->seq, tp->rcv_nxt)) { > ^1da177e4 Linus Torvalds 2005-04-16 4659 /* Partial packet, seq < rcv_next < end_seq */ ... But that line is not in net-next any more, after Eric's recent net-next commit: a3893637e1eb0e ("tcp: do not force quickack when receiving out-of-order packets") https://git.kernel.org/pub/scm/linux/kernel/git/davem/net-next.git/commit/net/ipv4/tcp_input.c?id=a3893637e1eb0ef5eb1bbc52b3a8d2dfa317a35d That commit removed that line: diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 0bf032839548f..f5622b2506651 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -4715,8 +4715,6 @@ drop: if (!before(TCP_SKB_CB(skb)->seq, tp->rcv_nxt + tcp_receive_window(tp))) goto out_of_window; - tcp_enter_quickack_mode(sk); - if (before(TCP_SKB_CB(skb)->seq, tp->rcv_nxt)) { /* Partial packet, seq < rcv_next < end_seq */ SOCK_DEBUG(sk, "partial packet: rcv_next %X seq %X - %X\n", cheers, neal ^ permalink raw reply related [flat|nested] 11+ messages in thread
* [PATCH net-next 2/2] tcp: do not aggressively quick ack after ECN events 2018-05-21 22:08 [PATCH net-next 0/2] tcp: reduce quickack pressure for ECN Eric Dumazet 2018-05-21 22:08 ` [PATCH net-next 1/2] tcp: add max_quickacks param to tcp_incr_quickack and tcp_enter_quickack_mode Eric Dumazet @ 2018-05-21 22:08 ` Eric Dumazet 2018-05-22 14:04 ` Soheil Hassas Yeganeh ` (2 more replies) 2018-05-22 19:43 ` [PATCH net-next 0/2] tcp: reduce quickack pressure for ECN David Miller 2 siblings, 3 replies; 11+ messages in thread From: Eric Dumazet @ 2018-05-21 22:08 UTC (permalink / raw) To: David S . Miller Cc: netdev, Van Jacobson, Neal Cardwell, Yuchung Cheng, Soheil Hassas Yeganeh, Eric Dumazet, Eric Dumazet ECN signals currently forces TCP to enter quickack mode for up to 16 (TCP_MAX_QUICKACKS) following incoming packets. We believe this is not needed, and only sending one immediate ack for the current packet should be enough. This should reduce the extra load noticed in DCTCP environments, after congestion events. This is part 2 of our effort to reduce pure ACK packets. Signed-off-by: Eric Dumazet <edumazet@google.com> --- net/ipv4/tcp_input.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 2e970e9f4e09d966b703af2d14d521a4328eba7e..1191cac72109f2f7e2b688ddbc1d404151d274d6 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -263,7 +263,7 @@ static void __tcp_ecn_check_ce(struct tcp_sock *tp, const struct sk_buff *skb) * it is probably a retransmit. */ if (tp->ecn_flags & TCP_ECN_SEEN) - tcp_enter_quickack_mode((struct sock *)tp, TCP_MAX_QUICKACKS); + tcp_enter_quickack_mode((struct sock *)tp, 1); break; case INET_ECN_CE: if (tcp_ca_needs_ecn((struct sock *)tp)) @@ -271,7 +271,7 @@ static void __tcp_ecn_check_ce(struct tcp_sock *tp, const struct sk_buff *skb) if (!(tp->ecn_flags & TCP_ECN_DEMAND_CWR)) { /* Better not delay acks, sender can have a very low cwnd */ - tcp_enter_quickack_mode((struct sock *)tp, TCP_MAX_QUICKACKS); + tcp_enter_quickack_mode((struct sock *)tp, 1); tp->ecn_flags |= TCP_ECN_DEMAND_CWR; } tp->ecn_flags |= TCP_ECN_SEEN; -- 2.17.0.441.gb46fe60e1d-goog ^ permalink raw reply related [flat|nested] 11+ messages in thread
* Re: [PATCH net-next 2/2] tcp: do not aggressively quick ack after ECN events 2018-05-21 22:08 ` [PATCH net-next 2/2] tcp: do not aggressively quick ack after ECN events Eric Dumazet @ 2018-05-22 14:04 ` Soheil Hassas Yeganeh 2018-05-22 16:54 ` Yuchung Cheng 2018-05-22 17:36 ` Neal Cardwell 2 siblings, 0 replies; 11+ messages in thread From: Soheil Hassas Yeganeh @ 2018-05-22 14:04 UTC (permalink / raw) To: Eric Dumazet Cc: David S . Miller, netdev, Van Jacobson, Neal Cardwell, Yuchung Cheng, Eric Dumazet On Mon, May 21, 2018 at 6:08 PM, Eric Dumazet <edumazet@google.com> wrote: > ECN signals currently forces TCP to enter quickack mode for > up to 16 (TCP_MAX_QUICKACKS) following incoming packets. > > We believe this is not needed, and only sending one immediate ack > for the current packet should be enough. > > This should reduce the extra load noticed in DCTCP environments, > after congestion events. > > This is part 2 of our effort to reduce pure ACK packets. > > Signed-off-by: Eric Dumazet <edumazet@google.com> Acked-by: Soheil Hassas Yeganeh <soheil@google.com> Thanks for the patch! ^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: [PATCH net-next 2/2] tcp: do not aggressively quick ack after ECN events 2018-05-21 22:08 ` [PATCH net-next 2/2] tcp: do not aggressively quick ack after ECN events Eric Dumazet 2018-05-22 14:04 ` Soheil Hassas Yeganeh @ 2018-05-22 16:54 ` Yuchung Cheng 2018-05-22 17:36 ` Neal Cardwell 2 siblings, 0 replies; 11+ messages in thread From: Yuchung Cheng @ 2018-05-22 16:54 UTC (permalink / raw) To: Eric Dumazet Cc: David S . Miller, netdev, Van Jacobson, Neal Cardwell, Soheil Hassas Yeganeh, Eric Dumazet On Mon, May 21, 2018 at 3:08 PM, Eric Dumazet <edumazet@google.com> wrote: > ECN signals currently forces TCP to enter quickack mode for > up to 16 (TCP_MAX_QUICKACKS) following incoming packets. > > We believe this is not needed, and only sending one immediate ack > for the current packet should be enough. > > This should reduce the extra load noticed in DCTCP environments, > after congestion events. > > This is part 2 of our effort to reduce pure ACK packets. > > Signed-off-by: Eric Dumazet <edumazet@google.com> > --- Acked-by: Yuchung Cheng <ycheng@google.com> Thanks for this patch. I am still wondering how much does the "funny extension" help. but this patch definitely reduce the amount of unnecessary immediate ACKs on ECN. > net/ipv4/tcp_input.c | 4 ++-- > 1 file changed, 2 insertions(+), 2 deletions(-) > > diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c > index 2e970e9f4e09d966b703af2d14d521a4328eba7e..1191cac72109f2f7e2b688ddbc1d404151d274d6 100644 > --- a/net/ipv4/tcp_input.c > +++ b/net/ipv4/tcp_input.c > @@ -263,7 +263,7 @@ static void __tcp_ecn_check_ce(struct tcp_sock *tp, const struct sk_buff *skb) > * it is probably a retransmit. > */ > if (tp->ecn_flags & TCP_ECN_SEEN) > - tcp_enter_quickack_mode((struct sock *)tp, TCP_MAX_QUICKACKS); > + tcp_enter_quickack_mode((struct sock *)tp, 1); > break; > case INET_ECN_CE: > if (tcp_ca_needs_ecn((struct sock *)tp)) > @@ -271,7 +271,7 @@ static void __tcp_ecn_check_ce(struct tcp_sock *tp, const struct sk_buff *skb) > > if (!(tp->ecn_flags & TCP_ECN_DEMAND_CWR)) { > /* Better not delay acks, sender can have a very low cwnd */ > - tcp_enter_quickack_mode((struct sock *)tp, TCP_MAX_QUICKACKS); > + tcp_enter_quickack_mode((struct sock *)tp, 1); > tp->ecn_flags |= TCP_ECN_DEMAND_CWR; > } > tp->ecn_flags |= TCP_ECN_SEEN; > -- > 2.17.0.441.gb46fe60e1d-goog > ^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: [PATCH net-next 2/2] tcp: do not aggressively quick ack after ECN events 2018-05-21 22:08 ` [PATCH net-next 2/2] tcp: do not aggressively quick ack after ECN events Eric Dumazet 2018-05-22 14:04 ` Soheil Hassas Yeganeh 2018-05-22 16:54 ` Yuchung Cheng @ 2018-05-22 17:36 ` Neal Cardwell 2 siblings, 0 replies; 11+ messages in thread From: Neal Cardwell @ 2018-05-22 17:36 UTC (permalink / raw) To: Eric Dumazet Cc: David Miller, Netdev, Van Jacobson, Yuchung Cheng, Soheil Hassas Yeganeh, Eric Dumazet On Mon, May 21, 2018 at 6:09 PM Eric Dumazet <edumazet@google.com> wrote: > ECN signals currently forces TCP to enter quickack mode for > up to 16 (TCP_MAX_QUICKACKS) following incoming packets. > We believe this is not needed, and only sending one immediate ack > for the current packet should be enough. > This should reduce the extra load noticed in DCTCP environments, > after congestion events. > This is part 2 of our effort to reduce pure ACK packets. > Signed-off-by: Eric Dumazet <edumazet@google.com> > --- Acked-by: Neal Cardwell <ncardwell@google.com> Thanks! neal ^ permalink raw reply [flat|nested] 11+ messages in thread
* Re: [PATCH net-next 0/2] tcp: reduce quickack pressure for ECN 2018-05-21 22:08 [PATCH net-next 0/2] tcp: reduce quickack pressure for ECN Eric Dumazet 2018-05-21 22:08 ` [PATCH net-next 1/2] tcp: add max_quickacks param to tcp_incr_quickack and tcp_enter_quickack_mode Eric Dumazet 2018-05-21 22:08 ` [PATCH net-next 2/2] tcp: do not aggressively quick ack after ECN events Eric Dumazet @ 2018-05-22 19:43 ` David Miller 2 siblings, 0 replies; 11+ messages in thread From: David Miller @ 2018-05-22 19:43 UTC (permalink / raw) To: edumazet; +Cc: netdev, vanj, ncardwell, ycheng, soheil, eric.dumazet From: Eric Dumazet <edumazet@google.com> Date: Mon, 21 May 2018 15:08:55 -0700 > Small patch series changing TCP behavior vs quickack and ECN > > First patch is a refactoring, adding parameter to tcp_incr_quickack() > and tcp_enter_quickack_mode() helpers. > > Second patch implements the change, lowering number of ACK packets > sent after an ECN event. Series applied, thanks Eric. ^ permalink raw reply [flat|nested] 11+ messages in thread
end of thread, other threads:[~2018-05-23 1:37 UTC | newest] Thread overview: 11+ messages (download: mbox.gz / follow: Atom feed) -- links below jump to the message on this page -- 2018-05-21 22:08 [PATCH net-next 0/2] tcp: reduce quickack pressure for ECN Eric Dumazet 2018-05-21 22:08 ` [PATCH net-next 1/2] tcp: add max_quickacks param to tcp_incr_quickack and tcp_enter_quickack_mode Eric Dumazet 2018-05-22 14:03 ` Soheil Hassas Yeganeh 2018-05-22 17:27 ` Neal Cardwell 2018-05-23 0:30 ` kbuild test robot 2018-05-23 1:37 ` Neal Cardwell 2018-05-21 22:08 ` [PATCH net-next 2/2] tcp: do not aggressively quick ack after ECN events Eric Dumazet 2018-05-22 14:04 ` Soheil Hassas Yeganeh 2018-05-22 16:54 ` Yuchung Cheng 2018-05-22 17:36 ` Neal Cardwell 2018-05-22 19:43 ` [PATCH net-next 0/2] tcp: reduce quickack pressure for ECN David Miller
This is a public inbox, see mirroring instructions for how to clone and mirror all data and code used for this inbox; as well as URLs for NNTP newsgroup(s).