From d635fbe27ebee0f4b845abe5e9620c9400785a5c Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Tue, 16 May 2017 14:00:03 -0700 Subject: tcp: use tcp_jiffies32 to feed tp->lsndtime Use tcp_jiffies32 instead of tcp_time_stamp to feed tp->lsndtime. tcp_time_stamp will soon be a litle bit more expensive than simply reading 'jiffies'. Signed-off-by: Eric Dumazet Acked-by: Soheil Hassas Yeganeh Signed-off-by: David S. Miller --- net/ipv4/tcp_input.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'net/ipv4/tcp_input.c') diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 06e2dbc2b4a2..c0b3f909df39 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -5571,7 +5571,7 @@ void tcp_finish_connect(struct sock *sk, struct sk_buff *skb) /* Prevent spurious tcp_cwnd_restart() on first data * packet. */ - tp->lsndtime = tcp_time_stamp; + tp->lsndtime = tcp_jiffies32; tcp_init_buffer_space(sk); @@ -6008,7 +6008,7 @@ int tcp_rcv_state_process(struct sock *sk, struct sk_buff *skb) tcp_update_pacing_rate(sk); /* Prevent spurious tcp_cwnd_restart() on first data packet */ - tp->lsndtime = tcp_time_stamp; + tp->lsndtime = tcp_jiffies32; tcp_initialize_rcv_mss(sk); tcp_fast_path_on(tp); -- cgit v1.2.3 From c2203cf75ed7dfab8dfc7ac915a726880ee7512f Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Tue, 16 May 2017 14:00:04 -0700 Subject: tcp: use tcp_jiffies32 to feed tp->snd_cwnd_stamp Use tcp_jiffies32 instead of tcp_time_stamp to feed tp->snd_cwnd_stamp. tcp_time_stamp will soon be a litle bit more expensive than simply reading 'jiffies'. Signed-off-by: Eric Dumazet Acked-by: Soheil Hassas Yeganeh Signed-off-by: David S. Miller --- net/ipv4/tcp_input.c | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-) (limited to 'net/ipv4/tcp_input.c') diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index c0b3f909df39..6a15c9b80b09 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -463,7 +463,7 @@ void tcp_init_buffer_space(struct sock *sk) tp->window_clamp = max(2 * tp->advmss, maxwin - tp->advmss); tp->rcv_ssthresh = min(tp->rcv_ssthresh, tp->window_clamp); - tp->snd_cwnd_stamp = tcp_time_stamp; + tp->snd_cwnd_stamp = tcp_jiffies32; } /* 5. Recalculate window clamp after socket hit its memory bounds. */ @@ -1954,7 +1954,7 @@ void tcp_enter_loss(struct sock *sk) } tp->snd_cwnd = 1; tp->snd_cwnd_cnt = 0; - tp->snd_cwnd_stamp = tcp_time_stamp; + tp->snd_cwnd_stamp = tcp_jiffies32; tp->retrans_out = 0; tp->lost_out = 0; @@ -2383,7 +2383,7 @@ static void tcp_undo_cwnd_reduction(struct sock *sk, bool unmark_loss) tcp_ecn_withdraw_cwr(tp); } } - tp->snd_cwnd_stamp = tcp_time_stamp; + tp->snd_cwnd_stamp = tcp_jiffies32; tp->undo_marker = 0; } @@ -2520,7 +2520,7 @@ static inline void tcp_end_cwnd_reduction(struct sock *sk) if (inet_csk(sk)->icsk_ca_state == TCP_CA_CWR || (tp->undo_marker && tp->snd_ssthresh < TCP_INFINITE_SSTHRESH)) { tp->snd_cwnd = tp->snd_ssthresh; - tp->snd_cwnd_stamp = tcp_time_stamp; + tp->snd_cwnd_stamp = tcp_jiffies32; } tcp_ca_event(sk, CA_EVENT_COMPLETE_CWR); } @@ -2590,7 +2590,7 @@ static void tcp_mtup_probe_success(struct sock *sk) tcp_mss_to_mtu(sk, tp->mss_cache) / icsk->icsk_mtup.probe_size; tp->snd_cwnd_cnt = 0; - tp->snd_cwnd_stamp = tcp_time_stamp; + tp->snd_cwnd_stamp = tcp_jiffies32; tp->snd_ssthresh = tcp_current_ssthresh(sk); icsk->icsk_mtup.search_low = icsk->icsk_mtup.probe_size; @@ -2976,7 +2976,7 @@ static void tcp_cong_avoid(struct sock *sk, u32 ack, u32 acked) const struct inet_connection_sock *icsk = inet_csk(sk); icsk->icsk_ca_ops->cong_avoid(sk, ack, acked); - tcp_sk(sk)->snd_cwnd_stamp = tcp_time_stamp; + tcp_sk(sk)->snd_cwnd_stamp = tcp_jiffies32; } /* Restart timer after forward progress on connection. @@ -5019,7 +5019,7 @@ static void tcp_new_space(struct sock *sk) if (tcp_should_expand_sndbuf(sk)) { tcp_sndbuf_expand(sk); - tp->snd_cwnd_stamp = tcp_time_stamp; + tp->snd_cwnd_stamp = tcp_jiffies32; } sk->sk_write_space(sk); -- cgit v1.2.3 From 70eabf0e1b8fe11519f793416655266605f700b9 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Tue, 16 May 2017 14:00:07 -0700 Subject: tcp: use tcp_jiffies32 for rcv_tstamp and lrcvtime Use tcp_jiffies32 instead of tcp_time_stamp, since tcp_time_stamp will soon be only used for TCP TS option. Signed-off-by: Eric Dumazet Acked-by: Soheil Hassas Yeganeh Signed-off-by: David S. Miller --- net/ipv4/tcp_input.c | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'net/ipv4/tcp_input.c') diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 6a15c9b80b09..eeb4967df25a 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -672,7 +672,7 @@ static void tcp_event_data_recv(struct sock *sk, struct sk_buff *skb) tcp_rcv_rtt_measure(tp); - now = tcp_time_stamp; + now = tcp_jiffies32; if (!icsk->icsk_ack.ato) { /* The _first_ data packet received, initialize @@ -3636,7 +3636,7 @@ static int tcp_ack(struct sock *sk, const struct sk_buff *skb, int flag) */ sk->sk_err_soft = 0; icsk->icsk_probes_out = 0; - tp->rcv_tstamp = tcp_time_stamp; + tp->rcv_tstamp = tcp_jiffies32; if (!prior_packets) goto no_queue; @@ -5554,7 +5554,7 @@ void tcp_finish_connect(struct sock *sk, struct sk_buff *skb) struct inet_connection_sock *icsk = inet_csk(sk); tcp_set_state(sk, TCP_ESTABLISHED); - icsk->icsk_ack.lrcvtime = tcp_time_stamp; + icsk->icsk_ack.lrcvtime = tcp_jiffies32; if (skb) { icsk->icsk_af_ops->sk_rx_dst_set(sk, skb); -- cgit v1.2.3 From 594208afe40c448faca967235691ec04fe9f57e3 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Tue, 16 May 2017 14:00:10 -0700 Subject: tcp: use tcp_jiffies32 in __tcp_oow_rate_limited() This place wants to use tcp_jiffies32, this is good enough. Signed-off-by: Eric Dumazet Acked-by: Soheil Hassas Yeganeh Signed-off-by: David S. Miller --- net/ipv4/tcp_input.c | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) (limited to 'net/ipv4/tcp_input.c') diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index eeb4967df25a..85575888365a 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -3390,7 +3390,7 @@ static bool __tcp_oow_rate_limited(struct net *net, int mib_idx, u32 *last_oow_ack_time) { if (*last_oow_ack_time) { - s32 elapsed = (s32)(tcp_time_stamp - *last_oow_ack_time); + s32 elapsed = (s32)(tcp_jiffies32 - *last_oow_ack_time); if (0 <= elapsed && elapsed < sysctl_tcp_invalid_ratelimit) { NET_INC_STATS(net, mib_idx); @@ -3398,7 +3398,7 @@ static bool __tcp_oow_rate_limited(struct net *net, int mib_idx, } } - *last_oow_ack_time = tcp_time_stamp; + *last_oow_ack_time = tcp_jiffies32; return false; /* not rate-limited: go ahead, send dupack now! */ } -- cgit v1.2.3 From ac9517fcf310327fa3e3b0d8366e4b11236b1b4b Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Tue, 16 May 2017 14:00:13 -0700 Subject: tcp: replace misc tcp_time_stamp to tcp_jiffies32 After this patch, all uses of tcp_time_stamp will require a change when we introduce 1 ms and/or 1 us TCP TS option. Signed-off-by: Eric Dumazet Acked-by: Soheil Hassas Yeganeh Signed-off-by: David S. Miller --- net/ipv4/tcp_input.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) (limited to 'net/ipv4/tcp_input.c') diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 85575888365a..10e6775464f6 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -2911,7 +2911,7 @@ static void tcp_update_rtt_min(struct sock *sk, u32 rtt_us) struct tcp_sock *tp = tcp_sk(sk); u32 wlen = sysctl_tcp_min_rtt_wlen * HZ; - minmax_running_min(&tp->rtt_min, wlen, tcp_time_stamp, + minmax_running_min(&tp->rtt_min, wlen, tcp_jiffies32, rtt_us ? : jiffies_to_usecs(1)); } -- cgit v1.2.3 From 9a568de4818dea9a05af141046bd3e589245ab83 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Tue, 16 May 2017 14:00:14 -0700 Subject: tcp: switch TCP TS option (RFC 7323) to 1ms clock TCP Timestamps option is defined in RFC 7323 Traditionally on linux, it has been tied to the internal 'jiffies' variable, because it had been a cheap and good enough generator. For TCP flows on the Internet, 1 ms resolution would be much better than 4ms or 10ms (HZ=250 or HZ=100 respectively) For TCP flows in the DC, Google has used usec resolution for more than two years with great success [1] Receive size autotuning (DRS) is indeed more precise and converges faster to optimal window size. This patch converts tp->tcp_mstamp to a plain u64 value storing a 1 usec TCP clock. This choice will allow us to upstream the 1 usec TS option as discussed in IETF 97. [1] https://www.ietf.org/proceedings/97/slides/slides-97-tcpm-tcp-options-for-low-latency-00.pdf Signed-off-by: Eric Dumazet Acked-by: Soheil Hassas Yeganeh Signed-off-by: David S. Miller --- net/ipv4/tcp_input.c | 96 ++++++++++++++++++++++++++-------------------------- 1 file changed, 48 insertions(+), 48 deletions(-) (limited to 'net/ipv4/tcp_input.c') diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 10e6775464f6..9a5a9e8eda89 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -441,7 +441,7 @@ void tcp_init_buffer_space(struct sock *sk) tcp_sndbuf_expand(sk); tp->rcvq_space.space = tp->rcv_wnd; - skb_mstamp_get(&tp->tcp_mstamp); + tcp_mstamp_refresh(tp); tp->rcvq_space.time = tp->tcp_mstamp; tp->rcvq_space.seq = tp->copied_seq; @@ -555,11 +555,11 @@ static inline void tcp_rcv_rtt_measure(struct tcp_sock *tp) { u32 delta_us; - if (tp->rcv_rtt_est.time.v64 == 0) + if (tp->rcv_rtt_est.time == 0) goto new_measure; if (before(tp->rcv_nxt, tp->rcv_rtt_est.seq)) return; - delta_us = skb_mstamp_us_delta(&tp->tcp_mstamp, &tp->rcv_rtt_est.time); + delta_us = tcp_stamp_us_delta(tp->tcp_mstamp, tp->rcv_rtt_est.time); tcp_rcv_rtt_update(tp, delta_us, 1); new_measure: @@ -571,13 +571,15 @@ static inline void tcp_rcv_rtt_measure_ts(struct sock *sk, const struct sk_buff *skb) { struct tcp_sock *tp = tcp_sk(sk); + if (tp->rx_opt.rcv_tsecr && (TCP_SKB_CB(skb)->end_seq - - TCP_SKB_CB(skb)->seq >= inet_csk(sk)->icsk_ack.rcv_mss)) - tcp_rcv_rtt_update(tp, - jiffies_to_usecs(tcp_time_stamp - - tp->rx_opt.rcv_tsecr), - 0); + TCP_SKB_CB(skb)->seq >= inet_csk(sk)->icsk_ack.rcv_mss)) { + u32 delta = tcp_time_stamp(tp) - tp->rx_opt.rcv_tsecr; + u32 delta_us = delta * (USEC_PER_SEC / TCP_TS_HZ); + + tcp_rcv_rtt_update(tp, delta_us, 0); + } } /* @@ -590,7 +592,7 @@ void tcp_rcv_space_adjust(struct sock *sk) int time; int copied; - time = skb_mstamp_us_delta(&tp->tcp_mstamp, &tp->rcvq_space.time); + time = tcp_stamp_us_delta(tp->tcp_mstamp, tp->rcvq_space.time); if (time < (tp->rcv_rtt_est.rtt_us >> 3) || tp->rcv_rtt_est.rtt_us == 0) return; @@ -1134,8 +1136,8 @@ struct tcp_sacktag_state { * that was SACKed. RTO needs the earliest RTT to stay conservative, * but congestion control should still get an accurate delay signal. */ - struct skb_mstamp first_sackt; - struct skb_mstamp last_sackt; + u64 first_sackt; + u64 last_sackt; struct rate_sample *rate; int flag; }; @@ -1200,7 +1202,7 @@ static u8 tcp_sacktag_one(struct sock *sk, struct tcp_sacktag_state *state, u8 sacked, u32 start_seq, u32 end_seq, int dup_sack, int pcount, - const struct skb_mstamp *xmit_time) + u64 xmit_time) { struct tcp_sock *tp = tcp_sk(sk); int fack_count = state->fack_count; @@ -1242,9 +1244,9 @@ static u8 tcp_sacktag_one(struct sock *sk, state->reord); if (!after(end_seq, tp->high_seq)) state->flag |= FLAG_ORIG_SACK_ACKED; - if (state->first_sackt.v64 == 0) - state->first_sackt = *xmit_time; - state->last_sackt = *xmit_time; + if (state->first_sackt == 0) + state->first_sackt = xmit_time; + state->last_sackt = xmit_time; } if (sacked & TCPCB_LOST) { @@ -1304,7 +1306,7 @@ static bool tcp_shifted_skb(struct sock *sk, struct sk_buff *skb, */ tcp_sacktag_one(sk, state, TCP_SKB_CB(skb)->sacked, start_seq, end_seq, dup_sack, pcount, - &skb->skb_mstamp); + skb->skb_mstamp); tcp_rate_skb_delivered(sk, skb, state->rate); if (skb == tp->lost_skb_hint) @@ -1356,8 +1358,8 @@ static bool tcp_shifted_skb(struct sock *sk, struct sk_buff *skb, tcp_advance_highest_sack(sk, skb); tcp_skb_collapse_tstamp(prev, skb); - if (unlikely(TCP_SKB_CB(prev)->tx.delivered_mstamp.v64)) - TCP_SKB_CB(prev)->tx.delivered_mstamp.v64 = 0; + if (unlikely(TCP_SKB_CB(prev)->tx.delivered_mstamp)) + TCP_SKB_CB(prev)->tx.delivered_mstamp = 0; tcp_unlink_write_queue(skb, sk); sk_wmem_free_skb(sk, skb); @@ -1587,7 +1589,7 @@ static struct sk_buff *tcp_sacktag_walk(struct sk_buff *skb, struct sock *sk, TCP_SKB_CB(skb)->end_seq, dup_sack, tcp_skb_pcount(skb), - &skb->skb_mstamp); + skb->skb_mstamp); tcp_rate_skb_delivered(sk, skb, state->rate); if (!before(TCP_SKB_CB(skb)->seq, @@ -2936,9 +2938,12 @@ static inline bool tcp_ack_update_rtt(struct sock *sk, const int flag, * See draft-ietf-tcplw-high-performance-00, section 3.3. */ if (seq_rtt_us < 0 && tp->rx_opt.saw_tstamp && tp->rx_opt.rcv_tsecr && - flag & FLAG_ACKED) - seq_rtt_us = ca_rtt_us = jiffies_to_usecs(tcp_time_stamp - - tp->rx_opt.rcv_tsecr); + flag & FLAG_ACKED) { + u32 delta = tcp_time_stamp(tp) - tp->rx_opt.rcv_tsecr; + u32 delta_us = delta * (USEC_PER_SEC / TCP_TS_HZ); + + seq_rtt_us = ca_rtt_us = delta_us; + } if (seq_rtt_us < 0) return false; @@ -2960,12 +2965,8 @@ void tcp_synack_rtt_meas(struct sock *sk, struct request_sock *req) { long rtt_us = -1L; - if (req && !req->num_retrans && tcp_rsk(req)->snt_synack.v64) { - struct skb_mstamp now; - - skb_mstamp_get(&now); - rtt_us = skb_mstamp_us_delta(&now, &tcp_rsk(req)->snt_synack); - } + if (req && !req->num_retrans && tcp_rsk(req)->snt_synack) + rtt_us = tcp_stamp_us_delta(tcp_clock_us(), tcp_rsk(req)->snt_synack); tcp_ack_update_rtt(sk, FLAG_SYN_ACKED, rtt_us, -1L, rtt_us); } @@ -3003,7 +3004,7 @@ void tcp_rearm_rto(struct sock *sk) struct sk_buff *skb = tcp_write_queue_head(sk); const u32 rto_time_stamp = tcp_skb_timestamp(skb) + rto; - s32 delta = (s32)(rto_time_stamp - tcp_time_stamp); + s32 delta = (s32)(rto_time_stamp - tcp_jiffies32); /* delta may not be positive if the socket is locked * when the retrans timer fires and is rescheduled. */ @@ -3060,9 +3061,8 @@ static int tcp_clean_rtx_queue(struct sock *sk, int prior_fackets, struct tcp_sacktag_state *sack) { const struct inet_connection_sock *icsk = inet_csk(sk); - struct skb_mstamp first_ackt, last_ackt; + u64 first_ackt, last_ackt; struct tcp_sock *tp = tcp_sk(sk); - struct skb_mstamp *now = &tp->tcp_mstamp; u32 prior_sacked = tp->sacked_out; u32 reord = tp->packets_out; bool fully_acked = true; @@ -3075,7 +3075,7 @@ static int tcp_clean_rtx_queue(struct sock *sk, int prior_fackets, bool rtt_update; int flag = 0; - first_ackt.v64 = 0; + first_ackt = 0; while ((skb = tcp_write_queue_head(sk)) && skb != tcp_send_head(sk)) { struct tcp_skb_cb *scb = TCP_SKB_CB(skb); @@ -3106,8 +3106,8 @@ static int tcp_clean_rtx_queue(struct sock *sk, int prior_fackets, flag |= FLAG_RETRANS_DATA_ACKED; } else if (!(sacked & TCPCB_SACKED_ACKED)) { last_ackt = skb->skb_mstamp; - WARN_ON_ONCE(last_ackt.v64 == 0); - if (!first_ackt.v64) + WARN_ON_ONCE(last_ackt == 0); + if (!first_ackt) first_ackt = last_ackt; last_in_flight = TCP_SKB_CB(skb)->tx.in_flight; @@ -3122,7 +3122,7 @@ static int tcp_clean_rtx_queue(struct sock *sk, int prior_fackets, tp->delivered += acked_pcount; if (!tcp_skb_spurious_retrans(tp, skb)) tcp_rack_advance(tp, sacked, scb->end_seq, - &skb->skb_mstamp); + skb->skb_mstamp); } if (sacked & TCPCB_LOST) tp->lost_out -= acked_pcount; @@ -3165,13 +3165,13 @@ static int tcp_clean_rtx_queue(struct sock *sk, int prior_fackets, if (skb && (TCP_SKB_CB(skb)->sacked & TCPCB_SACKED_ACKED)) flag |= FLAG_SACK_RENEGING; - if (likely(first_ackt.v64) && !(flag & FLAG_RETRANS_DATA_ACKED)) { - seq_rtt_us = skb_mstamp_us_delta(now, &first_ackt); - ca_rtt_us = skb_mstamp_us_delta(now, &last_ackt); + if (likely(first_ackt) && !(flag & FLAG_RETRANS_DATA_ACKED)) { + seq_rtt_us = tcp_stamp_us_delta(tp->tcp_mstamp, first_ackt); + ca_rtt_us = tcp_stamp_us_delta(tp->tcp_mstamp, last_ackt); } - if (sack->first_sackt.v64) { - sack_rtt_us = skb_mstamp_us_delta(now, &sack->first_sackt); - ca_rtt_us = skb_mstamp_us_delta(now, &sack->last_sackt); + if (sack->first_sackt) { + sack_rtt_us = tcp_stamp_us_delta(tp->tcp_mstamp, sack->first_sackt); + ca_rtt_us = tcp_stamp_us_delta(tp->tcp_mstamp, sack->last_sackt); } sack->rate->rtt_us = ca_rtt_us; /* RTT of last (S)ACKed packet, or -1 */ rtt_update = tcp_ack_update_rtt(sk, flag, seq_rtt_us, sack_rtt_us, @@ -3201,7 +3201,7 @@ static int tcp_clean_rtx_queue(struct sock *sk, int prior_fackets, tp->fackets_out -= min(pkts_acked, tp->fackets_out); } else if (skb && rtt_update && sack_rtt_us >= 0 && - sack_rtt_us > skb_mstamp_us_delta(now, &skb->skb_mstamp)) { + sack_rtt_us > tcp_stamp_us_delta(tp->tcp_mstamp, skb->skb_mstamp)) { /* Do not re-arm RTO if the sack RTT is measured from data sent * after when the head was last (re)transmitted. Otherwise the * timeout may continue to extend in loss recovery. @@ -3553,7 +3553,7 @@ static int tcp_ack(struct sock *sk, const struct sk_buff *skb, int flag) int acked = 0; /* Number of packets newly acked */ int rexmit = REXMIT_NONE; /* Flag to (re)transmit to recover losses */ - sack_state.first_sackt.v64 = 0; + sack_state.first_sackt = 0; sack_state.rate = &rs; /* We very likely will need to access write queue head. */ @@ -5356,7 +5356,7 @@ void tcp_rcv_established(struct sock *sk, struct sk_buff *skb, { struct tcp_sock *tp = tcp_sk(sk); - skb_mstamp_get(&tp->tcp_mstamp); + tcp_mstamp_refresh(tp); if (unlikely(!sk->sk_rx_dst)) inet_csk(sk)->icsk_af_ops->sk_rx_dst_set(sk, skb); /* @@ -5672,7 +5672,7 @@ static int tcp_rcv_synsent_state_process(struct sock *sk, struct sk_buff *skb, if (tp->rx_opt.saw_tstamp && tp->rx_opt.rcv_tsecr && !between(tp->rx_opt.rcv_tsecr, tp->retrans_stamp, - tcp_time_stamp)) { + tcp_time_stamp(tp))) { NET_INC_STATS(sock_net(sk), LINUX_MIB_PAWSACTIVEREJECTED); goto reset_and_undo; @@ -5917,7 +5917,7 @@ int tcp_rcv_state_process(struct sock *sk, struct sk_buff *skb) case TCP_SYN_SENT: tp->rx_opt.saw_tstamp = 0; - skb_mstamp_get(&tp->tcp_mstamp); + tcp_mstamp_refresh(tp); queued = tcp_rcv_synsent_state_process(sk, skb, th); if (queued >= 0) return queued; @@ -5929,7 +5929,7 @@ int tcp_rcv_state_process(struct sock *sk, struct sk_buff *skb) return 0; } - skb_mstamp_get(&tp->tcp_mstamp); + tcp_mstamp_refresh(tp); tp->rx_opt.saw_tstamp = 0; req = tp->fastopen_rsk; if (req) { @@ -6202,7 +6202,7 @@ static void tcp_openreq_init(struct request_sock *req, req->cookie_ts = 0; tcp_rsk(req)->rcv_isn = TCP_SKB_CB(skb)->seq; tcp_rsk(req)->rcv_nxt = TCP_SKB_CB(skb)->seq + 1; - skb_mstamp_get(&tcp_rsk(req)->snt_synack); + tcp_rsk(req)->snt_synack = tcp_clock_us(); tcp_rsk(req)->last_oow_ack_time = 0; req->mss = rx_opt->mss_clamp; req->ts_recent = rx_opt->saw_tstamp ? rx_opt->rcv_tsval : 0; -- cgit v1.2.3 From b17b8a20c5cd4a264601eacf1fda29008047d05a Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Thu, 18 May 2017 09:15:58 -0700 Subject: tcp: fix tcp_rearm_rto() skbs in (re)transmit queue no longer have a copy of jiffies at the time of the transmit : skb->skb_mstamp is now in usec unit, with no correlation to tcp_jiffies32. We have to convert rto from jiffies to usec, compute a time difference in usec, then convert the delta to HZ units. Fixes: 9a568de4818d ("tcp: switch TCP TS option (RFC 7323) to 1ms clock") Signed-off-by: Eric Dumazet Acked-by: Soheil Hassas Yeganeh Signed-off-by: David S. Miller --- net/ipv4/tcp_input.c | 12 ++++++------ 1 file changed, 6 insertions(+), 6 deletions(-) (limited to 'net/ipv4/tcp_input.c') diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 9a5a9e8eda89..aa1eef150dc4 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -3002,14 +3002,14 @@ void tcp_rearm_rto(struct sock *sk) if (icsk->icsk_pending == ICSK_TIME_REO_TIMEOUT || icsk->icsk_pending == ICSK_TIME_LOSS_PROBE) { struct sk_buff *skb = tcp_write_queue_head(sk); - const u32 rto_time_stamp = - tcp_skb_timestamp(skb) + rto; - s32 delta = (s32)(rto_time_stamp - tcp_jiffies32); - /* delta may not be positive if the socket is locked + u64 rto_time_stamp = skb->skb_mstamp + + jiffies_to_usecs(rto); + s64 delta_us = rto_time_stamp - tp->tcp_mstamp; + /* delta_us may not be positive if the socket is locked * when the retrans timer fires and is rescheduled. */ - if (delta > 0) - rto = delta; + if (delta_us > 0) + rto = usecs_to_jiffies(delta_us); } inet_csk_reset_xmit_timer(sk, ICSK_TIME_RETRANS, rto, TCP_RTO_MAX); -- cgit v1.2.3 From 6f5b24eed0278136c29c27f2a7b3a2b6a202ac68 Mon Sep 17 00:00:00 2001 From: Soheil Hassas Yeganeh Date: Tue, 16 May 2017 17:39:02 -0400 Subject: tcp: warn on negative reordering values Commit bafbb9c73241 ("tcp: eliminate negative reordering in tcp_clean_rtx_queue") fixes an issue for negative reordering metrics. To be resilient to such errors, warn and return when a negative metric is passed to tcp_update_reordering(). Signed-off-by: Soheil Hassas Yeganeh Signed-off-by: Neal Cardwell Signed-off-by: Yuchung Cheng Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- net/ipv4/tcp_input.c | 3 +++ 1 file changed, 3 insertions(+) (limited to 'net/ipv4/tcp_input.c') diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index bbadd79815a4..2fa55f57ac06 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -887,6 +887,9 @@ static void tcp_update_reordering(struct sock *sk, const int metric, struct tcp_sock *tp = tcp_sk(sk); int mib_idx; + if (WARN_ON_ONCE(metric < 0)) + return; + if (metric > tp->reordering) { tp->reordering = min(sysctl_tcp_max_reordering, metric); -- cgit v1.2.3 From d0e1a1b5a833b625c93d3d49847609350ebd79db Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Tue, 23 May 2017 15:24:46 -0700 Subject: tcp: better validation of received ack sequences Paul Fiterau Brostean reported : Linux TCP stack we analyze exhibits behavior that seems odd to me. The scenario is as follows (all packets have empty payloads, no window scaling, rcv/snd window size should not be a factor): TEST HARNESS (CLIENT) LINUX SERVER 1. - LISTEN (server listen, then accepts) 2. - --> --> SYN-RECEIVED 3. - <-- <-- SYN-RECEIVED 4. - --> --> ESTABLISHED 5. - <-- <-- FIN WAIT-1 (server opts to close the data connection calling "close" on the connection socket) 6. - --> --> CLOSING (client sends FIN,ACK with not yet sent acknowledgement number) 7. - <-- <-- CLOSING (ACK is 102 instead of 101, why?) ... (silence from CLIENT) 8. - <-- <-- CLOSING (retransmission, again ACK is 102) Now, note that packet 6 while having the expected sequence number, acknowledges something that wasn't sent by the server. So I would expect the packet to maybe prompt an ACK response from the server, and then be ignored. Yet it is not ignored and actually leads to an increase of the acknowledgement number in the server's retransmission of the FIN,ACK packet. The explanation I found is that the FIN in packet 6 was processed, despite the acknowledgement number being unacceptable. Further experiments indeed show that the server processes this FIN, transitioning to CLOSING, then on receiving an ACK for the FIN it had send in packet 5, the server (or better said connection) transitions from CLOSING to TIME_WAIT (as signaled by netstat). Indeed, tcp_rcv_state_process() calls tcp_ack() but does not exploit the @acceptable status but for TCP_SYN_RECV state. What we want here is to send a challenge ACK, if not in TCP_SYN_RECV state. TCP_FIN_WAIT1 state is not the only state we should fix. Add a FLAG_NO_CHALLENGE_ACK so that tcp_rcv_state_process() can choose to send a challenge ACK and discard the packet instead of wrongly change socket state. With help from Neal Cardwell. Signed-off-by: Eric Dumazet Reported-by: Paul Fiterau Brostean Cc: Neal Cardwell Cc: Yuchung Cheng Cc: Soheil Hassas Yeganeh Signed-off-by: David S. Miller --- net/ipv4/tcp_input.c | 24 +++++++++++------------- 1 file changed, 11 insertions(+), 13 deletions(-) (limited to 'net/ipv4/tcp_input.c') diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 2fa55f57ac06..9f4380662196 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -112,6 +112,7 @@ int sysctl_tcp_invalid_ratelimit __read_mostly = HZ/2; #define FLAG_DSACKING_ACK 0x800 /* SACK blocks contained D-SACK info */ #define FLAG_SACK_RENEGING 0x2000 /* snd_una advanced to a sacked seq */ #define FLAG_UPDATE_TS_RECENT 0x4000 /* tcp_replace_ts_recent() */ +#define FLAG_NO_CHALLENGE_ACK 0x8000 /* do not call tcp_send_challenge_ack() */ #define FLAG_ACKED (FLAG_DATA_ACKED|FLAG_SYN_ACKED) #define FLAG_NOT_DUP (FLAG_DATA|FLAG_WIN_UPDATE|FLAG_ACKED) @@ -3568,7 +3569,8 @@ static int tcp_ack(struct sock *sk, const struct sk_buff *skb, int flag) if (before(ack, prior_snd_una)) { /* RFC 5961 5.2 [Blind Data Injection Attack].[Mitigation] */ if (before(ack, prior_snd_una - tp->max_window)) { - tcp_send_challenge_ack(sk, skb); + if (!(flag & FLAG_NO_CHALLENGE_ACK)) + tcp_send_challenge_ack(sk, skb); return -1; } goto old_ack; @@ -5951,13 +5953,17 @@ int tcp_rcv_state_process(struct sock *sk, struct sk_buff *skb) /* step 5: check the ACK field */ acceptable = tcp_ack(sk, skb, FLAG_SLOWPATH | - FLAG_UPDATE_TS_RECENT) > 0; + FLAG_UPDATE_TS_RECENT | + FLAG_NO_CHALLENGE_ACK) > 0; + if (!acceptable) { + if (sk->sk_state == TCP_SYN_RECV) + return 1; /* send one RST */ + tcp_send_challenge_ack(sk, skb); + goto discard; + } switch (sk->sk_state) { case TCP_SYN_RECV: - if (!acceptable) - return 1; - if (!tp->srtt_us) tcp_synack_rtt_meas(sk, req); @@ -6026,14 +6032,6 @@ int tcp_rcv_state_process(struct sock *sk, struct sk_buff *skb) * our SYNACK so stop the SYNACK timer. */ if (req) { - /* Return RST if ack_seq is invalid. - * Note that RFC793 only says to generate a - * DUPACK for it but for TCP Fast Open it seems - * better to treat this case like TCP_SYN_RECV - * above. - */ - if (!acceptable) - return 1; /* We no longer need the request sock. */ reqsk_fastopen_remove(sk, req, false); tcp_rearm_rto(sk); -- cgit v1.2.3 From 775e68a93fe4d33ec93949c8022ed84b97a97096 Mon Sep 17 00:00:00 2001 From: Yuchung Cheng Date: Wed, 31 May 2017 11:30:53 -0700 Subject: tcp: use TS opt on RTTs for congestion control Currently when a data packet is retransmitted, we do not compute an RTT sample for congestion control due to Kern's check. Therefore the congestion control that uses RTT signals may not receive any update during loss recovery which could last many round trips. For example, BBR and Vegas may not be able to update its min RTT estimation if the network path has shortened until it recovers from losses. This patch mitigates that by using TCP timestamp options for RTT measurement for congestion control. Note that we already use timestamps for RTT estimation. Signed-off-by: Yuchung Cheng Signed-off-by: Soheil Hassas Yeganeh Signed-off-by: Neal Cardwell Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- net/ipv4/tcp_input.c | 15 ++++++++------- 1 file changed, 8 insertions(+), 7 deletions(-) (limited to 'net/ipv4/tcp_input.c') diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 9f4380662196..4ea8ec5c7bb4 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -2921,9 +2921,9 @@ static void tcp_update_rtt_min(struct sock *sk, u32 rtt_us) rtt_us ? : jiffies_to_usecs(1)); } -static inline bool tcp_ack_update_rtt(struct sock *sk, const int flag, - long seq_rtt_us, long sack_rtt_us, - long ca_rtt_us) +static bool tcp_ack_update_rtt(struct sock *sk, const int flag, + long seq_rtt_us, long sack_rtt_us, + long ca_rtt_us, struct rate_sample *rs) { const struct tcp_sock *tp = tcp_sk(sk); @@ -2948,6 +2948,7 @@ static inline bool tcp_ack_update_rtt(struct sock *sk, const int flag, seq_rtt_us = ca_rtt_us = delta_us; } + rs->rtt_us = ca_rtt_us; /* RTT of last (S)ACKed packet (or -1) */ if (seq_rtt_us < 0) return false; @@ -2967,12 +2968,13 @@ static inline bool tcp_ack_update_rtt(struct sock *sk, const int flag, /* Compute time elapsed between (last) SYNACK and the ACK completing 3WHS. */ void tcp_synack_rtt_meas(struct sock *sk, struct request_sock *req) { + struct rate_sample rs; long rtt_us = -1L; if (req && !req->num_retrans && tcp_rsk(req)->snt_synack) rtt_us = tcp_stamp_us_delta(tcp_clock_us(), tcp_rsk(req)->snt_synack); - tcp_ack_update_rtt(sk, FLAG_SYN_ACKED, rtt_us, -1L, rtt_us); + tcp_ack_update_rtt(sk, FLAG_SYN_ACKED, rtt_us, -1L, rtt_us, &rs); } @@ -3177,9 +3179,8 @@ static int tcp_clean_rtx_queue(struct sock *sk, int prior_fackets, sack_rtt_us = tcp_stamp_us_delta(tp->tcp_mstamp, sack->first_sackt); ca_rtt_us = tcp_stamp_us_delta(tp->tcp_mstamp, sack->last_sackt); } - sack->rate->rtt_us = ca_rtt_us; /* RTT of last (S)ACKed packet, or -1 */ rtt_update = tcp_ack_update_rtt(sk, flag, seq_rtt_us, sack_rtt_us, - ca_rtt_us); + ca_rtt_us, sack->rate); if (flag & FLAG_ACKED) { tcp_rearm_rto(sk); @@ -3215,7 +3216,7 @@ static int tcp_clean_rtx_queue(struct sock *sk, int prior_fackets, if (icsk->icsk_ca_ops->pkts_acked) { struct ack_sample sample = { .pkts_acked = pkts_acked, - .rtt_us = ca_rtt_us, + .rtt_us = sack->rate->rtt_us, .in_flight = last_in_flight }; icsk->icsk_ca_ops->pkts_acked(sk, &sample); -- cgit v1.2.3 From eed29f17f09ad7f400bc245f209acad6a8214fac Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Wed, 7 Jun 2017 10:34:36 -0700 Subject: tcp: add a struct net parameter to tcp_parse_options() We want to move some TCP sysctls to net namespaces in the future. tcp_window_scaling, tcp_sack and tcp_timestamps being fetched from tcp_parse_options(), we need to pass an extra parameter. Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- net/ipv4/tcp_input.c | 18 +++++++++++------- 1 file changed, 11 insertions(+), 7 deletions(-) (limited to 'net/ipv4/tcp_input.c') diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 4ea8ec5c7bb4..99ee707f0ef4 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -3724,7 +3724,8 @@ static void tcp_parse_fastopen_option(int len, const unsigned char *cookie, * But, this can also be called on packets in the established flow when * the fast version below fails. */ -void tcp_parse_options(const struct sk_buff *skb, +void tcp_parse_options(const struct net *net, + const struct sk_buff *skb, struct tcp_options_received *opt_rx, int estab, struct tcp_fastopen_cookie *foc) { @@ -3858,7 +3859,8 @@ static bool tcp_parse_aligned_timestamp(struct tcp_sock *tp, const struct tcphdr /* Fast parse options. This hopes to only see timestamps. * If it is wrong it falls back on tcp_parse_options(). */ -static bool tcp_fast_parse_options(const struct sk_buff *skb, +static bool tcp_fast_parse_options(const struct net *net, + const struct sk_buff *skb, const struct tcphdr *th, struct tcp_sock *tp) { /* In the spirit of fast parsing, compare doff directly to constant @@ -3873,7 +3875,7 @@ static bool tcp_fast_parse_options(const struct sk_buff *skb, return true; } - tcp_parse_options(skb, &tp->rx_opt, 1, NULL); + tcp_parse_options(net, skb, &tp->rx_opt, 1, NULL); if (tp->rx_opt.saw_tstamp && tp->rx_opt.rcv_tsecr) tp->rx_opt.rcv_tsecr -= tp->tsoffset; @@ -5234,7 +5236,8 @@ static bool tcp_validate_incoming(struct sock *sk, struct sk_buff *skb, bool rst_seq_match = false; /* RFC1323: H1. Apply PAWS check first. */ - if (tcp_fast_parse_options(skb, th, tp) && tp->rx_opt.saw_tstamp && + if (tcp_fast_parse_options(sock_net(sk), skb, th, tp) && + tp->rx_opt.saw_tstamp && tcp_paws_discard(sk, skb)) { if (!th->rst) { NET_INC_STATS(sock_net(sk), LINUX_MIB_PAWSESTABREJECTED); @@ -5605,7 +5608,7 @@ static bool tcp_rcv_fastopen_synack(struct sock *sk, struct sk_buff *synack, /* Get original SYNACK MSS value if user MSS sets mss_clamp */ tcp_clear_options(&opt); opt.user_mss = opt.mss_clamp = 0; - tcp_parse_options(synack, &opt, 0, NULL); + tcp_parse_options(sock_net(sk), synack, &opt, 0, NULL); mss = opt.mss_clamp; } @@ -5659,7 +5662,7 @@ static int tcp_rcv_synsent_state_process(struct sock *sk, struct sk_buff *skb, int saved_clamp = tp->rx_opt.mss_clamp; bool fastopen_fail; - tcp_parse_options(skb, &tp->rx_opt, 0, &foc); + tcp_parse_options(sock_net(sk), skb, &tp->rx_opt, 0, &foc); if (tp->rx_opt.saw_tstamp && tp->rx_opt.rcv_tsecr) tp->rx_opt.rcv_tsecr -= tp->tsoffset; @@ -6332,7 +6335,8 @@ int tcp_conn_request(struct request_sock_ops *rsk_ops, tcp_clear_options(&tmp_opt); tmp_opt.mss_clamp = af_ops->mss_clamp; tmp_opt.user_mss = tp->rx_opt.user_mss; - tcp_parse_options(skb, &tmp_opt, 0, want_cookie ? NULL : &foc); + tcp_parse_options(sock_net(sk), skb, &tmp_opt, 0, + want_cookie ? NULL : &foc); if (want_cookie && !tmp_opt.saw_tstamp) tcp_clear_options(&tmp_opt); -- cgit v1.2.3 From f930103421f6579719b8252285c94c1195f6e032 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Wed, 7 Jun 2017 10:34:37 -0700 Subject: tcp: Namespaceify sysctl_tcp_sack Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- net/ipv4/tcp_input.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'net/ipv4/tcp_input.c') diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 99ee707f0ef4..2eacfcaf1257 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -78,7 +78,6 @@ int sysctl_tcp_timestamps __read_mostly = 1; int sysctl_tcp_window_scaling __read_mostly = 1; -int sysctl_tcp_sack __read_mostly = 1; int sysctl_tcp_fack __read_mostly; int sysctl_tcp_max_reordering __read_mostly = 300; int sysctl_tcp_dsack __read_mostly = 1; @@ -3790,7 +3789,7 @@ void tcp_parse_options(const struct net *net, break; case TCPOPT_SACK_PERM: if (opsize == TCPOLEN_SACK_PERM && th->syn && - !estab && sysctl_tcp_sack) { + !estab && net->ipv4.sysctl_tcp_sack) { opt_rx->sack_ok = TCP_SACK_SEEN; tcp_sack_reset(opt_rx); } -- cgit v1.2.3 From 9bb37ef00e932eb4b989e855245468feb3980700 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Wed, 7 Jun 2017 10:34:38 -0700 Subject: tcp: Namespaceify sysctl_tcp_window_scaling Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- net/ipv4/tcp_input.c | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) (limited to 'net/ipv4/tcp_input.c') diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 2eacfcaf1257..675ee903370f 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -77,7 +77,6 @@ #include int sysctl_tcp_timestamps __read_mostly = 1; -int sysctl_tcp_window_scaling __read_mostly = 1; int sysctl_tcp_fack __read_mostly; int sysctl_tcp_max_reordering __read_mostly = 300; int sysctl_tcp_dsack __read_mostly = 1; @@ -3765,7 +3764,7 @@ void tcp_parse_options(const struct net *net, break; case TCPOPT_WINDOW: if (opsize == TCPOLEN_WINDOW && th->syn && - !estab && sysctl_tcp_window_scaling) { + !estab && net->ipv4.sysctl_tcp_window_scaling) { __u8 snd_wscale = *(__u8 *)ptr; opt_rx->wscale_ok = 1; if (snd_wscale > TCP_MAX_WSCALE) { -- cgit v1.2.3 From 5d2ed0521ac98f1ae0243fe52b8ebf95e2abf791 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Wed, 7 Jun 2017 10:34:39 -0700 Subject: tcp: Namespaceify sysctl_tcp_timestamps Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- net/ipv4/tcp_input.c | 5 ++--- 1 file changed, 2 insertions(+), 3 deletions(-) (limited to 'net/ipv4/tcp_input.c') diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 675ee903370f..2ab7e2fa9bb9 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -76,7 +76,6 @@ #include #include -int sysctl_tcp_timestamps __read_mostly = 1; int sysctl_tcp_fack __read_mostly; int sysctl_tcp_max_reordering __read_mostly = 300; int sysctl_tcp_dsack __read_mostly = 1; @@ -3780,7 +3779,7 @@ void tcp_parse_options(const struct net *net, case TCPOPT_TIMESTAMP: if ((opsize == TCPOLEN_TIMESTAMP) && ((estab && opt_rx->tstamp_ok) || - (!estab && sysctl_tcp_timestamps))) { + (!estab && net->ipv4.sysctl_tcp_timestamps))) { opt_rx->saw_tstamp = 1; opt_rx->rcv_tsval = get_unaligned_be32(ptr); opt_rx->rcv_tsecr = get_unaligned_be32(ptr + 4); @@ -6352,7 +6351,7 @@ int tcp_conn_request(struct request_sock_ops *rsk_ops, goto drop_and_free; if (tmp_opt.tstamp_ok) - tcp_rsk(req)->ts_off = af_ops->init_ts_off(skb); + tcp_rsk(req)->ts_off = af_ops->init_ts_off(net, skb); if (!want_cookie && !isn) { /* Kill the following clause, if you dislike this way. */ -- cgit v1.2.3 From 8550f328f45db6d37981eb2041bc465810245c03 Mon Sep 17 00:00:00 2001 From: Lawrence Brakmo Date: Fri, 30 Jun 2017 20:02:42 -0700 Subject: bpf: Support for per connection SYN/SYN-ACK RTOs This patch adds support for setting a per connection SYN and SYN_ACK RTOs from within a BPF_SOCK_OPS program. For example, to set small RTOs when it is known both hosts are within a datacenter. Signed-off-by: Lawrence Brakmo Signed-off-by: David S. Miller --- net/ipv4/tcp_input.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'net/ipv4/tcp_input.c') diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 2ab7e2fa9bb9..bcc96654cd7e 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -6406,7 +6406,8 @@ int tcp_conn_request(struct request_sock_ops *rsk_ops, } else { tcp_rsk(req)->tfo_listener = false; if (!want_cookie) - inet_csk_reqsk_queue_hash_add(sk, req, TCP_TIMEOUT_INIT); + inet_csk_reqsk_queue_hash_add(sk, req, + tcp_timeout_init((struct sock *)req)); af_ops->send_synack(sk, dst, &fl, req, &foc, !want_cookie ? TCP_SYNACK_NORMAL : TCP_SYNACK_COOKIE); -- cgit v1.2.3 From 9872a4bde31b0b055448e9ac1f4c9ee62d978766 Mon Sep 17 00:00:00 2001 From: Lawrence Brakmo Date: Fri, 30 Jun 2017 20:02:47 -0700 Subject: bpf: Add TCP connection BPF callbacks Added callbacks to BPF SOCK_OPS type program before an active connection is intialized and after a passive or active connection is established. The following patch demostrates how they can be used to set send and receive buffer sizes. Signed-off-by: Lawrence Brakmo Signed-off-by: David S. Miller --- net/ipv4/tcp_input.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'net/ipv4/tcp_input.c') diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index bcc96654cd7e..664210e5e4a7 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -5571,7 +5571,7 @@ void tcp_finish_connect(struct sock *sk, struct sk_buff *skb) icsk->icsk_af_ops->rebuild_header(sk); tcp_init_metrics(sk); - + tcp_call_bpf(sk, BPF_SOCK_OPS_ACTIVE_ESTABLISHED_CB); tcp_init_congestion_control(sk); /* Prevent spurious tcp_cwnd_restart() on first data @@ -5977,6 +5977,7 @@ int tcp_rcv_state_process(struct sock *sk, struct sk_buff *skb) } else { /* Make sure socket is routed, for correct metrics. */ icsk->icsk_af_ops->rebuild_header(sk); + tcp_call_bpf(sk, BPF_SOCK_OPS_PASSIVE_ESTABLISHED_CB); tcp_init_congestion_control(sk); tcp_mtup_init(sk); -- cgit v1.2.3 From 91b5b21c7c16899abb37f4a9e4388b4e9aae0b9d Mon Sep 17 00:00:00 2001 From: Lawrence Brakmo Date: Fri, 30 Jun 2017 20:02:49 -0700 Subject: bpf: Add support for changing congestion control Added support for changing congestion control for SOCK_OPS bpf programs through the setsockopt bpf helper function. It also adds a new SOCK_OPS op, BPF_SOCK_OPS_NEEDS_ECN, that is needed for congestion controls, like dctcp, that need to enable ECN in the SYN packets. Signed-off-by: Lawrence Brakmo Signed-off-by: David S. Miller --- net/ipv4/tcp_input.c | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) (limited to 'net/ipv4/tcp_input.c') diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 664210e5e4a7..2920e0cb09f8 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -6191,7 +6191,8 @@ static void tcp_ecn_create_request(struct request_sock *req, ecn_ok = net->ipv4.sysctl_tcp_ecn || ecn_ok_dst; if ((!ect && ecn_ok) || tcp_ca_needs_ecn(listen_sk) || - (ecn_ok_dst & DST_FEATURE_ECN_CA)) + (ecn_ok_dst & DST_FEATURE_ECN_CA) || + tcp_bpf_ca_needs_ecn((struct sock *)req)) inet_rsk(req)->ecn_ok = 1; } -- cgit v1.2.3