From 2ae21cf527da0e5cf9d7ee14bd5b0909bb9d1a75 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Thu, 26 Oct 2017 21:54:56 -0700 Subject: [PATCH 01/15] tcp: Namespace-ify sysctl_tcp_early_retrans Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- include/net/netns/ipv4.h | 1 + include/net/tcp.h | 1 - net/ipv4/sysctl_net_ipv4.c | 18 +++++++++--------- net/ipv4/tcp_input.c | 1 - net/ipv4/tcp_ipv4.c | 1 + net/ipv4/tcp_output.c | 4 +++- 6 files changed, 14 insertions(+), 12 deletions(-) diff --git a/include/net/netns/ipv4.h b/include/net/netns/ipv4.h index 2c4222a5d102..a7f39e3ea666 100644 --- a/include/net/netns/ipv4.h +++ b/include/net/netns/ipv4.h @@ -128,6 +128,7 @@ struct netns_ipv4 { int sysctl_tcp_sack; int sysctl_tcp_window_scaling; int sysctl_tcp_timestamps; + int sysctl_tcp_early_retrans; struct inet_timewait_death_row tcp_death_row; int sysctl_max_syn_backlog; int sysctl_tcp_fastopen; diff --git a/include/net/tcp.h b/include/net/tcp.h index 285bc82dea41..a12b71d4118b 100644 --- a/include/net/tcp.h +++ b/include/net/tcp.h @@ -265,7 +265,6 @@ extern int sysctl_tcp_workaround_signed_windows; extern int sysctl_tcp_slow_start_after_idle; extern int sysctl_tcp_thin_linear_timeouts; extern int sysctl_tcp_thin_dupack; -extern int sysctl_tcp_early_retrans; extern int sysctl_tcp_recovery; #define TCP_RACK_LOSS_DETECTION 0x1 /* Use RACK to detect losses */ diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c index 81d218346cf7..f0f650f020af 100644 --- a/net/ipv4/sysctl_net_ipv4.c +++ b/net/ipv4/sysctl_net_ipv4.c @@ -634,15 +634,6 @@ static struct ctl_table ipv4_table[] = { .mode = 0644, .proc_handler = proc_dointvec }, - { - .procname = "tcp_early_retrans", - .data = &sysctl_tcp_early_retrans, - .maxlen = sizeof(int), - .mode = 0644, - .proc_handler = proc_dointvec_minmax, - .extra1 = &zero, - .extra2 = &four, - }, { .procname = "tcp_min_tso_segs", .data = &sysctl_tcp_min_tso_segs, @@ -1145,6 +1136,15 @@ static struct ctl_table ipv4_net_table[] = { .mode = 0644, .proc_handler = proc_dointvec }, + { + .procname = "tcp_early_retrans", + .data = &init_net.ipv4.sysctl_tcp_early_retrans, + .maxlen = sizeof(int), + .mode = 0644, + .proc_handler = proc_dointvec_minmax, + .extra1 = &zero, + .extra2 = &four, + }, { } }; diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 337f6011528a..7656b1e6d504 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -95,7 +95,6 @@ int sysctl_tcp_max_orphans __read_mostly = NR_FILE; int sysctl_tcp_frto __read_mostly = 2; int sysctl_tcp_min_rtt_wlen __read_mostly = 300; int sysctl_tcp_moderate_rcvbuf __read_mostly = 1; -int sysctl_tcp_early_retrans __read_mostly = 3; int sysctl_tcp_invalid_ratelimit __read_mostly = HZ/2; #define FLAG_DATA 0x01 /* Incoming frame contained data. */ diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c index 23a8100af5ad..7ab313f6768e 100644 --- a/net/ipv4/tcp_ipv4.c +++ b/net/ipv4/tcp_ipv4.c @@ -2484,6 +2484,7 @@ static int __net_init tcp_sk_init(struct net *net) net->ipv4.sysctl_tcp_sack = 1; net->ipv4.sysctl_tcp_window_scaling = 1; net->ipv4.sysctl_tcp_timestamps = 1; + net->ipv4.sysctl_tcp_early_retrans = 3; net->ipv4.sysctl_tcp_fastopen = TFO_CLIENT_ENABLE; spin_lock_init(&net->ipv4.tcp_fastopen_ctx_lock); diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c index c8fc512e0bbb..21713836d46a 100644 --- a/net/ipv4/tcp_output.c +++ b/net/ipv4/tcp_output.c @@ -2435,6 +2435,7 @@ bool tcp_schedule_loss_probe(struct sock *sk) struct inet_connection_sock *icsk = inet_csk(sk); struct tcp_sock *tp = tcp_sk(sk); u32 timeout, rto_delta_us; + int early_retrans; /* Don't do any loss probe on a Fast Open connection before 3WHS * finishes. @@ -2442,10 +2443,11 @@ bool tcp_schedule_loss_probe(struct sock *sk) if (tp->fastopen_rsk) return false; + early_retrans = sock_net(sk)->ipv4.sysctl_tcp_early_retrans; /* Schedule a loss probe in 2*RTT for SACK capable connections * in Open state, that are either limited by cwnd or application. */ - if ((sysctl_tcp_early_retrans != 3 && sysctl_tcp_early_retrans != 4) || + if ((early_retrans != 3 && early_retrans != 4) || !tp->packets_out || !tcp_is_sack(tp) || icsk->icsk_ca_state != TCP_CA_Open) return false; From e20223f1962831d1b1c416d59d259879d0639d68 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Thu, 26 Oct 2017 21:54:57 -0700 Subject: [PATCH 02/15] tcp: Namespace-ify sysctl_tcp_recovery Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- include/net/netns/ipv4.h | 1 + include/net/tcp.h | 2 +- net/ipv4/sysctl_net_ipv4.c | 14 +++++++------- net/ipv4/tcp_input.c | 2 +- net/ipv4/tcp_ipv4.c | 1 + net/ipv4/tcp_recovery.c | 2 -- 6 files changed, 11 insertions(+), 11 deletions(-) diff --git a/include/net/netns/ipv4.h b/include/net/netns/ipv4.h index a7f39e3ea666..d6ed718075d4 100644 --- a/include/net/netns/ipv4.h +++ b/include/net/netns/ipv4.h @@ -129,6 +129,7 @@ struct netns_ipv4 { int sysctl_tcp_window_scaling; int sysctl_tcp_timestamps; int sysctl_tcp_early_retrans; + int sysctl_tcp_recovery; struct inet_timewait_death_row tcp_death_row; int sysctl_max_syn_backlog; int sysctl_tcp_fastopen; diff --git a/include/net/tcp.h b/include/net/tcp.h index a12b71d4118b..c7f51534fc44 100644 --- a/include/net/tcp.h +++ b/include/net/tcp.h @@ -265,7 +265,7 @@ extern int sysctl_tcp_workaround_signed_windows; extern int sysctl_tcp_slow_start_after_idle; extern int sysctl_tcp_thin_linear_timeouts; extern int sysctl_tcp_thin_dupack; -extern int sysctl_tcp_recovery; + #define TCP_RACK_LOSS_DETECTION 0x1 /* Use RACK to detect losses */ extern int sysctl_tcp_limit_output_bytes; diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c index f0f650f020af..78019adcae87 100644 --- a/net/ipv4/sysctl_net_ipv4.c +++ b/net/ipv4/sysctl_net_ipv4.c @@ -449,13 +449,6 @@ static struct ctl_table ipv4_table[] = { .mode = 0644, .proc_handler = proc_dointvec }, - { - .procname = "tcp_recovery", - .data = &sysctl_tcp_recovery, - .maxlen = sizeof(int), - .mode = 0644, - .proc_handler = proc_dointvec, - }, { .procname = "tcp_max_reordering", .data = &sysctl_tcp_max_reordering, @@ -1145,6 +1138,13 @@ static struct ctl_table ipv4_net_table[] = { .extra1 = &zero, .extra2 = &four, }, + { + .procname = "tcp_recovery", + .data = &init_net.ipv4.sysctl_tcp_recovery, + .maxlen = sizeof(int), + .mode = 0644, + .proc_handler = proc_dointvec, + }, { } }; diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 7656b1e6d504..5b2272dbf6a9 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -2788,7 +2788,7 @@ static void tcp_rack_identify_loss(struct sock *sk, int *ack_flag) struct tcp_sock *tp = tcp_sk(sk); /* Use RACK to detect loss */ - if (sysctl_tcp_recovery & TCP_RACK_LOSS_DETECTION) { + if (sock_net(sk)->ipv4.sysctl_tcp_recovery & TCP_RACK_LOSS_DETECTION) { u32 prior_retrans = tp->retrans_out; tcp_rack_mark_lost(sk); diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c index 7ab313f6768e..517ff1948a71 100644 --- a/net/ipv4/tcp_ipv4.c +++ b/net/ipv4/tcp_ipv4.c @@ -2485,6 +2485,7 @@ static int __net_init tcp_sk_init(struct net *net) net->ipv4.sysctl_tcp_window_scaling = 1; net->ipv4.sysctl_tcp_timestamps = 1; net->ipv4.sysctl_tcp_early_retrans = 3; + net->ipv4.sysctl_tcp_recovery = TCP_RACK_LOSS_DETECTION; net->ipv4.sysctl_tcp_fastopen = TFO_CLIENT_ENABLE; spin_lock_init(&net->ipv4.tcp_fastopen_ctx_lock); diff --git a/net/ipv4/tcp_recovery.c b/net/ipv4/tcp_recovery.c index cda6074a429a..d3603a9e24ea 100644 --- a/net/ipv4/tcp_recovery.c +++ b/net/ipv4/tcp_recovery.c @@ -1,8 +1,6 @@ #include #include -int sysctl_tcp_recovery __read_mostly = TCP_RACK_LOSS_DETECTION; - static void tcp_rack_mark_skb_lost(struct sock *sk, struct sk_buff *skb) { struct tcp_sock *tp = tcp_sk(sk); From 2c04ac8ae0b61e0780a30b7069a11bb202b21f26 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Thu, 26 Oct 2017 21:54:58 -0700 Subject: [PATCH 03/15] tcp: Namespace-ify sysctl_tcp_thin_linear_timeouts Note that sysctl_tcp_thin_dupack was not used, I deleted it. Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- include/net/netns/ipv4.h | 1 + include/net/tcp.h | 2 -- net/ipv4/sysctl_net_ipv4.c | 14 +++++++------- net/ipv4/tcp_timer.c | 4 +--- 4 files changed, 9 insertions(+), 12 deletions(-) diff --git a/include/net/netns/ipv4.h b/include/net/netns/ipv4.h index d6ed718075d4..2a9f37b39c45 100644 --- a/include/net/netns/ipv4.h +++ b/include/net/netns/ipv4.h @@ -130,6 +130,7 @@ struct netns_ipv4 { int sysctl_tcp_timestamps; int sysctl_tcp_early_retrans; int sysctl_tcp_recovery; + int sysctl_tcp_thin_linear_timeouts; struct inet_timewait_death_row tcp_death_row; int sysctl_max_syn_backlog; int sysctl_tcp_fastopen; diff --git a/include/net/tcp.h b/include/net/tcp.h index c7f51534fc44..063a7a48b7fe 100644 --- a/include/net/tcp.h +++ b/include/net/tcp.h @@ -263,8 +263,6 @@ extern int sysctl_tcp_moderate_rcvbuf; extern int sysctl_tcp_tso_win_divisor; extern int sysctl_tcp_workaround_signed_windows; extern int sysctl_tcp_slow_start_after_idle; -extern int sysctl_tcp_thin_linear_timeouts; -extern int sysctl_tcp_thin_dupack; #define TCP_RACK_LOSS_DETECTION 0x1 /* Use RACK to detect losses */ diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c index 78019adcae87..12003214f4d8 100644 --- a/net/ipv4/sysctl_net_ipv4.c +++ b/net/ipv4/sysctl_net_ipv4.c @@ -620,13 +620,6 @@ static struct ctl_table ipv4_table[] = { .mode = 0644, .proc_handler = proc_allowed_congestion_control, }, - { - .procname = "tcp_thin_linear_timeouts", - .data = &sysctl_tcp_thin_linear_timeouts, - .maxlen = sizeof(int), - .mode = 0644, - .proc_handler = proc_dointvec - }, { .procname = "tcp_min_tso_segs", .data = &sysctl_tcp_min_tso_segs, @@ -1145,6 +1138,13 @@ static struct ctl_table ipv4_net_table[] = { .mode = 0644, .proc_handler = proc_dointvec, }, + { + .procname = "tcp_thin_linear_timeouts", + .data = &init_net.ipv4.sysctl_tcp_thin_linear_timeouts, + .maxlen = sizeof(int), + .mode = 0644, + .proc_handler = proc_dointvec + }, { } }; diff --git a/net/ipv4/tcp_timer.c b/net/ipv4/tcp_timer.c index 804a8d34ce86..035a1ef1f2d8 100644 --- a/net/ipv4/tcp_timer.c +++ b/net/ipv4/tcp_timer.c @@ -22,8 +22,6 @@ #include #include -int sysctl_tcp_thin_linear_timeouts __read_mostly; - /** * tcp_write_err() - close socket and save error info * @sk: The socket the error has appeared on. @@ -522,7 +520,7 @@ void tcp_retransmit_timer(struct sock *sk) * linear-timeout retransmissions into a black hole */ if (sk->sk_state == TCP_ESTABLISHED && - (tp->thin_lto || sysctl_tcp_thin_linear_timeouts) && + (tp->thin_lto || net->ipv4.sysctl_tcp_thin_linear_timeouts) && tcp_stream_is_thin(tp) && icsk->icsk_retransmits <= TCP_THIN_LINEAR_RETRIES) { icsk->icsk_backoff = 0; From b510f0d23a47c3d1f074fe583e7867dc4918fe02 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Thu, 26 Oct 2017 21:54:59 -0700 Subject: [PATCH 04/15] tcp: Namespace-ify sysctl_tcp_slow_start_after_idle Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- include/net/netns/ipv4.h | 1 + include/net/tcp.h | 3 +-- net/ipv4/sysctl_net_ipv4.c | 14 +++++++------- net/ipv4/tcp_ipv4.c | 1 + net/ipv4/tcp_output.c | 5 +---- 5 files changed, 11 insertions(+), 13 deletions(-) diff --git a/include/net/netns/ipv4.h b/include/net/netns/ipv4.h index 2a9f37b39c45..8662692686b3 100644 --- a/include/net/netns/ipv4.h +++ b/include/net/netns/ipv4.h @@ -131,6 +131,7 @@ struct netns_ipv4 { int sysctl_tcp_early_retrans; int sysctl_tcp_recovery; int sysctl_tcp_thin_linear_timeouts; + int sysctl_tcp_slow_start_after_idle; struct inet_timewait_death_row tcp_death_row; int sysctl_max_syn_backlog; int sysctl_tcp_fastopen; diff --git a/include/net/tcp.h b/include/net/tcp.h index 063a7a48b7fe..cc2ab522eb5c 100644 --- a/include/net/tcp.h +++ b/include/net/tcp.h @@ -262,7 +262,6 @@ extern int sysctl_tcp_nometrics_save; extern int sysctl_tcp_moderate_rcvbuf; extern int sysctl_tcp_tso_win_divisor; extern int sysctl_tcp_workaround_signed_windows; -extern int sysctl_tcp_slow_start_after_idle; #define TCP_RACK_LOSS_DETECTION 0x1 /* Use RACK to detect losses */ @@ -1308,7 +1307,7 @@ static inline void tcp_slow_start_after_idle_check(struct sock *sk) struct tcp_sock *tp = tcp_sk(sk); s32 delta; - if (!sysctl_tcp_slow_start_after_idle || tp->packets_out || + if (!sock_net(sk)->ipv4.sysctl_tcp_slow_start_after_idle || tp->packets_out || ca_ops->cong_control) return; delta = tcp_jiffies32 - tp->lsndtime; diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c index 12003214f4d8..40d69af8b363 100644 --- a/net/ipv4/sysctl_net_ipv4.c +++ b/net/ipv4/sysctl_net_ipv4.c @@ -571,13 +571,6 @@ static struct ctl_table ipv4_table[] = { .mode = 0644, .proc_handler = proc_dointvec }, - { - .procname = "tcp_slow_start_after_idle", - .data = &sysctl_tcp_slow_start_after_idle, - .maxlen = sizeof(int), - .mode = 0644, - .proc_handler = proc_dointvec - }, #ifdef CONFIG_NETLABEL { .procname = "cipso_cache_enable", @@ -1145,6 +1138,13 @@ static struct ctl_table ipv4_net_table[] = { .mode = 0644, .proc_handler = proc_dointvec }, + { + .procname = "tcp_slow_start_after_idle", + .data = &init_net.ipv4.sysctl_tcp_slow_start_after_idle, + .maxlen = sizeof(int), + .mode = 0644, + .proc_handler = proc_dointvec + }, { } }; diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c index 517ff1948a71..cea63a4b5965 100644 --- a/net/ipv4/tcp_ipv4.c +++ b/net/ipv4/tcp_ipv4.c @@ -2486,6 +2486,7 @@ static int __net_init tcp_sk_init(struct net *net) net->ipv4.sysctl_tcp_timestamps = 1; net->ipv4.sysctl_tcp_early_retrans = 3; net->ipv4.sysctl_tcp_recovery = TCP_RACK_LOSS_DETECTION; + net->ipv4.sysctl_tcp_slow_start_after_idle = 1; /* By default, RFC2861 behavior. */ net->ipv4.sysctl_tcp_fastopen = TFO_CLIENT_ENABLE; spin_lock_init(&net->ipv4.tcp_fastopen_ctx_lock); diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c index 21713836d46a..bdc288a06f94 100644 --- a/net/ipv4/tcp_output.c +++ b/net/ipv4/tcp_output.c @@ -62,9 +62,6 @@ int sysctl_tcp_limit_output_bytes __read_mostly = 262144; */ int sysctl_tcp_tso_win_divisor __read_mostly = 3; -/* By default, RFC2861 behavior. */ -int sysctl_tcp_slow_start_after_idle __read_mostly = 1; - static bool tcp_write_xmit(struct sock *sk, unsigned int mss_now, int nonagle, int push_one, gfp_t gfp); @@ -1690,7 +1687,7 @@ static void tcp_cwnd_validate(struct sock *sk, bool is_cwnd_limited) if (tp->packets_out > tp->snd_cwnd_used) tp->snd_cwnd_used = tp->packets_out; - if (sysctl_tcp_slow_start_after_idle && + if (sock_net(sk)->ipv4.sysctl_tcp_slow_start_after_idle && (s32)(tcp_jiffies32 - tp->snd_cwnd_stamp) >= inet_csk(sk)->icsk_rto && !ca_ops->cong_control) tcp_cwnd_application_limited(sk); From e0a1e5b519236dc1662ff25e42560dd1be9e3776 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Thu, 26 Oct 2017 21:55:00 -0700 Subject: [PATCH 05/15] tcp: Namespace-ify sysctl_tcp_retrans_collapse Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- include/net/netns/ipv4.h | 1 + include/net/tcp.h | 1 - net/ipv4/sysctl_net_ipv4.c | 14 +++++++------- net/ipv4/tcp_ipv4.c | 2 +- net/ipv4/tcp_output.c | 5 +---- 5 files changed, 10 insertions(+), 13 deletions(-) diff --git a/include/net/netns/ipv4.h b/include/net/netns/ipv4.h index 8662692686b3..b28c172b10e4 100644 --- a/include/net/netns/ipv4.h +++ b/include/net/netns/ipv4.h @@ -132,6 +132,7 @@ struct netns_ipv4 { int sysctl_tcp_recovery; int sysctl_tcp_thin_linear_timeouts; int sysctl_tcp_slow_start_after_idle; + int sysctl_tcp_retrans_collapse; struct inet_timewait_death_row tcp_death_row; int sysctl_max_syn_backlog; int sysctl_tcp_fastopen; diff --git a/include/net/tcp.h b/include/net/tcp.h index cc2ab522eb5c..33cc86355b8f 100644 --- a/include/net/tcp.h +++ b/include/net/tcp.h @@ -243,7 +243,6 @@ void tcp_time_wait(struct sock *sk, int state, int timeo); /* sysctl variables for tcp */ -extern int sysctl_tcp_retrans_collapse; extern int sysctl_tcp_stdurg; extern int sysctl_tcp_rfc1337; extern int sysctl_tcp_abort_on_overflow; diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c index 40d69af8b363..533b92ad39dd 100644 --- a/net/ipv4/sysctl_net_ipv4.c +++ b/net/ipv4/sysctl_net_ipv4.c @@ -386,13 +386,6 @@ static int proc_tcp_available_ulp(struct ctl_table *ctl, } static struct ctl_table ipv4_table[] = { - { - .procname = "tcp_retrans_collapse", - .data = &sysctl_tcp_retrans_collapse, - .maxlen = sizeof(int), - .mode = 0644, - .proc_handler = proc_dointvec - }, { .procname = "tcp_max_orphans", .data = &sysctl_tcp_max_orphans, @@ -1145,6 +1138,13 @@ static struct ctl_table ipv4_net_table[] = { .mode = 0644, .proc_handler = proc_dointvec }, + { + .procname = "tcp_retrans_collapse", + .data = &init_net.ipv4.sysctl_tcp_retrans_collapse, + .maxlen = sizeof(int), + .mode = 0644, + .proc_handler = proc_dointvec + }, { } }; diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c index cea63a4b5965..2bc6ba2059d3 100644 --- a/net/ipv4/tcp_ipv4.c +++ b/net/ipv4/tcp_ipv4.c @@ -2487,7 +2487,7 @@ static int __net_init tcp_sk_init(struct net *net) net->ipv4.sysctl_tcp_early_retrans = 3; net->ipv4.sysctl_tcp_recovery = TCP_RACK_LOSS_DETECTION; net->ipv4.sysctl_tcp_slow_start_after_idle = 1; /* By default, RFC2861 behavior. */ - + net->ipv4.sysctl_tcp_retrans_collapse = 1; net->ipv4.sysctl_tcp_fastopen = TFO_CLIENT_ENABLE; spin_lock_init(&net->ipv4.tcp_fastopen_ctx_lock); net->ipv4.sysctl_tcp_fastopen_blackhole_timeout = 60 * 60; diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c index bdc288a06f94..55a0aa4b96df 100644 --- a/net/ipv4/tcp_output.c +++ b/net/ipv4/tcp_output.c @@ -45,9 +45,6 @@ #include -/* People can turn this off for buggy TCP's found in printers etc. */ -int sysctl_tcp_retrans_collapse __read_mostly = 1; - /* People can turn this on to work with those rare, broken TCPs that * interpret the window field as a signed quantity. */ @@ -2804,7 +2801,7 @@ static void tcp_retrans_try_collapse(struct sock *sk, struct sk_buff *to, struct sk_buff *skb = to, *tmp; bool first = true; - if (!sysctl_tcp_retrans_collapse) + if (!sock_net(sk)->ipv4.sysctl_tcp_retrans_collapse) return; if (TCP_SKB_CB(skb)->tcp_flags & TCPHDR_SYN) return; From 3f4c7c6f6a9053493ce7dd8a0f17ed8eafc53893 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Thu, 26 Oct 2017 21:55:01 -0700 Subject: [PATCH 06/15] tcp: Namespace-ify sysctl_tcp_stdurg Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- include/net/netns/ipv4.h | 1 + include/net/tcp.h | 1 - net/ipv4/sysctl_net_ipv4.c | 14 +++++++------- net/ipv4/tcp_input.c | 3 +-- 4 files changed, 9 insertions(+), 10 deletions(-) diff --git a/include/net/netns/ipv4.h b/include/net/netns/ipv4.h index b28c172b10e4..ffa2cf3dc747 100644 --- a/include/net/netns/ipv4.h +++ b/include/net/netns/ipv4.h @@ -133,6 +133,7 @@ struct netns_ipv4 { int sysctl_tcp_thin_linear_timeouts; int sysctl_tcp_slow_start_after_idle; int sysctl_tcp_retrans_collapse; + int sysctl_tcp_stdurg; struct inet_timewait_death_row tcp_death_row; int sysctl_max_syn_backlog; int sysctl_tcp_fastopen; diff --git a/include/net/tcp.h b/include/net/tcp.h index 33cc86355b8f..cf3fac7008d7 100644 --- a/include/net/tcp.h +++ b/include/net/tcp.h @@ -243,7 +243,6 @@ void tcp_time_wait(struct sock *sk, int state, int timeo); /* sysctl variables for tcp */ -extern int sysctl_tcp_stdurg; extern int sysctl_tcp_rfc1337; extern int sysctl_tcp_abort_on_overflow; extern int sysctl_tcp_max_orphans; diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c index 533b92ad39dd..a34bb75815c1 100644 --- a/net/ipv4/sysctl_net_ipv4.c +++ b/net/ipv4/sysctl_net_ipv4.c @@ -400,13 +400,6 @@ static struct ctl_table ipv4_table[] = { .mode = 0644, .proc_handler = proc_dointvec }, - { - .procname = "tcp_stdurg", - .data = &sysctl_tcp_stdurg, - .maxlen = sizeof(int), - .mode = 0644, - .proc_handler = proc_dointvec - }, { .procname = "tcp_rfc1337", .data = &sysctl_tcp_rfc1337, @@ -1145,6 +1138,13 @@ static struct ctl_table ipv4_net_table[] = { .mode = 0644, .proc_handler = proc_dointvec }, + { + .procname = "tcp_stdurg", + .data = &init_net.ipv4.sysctl_tcp_stdurg, + .maxlen = sizeof(int), + .mode = 0644, + .proc_handler = proc_dointvec + }, { } }; diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 5b2272dbf6a9..14b06963c102 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -89,7 +89,6 @@ EXPORT_SYMBOL(sysctl_tcp_adv_win_scale); /* rfc5961 challenge ack rate limiting */ int sysctl_tcp_challenge_ack_limit = 1000; -int sysctl_tcp_stdurg __read_mostly; int sysctl_tcp_rfc1337 __read_mostly; int sysctl_tcp_max_orphans __read_mostly = NR_FILE; int sysctl_tcp_frto __read_mostly = 2; @@ -5123,7 +5122,7 @@ static void tcp_check_urg(struct sock *sk, const struct tcphdr *th) struct tcp_sock *tp = tcp_sk(sk); u32 ptr = ntohs(th->urg_ptr); - if (ptr && !sysctl_tcp_stdurg) + if (ptr && !sock_net(sk)->ipv4.sysctl_tcp_stdurg) ptr--; ptr += ntohl(th->seq); From 625357aa175c688d219da43c8cfaa2e1629e0e1a Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Thu, 26 Oct 2017 21:55:02 -0700 Subject: [PATCH 07/15] tcp: Namespace-ify sysctl_tcp_rfc1337 Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- include/net/netns/ipv4.h | 1 + include/net/tcp.h | 1 - net/ipv4/sysctl_net_ipv4.c | 14 +++++++------- net/ipv4/tcp_input.c | 1 - net/ipv4/tcp_minisocks.c | 2 +- 5 files changed, 9 insertions(+), 10 deletions(-) diff --git a/include/net/netns/ipv4.h b/include/net/netns/ipv4.h index ffa2cf3dc747..968edce38eb5 100644 --- a/include/net/netns/ipv4.h +++ b/include/net/netns/ipv4.h @@ -134,6 +134,7 @@ struct netns_ipv4 { int sysctl_tcp_slow_start_after_idle; int sysctl_tcp_retrans_collapse; int sysctl_tcp_stdurg; + int sysctl_tcp_rfc1337; struct inet_timewait_death_row tcp_death_row; int sysctl_max_syn_backlog; int sysctl_tcp_fastopen; diff --git a/include/net/tcp.h b/include/net/tcp.h index cf3fac7008d7..2aea2b3373b3 100644 --- a/include/net/tcp.h +++ b/include/net/tcp.h @@ -243,7 +243,6 @@ void tcp_time_wait(struct sock *sk, int state, int timeo); /* sysctl variables for tcp */ -extern int sysctl_tcp_rfc1337; extern int sysctl_tcp_abort_on_overflow; extern int sysctl_tcp_max_orphans; extern int sysctl_tcp_fack; diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c index a34bb75815c1..832e554235df 100644 --- a/net/ipv4/sysctl_net_ipv4.c +++ b/net/ipv4/sysctl_net_ipv4.c @@ -400,13 +400,6 @@ static struct ctl_table ipv4_table[] = { .mode = 0644, .proc_handler = proc_dointvec }, - { - .procname = "tcp_rfc1337", - .data = &sysctl_tcp_rfc1337, - .maxlen = sizeof(int), - .mode = 0644, - .proc_handler = proc_dointvec - }, { .procname = "inet_peer_threshold", .data = &inet_peer_threshold, @@ -1145,6 +1138,13 @@ static struct ctl_table ipv4_net_table[] = { .mode = 0644, .proc_handler = proc_dointvec }, + { + .procname = "tcp_rfc1337", + .data = &init_net.ipv4.sysctl_tcp_rfc1337, + .maxlen = sizeof(int), + .mode = 0644, + .proc_handler = proc_dointvec + }, { } }; diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 14b06963c102..64fde81b0eb7 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -89,7 +89,6 @@ EXPORT_SYMBOL(sysctl_tcp_adv_win_scale); /* rfc5961 challenge ack rate limiting */ int sysctl_tcp_challenge_ack_limit = 1000; -int sysctl_tcp_rfc1337 __read_mostly; int sysctl_tcp_max_orphans __read_mostly = NR_FILE; int sysctl_tcp_frto __read_mostly = 2; int sysctl_tcp_min_rtt_wlen __read_mostly = 300; diff --git a/net/ipv4/tcp_minisocks.c b/net/ipv4/tcp_minisocks.c index 056009f1c14f..11836667763c 100644 --- a/net/ipv4/tcp_minisocks.c +++ b/net/ipv4/tcp_minisocks.c @@ -181,7 +181,7 @@ tcp_timewait_state_process(struct inet_timewait_sock *tw, struct sk_buff *skb, * Oh well... nobody has a sufficient solution to this * protocol bug yet. */ - if (sysctl_tcp_rfc1337 == 0) { + if (twsk_net(tw)->ipv4.sysctl_tcp_rfc1337 == 0) { kill: inet_twsk_deschedule_put(tw); return TCP_TW_SUCCESS; From 65c9410cf55ecf32da1b720f563365d565d6289a Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Thu, 26 Oct 2017 21:55:03 -0700 Subject: [PATCH 08/15] tcp: Namespace-ify sysctl_tcp_abort_on_overflow Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- include/net/netns/ipv4.h | 1 + include/net/tcp.h | 1 - net/ipv4/sysctl_net_ipv4.c | 14 +++++++------- net/ipv4/tcp_minisocks.c | 4 +--- 4 files changed, 9 insertions(+), 11 deletions(-) diff --git a/include/net/netns/ipv4.h b/include/net/netns/ipv4.h index 968edce38eb5..3875fdf6b186 100644 --- a/include/net/netns/ipv4.h +++ b/include/net/netns/ipv4.h @@ -135,6 +135,7 @@ struct netns_ipv4 { int sysctl_tcp_retrans_collapse; int sysctl_tcp_stdurg; int sysctl_tcp_rfc1337; + int sysctl_tcp_abort_on_overflow; struct inet_timewait_death_row tcp_death_row; int sysctl_max_syn_backlog; int sysctl_tcp_fastopen; diff --git a/include/net/tcp.h b/include/net/tcp.h index 2aea2b3373b3..7331281a2292 100644 --- a/include/net/tcp.h +++ b/include/net/tcp.h @@ -243,7 +243,6 @@ void tcp_time_wait(struct sock *sk, int state, int timeo); /* sysctl variables for tcp */ -extern int sysctl_tcp_abort_on_overflow; extern int sysctl_tcp_max_orphans; extern int sysctl_tcp_fack; extern int sysctl_tcp_reordering; diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c index 832e554235df..ffd1fd769bba 100644 --- a/net/ipv4/sysctl_net_ipv4.c +++ b/net/ipv4/sysctl_net_ipv4.c @@ -393,13 +393,6 @@ static struct ctl_table ipv4_table[] = { .mode = 0644, .proc_handler = proc_dointvec }, - { - .procname = "tcp_abort_on_overflow", - .data = &sysctl_tcp_abort_on_overflow, - .maxlen = sizeof(int), - .mode = 0644, - .proc_handler = proc_dointvec - }, { .procname = "inet_peer_threshold", .data = &inet_peer_threshold, @@ -1145,6 +1138,13 @@ static struct ctl_table ipv4_net_table[] = { .mode = 0644, .proc_handler = proc_dointvec }, + { + .procname = "tcp_abort_on_overflow", + .data = &init_net.ipv4.sysctl_tcp_abort_on_overflow, + .maxlen = sizeof(int), + .mode = 0644, + .proc_handler = proc_dointvec + }, { } }; diff --git a/net/ipv4/tcp_minisocks.c b/net/ipv4/tcp_minisocks.c index 11836667763c..3674d63170b2 100644 --- a/net/ipv4/tcp_minisocks.c +++ b/net/ipv4/tcp_minisocks.c @@ -29,8 +29,6 @@ #include #include -int sysctl_tcp_abort_on_overflow __read_mostly; - static bool tcp_in_window(u32 seq, u32 end_seq, u32 s_win, u32 e_win) { if (seq == s_win) @@ -783,7 +781,7 @@ struct sock *tcp_check_req(struct sock *sk, struct sk_buff *skb, return inet_csk_complete_hashdance(sk, child, req, own_req); listen_overflow: - if (!sysctl_tcp_abort_on_overflow) { + if (!sock_net(sk)->ipv4.sysctl_tcp_abort_on_overflow) { inet_rsk(req)->acked = 1; return NULL; } From 0bc65a28ae2aeb14aab7f4a930e0d8cf4cad9dc4 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Thu, 26 Oct 2017 21:55:04 -0700 Subject: [PATCH 09/15] tcp: Namespace-ify sysctl_tcp_fack Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- include/net/netns/ipv4.h | 1 + include/net/tcp.h | 1 - net/ipv4/sysctl_net_ipv4.c | 14 +++++++------- net/ipv4/tcp.c | 2 +- net/ipv4/tcp_input.c | 3 +-- net/ipv4/tcp_minisocks.c | 2 +- 6 files changed, 11 insertions(+), 12 deletions(-) diff --git a/include/net/netns/ipv4.h b/include/net/netns/ipv4.h index 3875fdf6b186..f0e792beeea9 100644 --- a/include/net/netns/ipv4.h +++ b/include/net/netns/ipv4.h @@ -136,6 +136,7 @@ struct netns_ipv4 { int sysctl_tcp_stdurg; int sysctl_tcp_rfc1337; int sysctl_tcp_abort_on_overflow; + int sysctl_tcp_fack; struct inet_timewait_death_row tcp_death_row; int sysctl_max_syn_backlog; int sysctl_tcp_fastopen; diff --git a/include/net/tcp.h b/include/net/tcp.h index 7331281a2292..e7b15e9f6e28 100644 --- a/include/net/tcp.h +++ b/include/net/tcp.h @@ -244,7 +244,6 @@ void tcp_time_wait(struct sock *sk, int state, int timeo); /* sysctl variables for tcp */ extern int sysctl_tcp_max_orphans; -extern int sysctl_tcp_fack; extern int sysctl_tcp_reordering; extern int sysctl_tcp_max_reordering; extern int sysctl_tcp_dsack; diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c index ffd1fd769bba..1f23be13ce7b 100644 --- a/net/ipv4/sysctl_net_ipv4.c +++ b/net/ipv4/sysctl_net_ipv4.c @@ -414,13 +414,6 @@ static struct ctl_table ipv4_table[] = { .mode = 0644, .proc_handler = proc_dointvec_jiffies, }, - { - .procname = "tcp_fack", - .data = &sysctl_tcp_fack, - .maxlen = sizeof(int), - .mode = 0644, - .proc_handler = proc_dointvec - }, { .procname = "tcp_max_reordering", .data = &sysctl_tcp_max_reordering, @@ -1145,6 +1138,13 @@ static struct ctl_table ipv4_net_table[] = { .mode = 0644, .proc_handler = proc_dointvec }, + { + .procname = "tcp_fack", + .data = &init_net.ipv4.sysctl_tcp_fack, + .maxlen = sizeof(int), + .mode = 0644, + .proc_handler = proc_dointvec + }, { } }; diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c index f6e1c00e300e..c7c983f0f817 100644 --- a/net/ipv4/tcp.c +++ b/net/ipv4/tcp.c @@ -2517,7 +2517,7 @@ static int tcp_repair_options_est(struct sock *sk, return -EINVAL; tp->rx_opt.sack_ok |= TCP_SACK_SEEN; - if (sysctl_tcp_fack) + if (sock_net(sk)->ipv4.sysctl_tcp_fack) tcp_enable_fack(tp); break; case TCPOPT_TIMESTAMP: diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 64fde81b0eb7..c5b94460793f 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -79,7 +79,6 @@ #include #include -int sysctl_tcp_fack __read_mostly; int sysctl_tcp_max_reordering __read_mostly = 300; int sysctl_tcp_dsack __read_mostly = 1; int sysctl_tcp_app_win __read_mostly = 31; @@ -5720,7 +5719,7 @@ static int tcp_rcv_synsent_state_process(struct sock *sk, struct sk_buff *skb, tp->tcp_header_len = sizeof(struct tcphdr); } - if (tcp_is_sack(tp) && sysctl_tcp_fack) + if (tcp_is_sack(tp) && sock_net(sk)->ipv4.sysctl_tcp_fack) tcp_enable_fack(tp); tcp_sync_mss(sk, icsk->icsk_pmtu_cookie); diff --git a/net/ipv4/tcp_minisocks.c b/net/ipv4/tcp_minisocks.c index 3674d63170b2..3270ab8416ce 100644 --- a/net/ipv4/tcp_minisocks.c +++ b/net/ipv4/tcp_minisocks.c @@ -510,7 +510,7 @@ struct sock *tcp_create_openreq_child(const struct sock *sk, newtp->rx_opt.tstamp_ok = ireq->tstamp_ok; if ((newtp->rx_opt.sack_ok = ireq->sack_ok) != 0) { - if (sysctl_tcp_fack) + if (sock_net(sk)->ipv4.sysctl_tcp_fack) tcp_enable_fack(newtp); } newtp->window_clamp = req->rsk_window_clamp; From 773d4bb96ceca6829ae9928f6b002b93e2e62cdc Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Thu, 26 Oct 2017 21:55:05 -0700 Subject: [PATCH 10/15] tcp: remove stale sysctl_tcp_reordering This extern is no longer used. Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- include/net/tcp.h | 1 - 1 file changed, 1 deletion(-) diff --git a/include/net/tcp.h b/include/net/tcp.h index e7b15e9f6e28..fc134ba74c7d 100644 --- a/include/net/tcp.h +++ b/include/net/tcp.h @@ -244,7 +244,6 @@ void tcp_time_wait(struct sock *sk, int state, int timeo); /* sysctl variables for tcp */ extern int sysctl_tcp_max_orphans; -extern int sysctl_tcp_reordering; extern int sysctl_tcp_max_reordering; extern int sysctl_tcp_dsack; extern long sysctl_tcp_mem[3]; From c6e218035913e14952b04ceecf1a543205106fdb Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Thu, 26 Oct 2017 21:55:06 -0700 Subject: [PATCH 11/15] tcp: Namespace-ify sysctl_tcp_max_reordering Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- include/net/netns/ipv4.h | 1 + include/net/tcp.h | 1 - net/ipv4/sysctl_net_ipv4.c | 14 +++++++------- net/ipv4/tcp_input.c | 3 +-- net/ipv4/tcp_ipv4.c | 2 ++ 5 files changed, 11 insertions(+), 10 deletions(-) diff --git a/include/net/netns/ipv4.h b/include/net/netns/ipv4.h index f0e792beeea9..3f6844665a2f 100644 --- a/include/net/netns/ipv4.h +++ b/include/net/netns/ipv4.h @@ -137,6 +137,7 @@ struct netns_ipv4 { int sysctl_tcp_rfc1337; int sysctl_tcp_abort_on_overflow; int sysctl_tcp_fack; + int sysctl_tcp_max_reordering; struct inet_timewait_death_row tcp_death_row; int sysctl_max_syn_backlog; int sysctl_tcp_fastopen; diff --git a/include/net/tcp.h b/include/net/tcp.h index fc134ba74c7d..8cd286226a1e 100644 --- a/include/net/tcp.h +++ b/include/net/tcp.h @@ -244,7 +244,6 @@ void tcp_time_wait(struct sock *sk, int state, int timeo); /* sysctl variables for tcp */ extern int sysctl_tcp_max_orphans; -extern int sysctl_tcp_max_reordering; extern int sysctl_tcp_dsack; extern long sysctl_tcp_mem[3]; extern int sysctl_tcp_wmem[3]; diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c index 1f23be13ce7b..18cd228a2069 100644 --- a/net/ipv4/sysctl_net_ipv4.c +++ b/net/ipv4/sysctl_net_ipv4.c @@ -414,13 +414,6 @@ static struct ctl_table ipv4_table[] = { .mode = 0644, .proc_handler = proc_dointvec_jiffies, }, - { - .procname = "tcp_max_reordering", - .data = &sysctl_tcp_max_reordering, - .maxlen = sizeof(int), - .mode = 0644, - .proc_handler = proc_dointvec - }, { .procname = "tcp_dsack", .data = &sysctl_tcp_dsack, @@ -1145,6 +1138,13 @@ static struct ctl_table ipv4_net_table[] = { .mode = 0644, .proc_handler = proc_dointvec }, + { + .procname = "tcp_max_reordering", + .data = &init_net.ipv4.sysctl_tcp_max_reordering, + .maxlen = sizeof(int), + .mode = 0644, + .proc_handler = proc_dointvec + }, { } }; diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index c5b94460793f..c118657f06ee 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -79,7 +79,6 @@ #include #include -int sysctl_tcp_max_reordering __read_mostly = 300; int sysctl_tcp_dsack __read_mostly = 1; int sysctl_tcp_app_win __read_mostly = 31; int sysctl_tcp_adv_win_scale __read_mostly = 1; @@ -889,7 +888,7 @@ static void tcp_update_reordering(struct sock *sk, const int metric, return; if (metric > tp->reordering) { - tp->reordering = min(sysctl_tcp_max_reordering, metric); + tp->reordering = min(sock_net(sk)->ipv4.sysctl_tcp_max_reordering, metric); #if FASTRETRANS_DEBUG > 1 pr_debug("Disorder%d %d %u f%u s%u rr%d\n", diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c index 2bc6ba2059d3..c379a242abb3 100644 --- a/net/ipv4/tcp_ipv4.c +++ b/net/ipv4/tcp_ipv4.c @@ -2488,6 +2488,8 @@ static int __net_init tcp_sk_init(struct net *net) net->ipv4.sysctl_tcp_recovery = TCP_RACK_LOSS_DETECTION; net->ipv4.sysctl_tcp_slow_start_after_idle = 1; /* By default, RFC2861 behavior. */ net->ipv4.sysctl_tcp_retrans_collapse = 1; + net->ipv4.sysctl_tcp_max_reordering = 300; + net->ipv4.sysctl_tcp_fastopen = TFO_CLIENT_ENABLE; spin_lock_init(&net->ipv4.tcp_fastopen_ctx_lock); net->ipv4.sysctl_tcp_fastopen_blackhole_timeout = 60 * 60; From 6496f6bde0c323fba5e8c5b5cbf3a7bf28dad7ed Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Thu, 26 Oct 2017 21:55:07 -0700 Subject: [PATCH 12/15] tcp: Namespace-ify sysctl_tcp_dsack Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- include/net/netns/ipv4.h | 1 + include/net/tcp.h | 1 - net/ipv4/sysctl_net_ipv4.c | 14 +++++++------- net/ipv4/tcp_input.c | 5 ++--- net/ipv4/tcp_ipv4.c | 1 + 5 files changed, 11 insertions(+), 11 deletions(-) diff --git a/include/net/netns/ipv4.h b/include/net/netns/ipv4.h index 3f6844665a2f..956957a77db9 100644 --- a/include/net/netns/ipv4.h +++ b/include/net/netns/ipv4.h @@ -138,6 +138,7 @@ struct netns_ipv4 { int sysctl_tcp_abort_on_overflow; int sysctl_tcp_fack; int sysctl_tcp_max_reordering; + int sysctl_tcp_dsack; struct inet_timewait_death_row tcp_death_row; int sysctl_max_syn_backlog; int sysctl_tcp_fastopen; diff --git a/include/net/tcp.h b/include/net/tcp.h index 8cd286226a1e..8b2ae3e8d79f 100644 --- a/include/net/tcp.h +++ b/include/net/tcp.h @@ -244,7 +244,6 @@ void tcp_time_wait(struct sock *sk, int state, int timeo); /* sysctl variables for tcp */ extern int sysctl_tcp_max_orphans; -extern int sysctl_tcp_dsack; extern long sysctl_tcp_mem[3]; extern int sysctl_tcp_wmem[3]; extern int sysctl_tcp_rmem[3]; diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c index 18cd228a2069..7652a9c2a65d 100644 --- a/net/ipv4/sysctl_net_ipv4.c +++ b/net/ipv4/sysctl_net_ipv4.c @@ -414,13 +414,6 @@ static struct ctl_table ipv4_table[] = { .mode = 0644, .proc_handler = proc_dointvec_jiffies, }, - { - .procname = "tcp_dsack", - .data = &sysctl_tcp_dsack, - .maxlen = sizeof(int), - .mode = 0644, - .proc_handler = proc_dointvec - }, { .procname = "tcp_mem", .maxlen = sizeof(sysctl_tcp_mem), @@ -1145,6 +1138,13 @@ static struct ctl_table ipv4_net_table[] = { .mode = 0644, .proc_handler = proc_dointvec }, + { + .procname = "tcp_dsack", + .data = &init_net.ipv4.sysctl_tcp_dsack, + .maxlen = sizeof(int), + .mode = 0644, + .proc_handler = proc_dointvec + }, { } }; diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index c118657f06ee..fd77037ac800 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -79,7 +79,6 @@ #include #include -int sysctl_tcp_dsack __read_mostly = 1; int sysctl_tcp_app_win __read_mostly = 31; int sysctl_tcp_adv_win_scale __read_mostly = 1; EXPORT_SYMBOL(sysctl_tcp_adv_win_scale); @@ -4150,7 +4149,7 @@ static void tcp_dsack_set(struct sock *sk, u32 seq, u32 end_seq) { struct tcp_sock *tp = tcp_sk(sk); - if (tcp_is_sack(tp) && sysctl_tcp_dsack) { + if (tcp_is_sack(tp) && sock_net(sk)->ipv4.sysctl_tcp_dsack) { int mib_idx; if (before(seq, tp->rcv_nxt)) @@ -4185,7 +4184,7 @@ static void tcp_send_dupack(struct sock *sk, const struct sk_buff *skb) NET_INC_STATS(sock_net(sk), LINUX_MIB_DELAYEDACKLOST); tcp_enter_quickack_mode(sk); - if (tcp_is_sack(tp) && sysctl_tcp_dsack) { + if (tcp_is_sack(tp) && sock_net(sk)->ipv4.sysctl_tcp_dsack) { u32 end_seq = TCP_SKB_CB(skb)->end_seq; if (after(TCP_SKB_CB(skb)->end_seq, tp->rcv_nxt)) diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c index c379a242abb3..d9d4d191e8f3 100644 --- a/net/ipv4/tcp_ipv4.c +++ b/net/ipv4/tcp_ipv4.c @@ -2489,6 +2489,7 @@ static int __net_init tcp_sk_init(struct net *net) net->ipv4.sysctl_tcp_slow_start_after_idle = 1; /* By default, RFC2861 behavior. */ net->ipv4.sysctl_tcp_retrans_collapse = 1; net->ipv4.sysctl_tcp_max_reordering = 300; + net->ipv4.sysctl_tcp_dsack = 1; net->ipv4.sysctl_tcp_fastopen = TFO_CLIENT_ENABLE; spin_lock_init(&net->ipv4.tcp_fastopen_ctx_lock); From 0c12654ac6d9004b9538b2a969b2b59e9a5ed831 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Thu, 26 Oct 2017 21:55:08 -0700 Subject: [PATCH 13/15] tcp: Namespace-ify sysctl_tcp_app_win Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- include/net/netns/ipv4.h | 1 + include/net/tcp.h | 1 - net/ipv4/sysctl_net_ipv4.c | 14 +++++++------- net/ipv4/tcp_input.c | 8 ++++---- net/ipv4/tcp_ipv4.c | 1 + 5 files changed, 13 insertions(+), 12 deletions(-) diff --git a/include/net/netns/ipv4.h b/include/net/netns/ipv4.h index 956957a77db9..63f91d52cbc0 100644 --- a/include/net/netns/ipv4.h +++ b/include/net/netns/ipv4.h @@ -139,6 +139,7 @@ struct netns_ipv4 { int sysctl_tcp_fack; int sysctl_tcp_max_reordering; int sysctl_tcp_dsack; + int sysctl_tcp_app_win; struct inet_timewait_death_row tcp_death_row; int sysctl_max_syn_backlog; int sysctl_tcp_fastopen; diff --git a/include/net/tcp.h b/include/net/tcp.h index 8b2ae3e8d79f..7aa3d65062a1 100644 --- a/include/net/tcp.h +++ b/include/net/tcp.h @@ -247,7 +247,6 @@ extern int sysctl_tcp_max_orphans; extern long sysctl_tcp_mem[3]; extern int sysctl_tcp_wmem[3]; extern int sysctl_tcp_rmem[3]; -extern int sysctl_tcp_app_win; extern int sysctl_tcp_adv_win_scale; extern int sysctl_tcp_frto; extern int sysctl_tcp_nometrics_save; diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c index 7652a9c2a65d..e057788834a9 100644 --- a/net/ipv4/sysctl_net_ipv4.c +++ b/net/ipv4/sysctl_net_ipv4.c @@ -437,13 +437,6 @@ static struct ctl_table ipv4_table[] = { .proc_handler = proc_dointvec_minmax, .extra1 = &one, }, - { - .procname = "tcp_app_win", - .data = &sysctl_tcp_app_win, - .maxlen = sizeof(int), - .mode = 0644, - .proc_handler = proc_dointvec - }, { .procname = "tcp_adv_win_scale", .data = &sysctl_tcp_adv_win_scale, @@ -1145,6 +1138,13 @@ static struct ctl_table ipv4_net_table[] = { .mode = 0644, .proc_handler = proc_dointvec }, + { + .procname = "tcp_app_win", + .data = &init_net.ipv4.sysctl_tcp_app_win, + .maxlen = sizeof(int), + .mode = 0644, + .proc_handler = proc_dointvec + }, { } }; diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index fd77037ac800..6af4b58ac6d5 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -79,7 +79,6 @@ #include #include -int sysctl_tcp_app_win __read_mostly = 31; int sysctl_tcp_adv_win_scale __read_mostly = 1; EXPORT_SYMBOL(sysctl_tcp_adv_win_scale); @@ -428,6 +427,7 @@ static void tcp_fixup_rcvbuf(struct sock *sk) */ void tcp_init_buffer_space(struct sock *sk) { + int tcp_app_win = sock_net(sk)->ipv4.sysctl_tcp_app_win; struct tcp_sock *tp = tcp_sk(sk); int maxwin; @@ -446,14 +446,14 @@ void tcp_init_buffer_space(struct sock *sk) if (tp->window_clamp >= maxwin) { tp->window_clamp = maxwin; - if (sysctl_tcp_app_win && maxwin > 4 * tp->advmss) + if (tcp_app_win && maxwin > 4 * tp->advmss) tp->window_clamp = max(maxwin - - (maxwin >> sysctl_tcp_app_win), + (maxwin >> tcp_app_win), 4 * tp->advmss); } /* Force reservation of one segment. */ - if (sysctl_tcp_app_win && + if (tcp_app_win && tp->window_clamp > 2 * tp->advmss && tp->window_clamp + tp->advmss > maxwin) tp->window_clamp = max(2 * tp->advmss, maxwin - tp->advmss); diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c index d9d4d191e8f3..189664ebd28e 100644 --- a/net/ipv4/tcp_ipv4.c +++ b/net/ipv4/tcp_ipv4.c @@ -2490,6 +2490,7 @@ static int __net_init tcp_sk_init(struct net *net) net->ipv4.sysctl_tcp_retrans_collapse = 1; net->ipv4.sysctl_tcp_max_reordering = 300; net->ipv4.sysctl_tcp_dsack = 1; + net->ipv4.sysctl_tcp_app_win = 31; net->ipv4.sysctl_tcp_fastopen = TFO_CLIENT_ENABLE; spin_lock_init(&net->ipv4.tcp_fastopen_ctx_lock); From 94f0893e0c27219f4a726932618505aab6795973 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Thu, 26 Oct 2017 21:55:09 -0700 Subject: [PATCH 14/15] tcp: Namespace-ify sysctl_tcp_adv_win_scale Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- include/net/netns/ipv4.h | 1 + include/net/tcp.h | 9 ++++----- net/ipv4/sysctl_net_ipv4.c | 18 +++++++++--------- net/ipv4/tcp_input.c | 13 +++++-------- net/ipv4/tcp_ipv4.c | 1 + 5 files changed, 20 insertions(+), 22 deletions(-) diff --git a/include/net/netns/ipv4.h b/include/net/netns/ipv4.h index 63f91d52cbc0..9dbb07d4eff4 100644 --- a/include/net/netns/ipv4.h +++ b/include/net/netns/ipv4.h @@ -140,6 +140,7 @@ struct netns_ipv4 { int sysctl_tcp_max_reordering; int sysctl_tcp_dsack; int sysctl_tcp_app_win; + int sysctl_tcp_adv_win_scale; struct inet_timewait_death_row tcp_death_row; int sysctl_max_syn_backlog; int sysctl_tcp_fastopen; diff --git a/include/net/tcp.h b/include/net/tcp.h index 7aa3d65062a1..0dc27cd24899 100644 --- a/include/net/tcp.h +++ b/include/net/tcp.h @@ -247,7 +247,6 @@ extern int sysctl_tcp_max_orphans; extern long sysctl_tcp_mem[3]; extern int sysctl_tcp_wmem[3]; extern int sysctl_tcp_rmem[3]; -extern int sysctl_tcp_adv_win_scale; extern int sysctl_tcp_frto; extern int sysctl_tcp_nometrics_save; extern int sysctl_tcp_moderate_rcvbuf; @@ -1311,9 +1310,9 @@ void tcp_select_initial_window(int __space, __u32 mss, __u32 *rcv_wnd, __u32 *window_clamp, int wscale_ok, __u8 *rcv_wscale, __u32 init_rcv_wnd); -static inline int tcp_win_from_space(int space) +static inline int tcp_win_from_space(const struct sock *sk, int space) { - int tcp_adv_win_scale = sysctl_tcp_adv_win_scale; + int tcp_adv_win_scale = sock_net(sk)->ipv4.sysctl_tcp_adv_win_scale; return tcp_adv_win_scale <= 0 ? (space>>(-tcp_adv_win_scale)) : @@ -1323,13 +1322,13 @@ static inline int tcp_win_from_space(int space) /* Note: caller must be prepared to deal with negative returns */ static inline int tcp_space(const struct sock *sk) { - return tcp_win_from_space(sk->sk_rcvbuf - + return tcp_win_from_space(sk, sk->sk_rcvbuf - atomic_read(&sk->sk_rmem_alloc)); } static inline int tcp_full_space(const struct sock *sk) { - return tcp_win_from_space(sk->sk_rcvbuf); + return tcp_win_from_space(sk, sk->sk_rcvbuf); } extern void tcp_openreq_init_rwin(struct request_sock *req, diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c index e057788834a9..a95123e1e7da 100644 --- a/net/ipv4/sysctl_net_ipv4.c +++ b/net/ipv4/sysctl_net_ipv4.c @@ -437,15 +437,6 @@ static struct ctl_table ipv4_table[] = { .proc_handler = proc_dointvec_minmax, .extra1 = &one, }, - { - .procname = "tcp_adv_win_scale", - .data = &sysctl_tcp_adv_win_scale, - .maxlen = sizeof(int), - .mode = 0644, - .proc_handler = proc_dointvec_minmax, - .extra1 = &tcp_adv_win_scale_min, - .extra2 = &tcp_adv_win_scale_max, - }, { .procname = "tcp_frto", .data = &sysctl_tcp_frto, @@ -1145,6 +1136,15 @@ static struct ctl_table ipv4_net_table[] = { .mode = 0644, .proc_handler = proc_dointvec }, + { + .procname = "tcp_adv_win_scale", + .data = &init_net.ipv4.sysctl_tcp_adv_win_scale, + .maxlen = sizeof(int), + .mode = 0644, + .proc_handler = proc_dointvec_minmax, + .extra1 = &tcp_adv_win_scale_min, + .extra2 = &tcp_adv_win_scale_max, + }, { } }; diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 6af4b58ac6d5..8ee2c84b0bc6 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -79,9 +79,6 @@ #include #include -int sysctl_tcp_adv_win_scale __read_mostly = 1; -EXPORT_SYMBOL(sysctl_tcp_adv_win_scale); - /* rfc5961 challenge ack rate limiting */ int sysctl_tcp_challenge_ack_limit = 1000; @@ -363,8 +360,8 @@ static int __tcp_grow_window(const struct sock *sk, const struct sk_buff *skb) { struct tcp_sock *tp = tcp_sk(sk); /* Optimize this! */ - int truesize = tcp_win_from_space(skb->truesize) >> 1; - int window = tcp_win_from_space(sysctl_tcp_rmem[2]) >> 1; + int truesize = tcp_win_from_space(sk, skb->truesize) >> 1; + int window = tcp_win_from_space(sk, sysctl_tcp_rmem[2]) >> 1; while (tp->rcv_ssthresh <= window) { if (truesize <= skb->len) @@ -389,7 +386,7 @@ static void tcp_grow_window(struct sock *sk, const struct sk_buff *skb) /* Check #2. Increase window, if skb with such overhead * will fit to rcvbuf in future. */ - if (tcp_win_from_space(skb->truesize) <= skb->len) + if (tcp_win_from_space(sk, skb->truesize) <= skb->len) incr = 2 * tp->advmss; else incr = __tcp_grow_window(sk, skb); @@ -630,7 +627,7 @@ void tcp_rcv_space_adjust(struct sock *sk) } rcvmem = SKB_TRUESIZE(tp->advmss + MAX_TCP_HEADER); - while (tcp_win_from_space(rcvmem) < tp->advmss) + while (tcp_win_from_space(sk, rcvmem) < tp->advmss) rcvmem += 128; rcvbuf = min(rcvwin / tp->advmss * rcvmem, sysctl_tcp_rmem[2]); @@ -4809,7 +4806,7 @@ tcp_collapse(struct sock *sk, struct sk_buff_head *list, struct rb_root *root, * overlaps to the next one. */ if (!(TCP_SKB_CB(skb)->tcp_flags & (TCPHDR_SYN | TCPHDR_FIN)) && - (tcp_win_from_space(skb->truesize) > skb->len || + (tcp_win_from_space(sk, skb->truesize) > skb->len || before(TCP_SKB_CB(skb)->seq, start))) { end_of_skbs = false; break; diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c index 189664ebd28e..1fe30fb99308 100644 --- a/net/ipv4/tcp_ipv4.c +++ b/net/ipv4/tcp_ipv4.c @@ -2491,6 +2491,7 @@ static int __net_init tcp_sk_init(struct net *net) net->ipv4.sysctl_tcp_max_reordering = 300; net->ipv4.sysctl_tcp_dsack = 1; net->ipv4.sysctl_tcp_app_win = 31; + net->ipv4.sysctl_tcp_adv_win_scale = 1; net->ipv4.sysctl_tcp_fastopen = TFO_CLIENT_ENABLE; spin_lock_init(&net->ipv4.tcp_fastopen_ctx_lock); From af9b69a7a6ca6b817e8d6f416e7aa5b2a5bf1d91 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Thu, 26 Oct 2017 21:55:10 -0700 Subject: [PATCH 15/15] tcp: Namespace-ify sysctl_tcp_frto Signed-off-by: Eric Dumazet Signed-off-by: David S. Miller --- include/net/netns/ipv4.h | 1 + include/net/tcp.h | 1 - net/ipv4/sysctl_net_ipv4.c | 14 +++++++------- net/ipv4/tcp_input.c | 3 +-- net/ipv4/tcp_ipv4.c | 1 + 5 files changed, 10 insertions(+), 10 deletions(-) diff --git a/include/net/netns/ipv4.h b/include/net/netns/ipv4.h index 9dbb07d4eff4..f4622e28db3a 100644 --- a/include/net/netns/ipv4.h +++ b/include/net/netns/ipv4.h @@ -141,6 +141,7 @@ struct netns_ipv4 { int sysctl_tcp_dsack; int sysctl_tcp_app_win; int sysctl_tcp_adv_win_scale; + int sysctl_tcp_frto; struct inet_timewait_death_row tcp_death_row; int sysctl_max_syn_backlog; int sysctl_tcp_fastopen; diff --git a/include/net/tcp.h b/include/net/tcp.h index 0dc27cd24899..18f047501f53 100644 --- a/include/net/tcp.h +++ b/include/net/tcp.h @@ -247,7 +247,6 @@ extern int sysctl_tcp_max_orphans; extern long sysctl_tcp_mem[3]; extern int sysctl_tcp_wmem[3]; extern int sysctl_tcp_rmem[3]; -extern int sysctl_tcp_frto; extern int sysctl_tcp_nometrics_save; extern int sysctl_tcp_moderate_rcvbuf; extern int sysctl_tcp_tso_win_divisor; diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c index a95123e1e7da..f1bcb9b7e082 100644 --- a/net/ipv4/sysctl_net_ipv4.c +++ b/net/ipv4/sysctl_net_ipv4.c @@ -437,13 +437,6 @@ static struct ctl_table ipv4_table[] = { .proc_handler = proc_dointvec_minmax, .extra1 = &one, }, - { - .procname = "tcp_frto", - .data = &sysctl_tcp_frto, - .maxlen = sizeof(int), - .mode = 0644, - .proc_handler = proc_dointvec - }, { .procname = "tcp_min_rtt_wlen", .data = &sysctl_tcp_min_rtt_wlen, @@ -1145,6 +1138,13 @@ static struct ctl_table ipv4_net_table[] = { .extra1 = &tcp_adv_win_scale_min, .extra2 = &tcp_adv_win_scale_max, }, + { + .procname = "tcp_frto", + .data = &init_net.ipv4.sysctl_tcp_frto, + .maxlen = sizeof(int), + .mode = 0644, + .proc_handler = proc_dointvec + }, { } }; diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c index 8ee2c84b0bc6..90d76f1c8f96 100644 --- a/net/ipv4/tcp_input.c +++ b/net/ipv4/tcp_input.c @@ -83,7 +83,6 @@ int sysctl_tcp_challenge_ack_limit = 1000; int sysctl_tcp_max_orphans __read_mostly = NR_FILE; -int sysctl_tcp_frto __read_mostly = 2; int sysctl_tcp_min_rtt_wlen __read_mostly = 300; int sysctl_tcp_moderate_rcvbuf __read_mostly = 1; int sysctl_tcp_invalid_ratelimit __read_mostly = HZ/2; @@ -2026,7 +2025,7 @@ void tcp_enter_loss(struct sock *sk) * falsely raise the receive window, which results in repeated * timeouts and stop-and-go behavior. */ - tp->frto = sysctl_tcp_frto && + tp->frto = net->ipv4.sysctl_tcp_frto && (new_recovery || icsk->icsk_retransmits) && !inet_csk(sk)->icsk_mtup.probe_size; } diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c index 1fe30fb99308..49757c758211 100644 --- a/net/ipv4/tcp_ipv4.c +++ b/net/ipv4/tcp_ipv4.c @@ -2492,6 +2492,7 @@ static int __net_init tcp_sk_init(struct net *net) net->ipv4.sysctl_tcp_dsack = 1; net->ipv4.sysctl_tcp_app_win = 31; net->ipv4.sysctl_tcp_adv_win_scale = 1; + net->ipv4.sysctl_tcp_frto = 2; net->ipv4.sysctl_tcp_fastopen = TFO_CLIENT_ENABLE; spin_lock_init(&net->ipv4.tcp_fastopen_ctx_lock);