* [PATCH net-next 00/12] tcp: move 12 sysctls to namespaces
@ 2017-10-27 14:47 Eric Dumazet
2017-10-27 14:47 ` [PATCH net-next 01/12] tcp: Namespace-ify sysctl_tcp_nometrics_save Eric Dumazet
` (12 more replies)
0 siblings, 13 replies; 14+ messages in thread
From: Eric Dumazet @ 2017-10-27 14:47 UTC (permalink / raw)
To: David S . Miller; +Cc: netdev, Eric Dumazet, Eric Dumazet
Ideally all TCP sysctls should be per netns.
This patch series takes care of 12 sysctls.
Remains the ones that need discussion :
sysctl_tcp_mem, sysctl_tcp_rmem, sysctl_tcp_wmem, and sysctl_tcp_max_orphans
Eric Dumazet (12):
tcp: Namespace-ify sysctl_tcp_nometrics_save
tcp: Namespace-ify sysctl_tcp_moderate_rcvbuf
tcp: Namespace-ify sysctl_tcp_tso_win_divisor
tcp: Namespace-ify sysctl_tcp_workaround_signed_windows
tcp: Namespace-ify sysctl_tcp_limit_output_bytes
tcp: Namespace-ify sysctl_tcp_challenge_ack_limit
tcp: Namespace-ify sysctl_tcp_min_tso_segs
tcp: Namespace-ify sysctl_tcp_min_rtt_wlen
tcp: Namespace-ify sysctl_tcp_autocorking
tcp: Namespace-ify sysctl_tcp_invalid_ratelimit
tcp: Namespace-ify sysctl_tcp_pacing_ss_ratio
tcp: Namespace-ify sysctl_tcp_pacing_ca_ratio
include/net/netns/ipv4.h | 12 +++
include/net/tcp.h | 16 +---
net/ipv4/syncookies.c | 2 +-
net/ipv4/sysctl_net_ipv4.c | 184 ++++++++++++++++++++++-----------------------
net/ipv4/tcp.c | 6 +-
net/ipv4/tcp_input.c | 38 +++-------
net/ipv4/tcp_ipv4.c | 16 ++++
net/ipv4/tcp_metrics.c | 4 +-
net/ipv4/tcp_minisocks.c | 2 +-
net/ipv4/tcp_output.c | 31 +++-----
net/ipv6/syncookies.c | 2 +-
11 files changed, 149 insertions(+), 164 deletions(-)
--
2.15.0.rc2.357.g7e34df9404-goog
^ permalink raw reply [flat|nested] 14+ messages in thread
* [PATCH net-next 01/12] tcp: Namespace-ify sysctl_tcp_nometrics_save
2017-10-27 14:47 [PATCH net-next 00/12] tcp: move 12 sysctls to namespaces Eric Dumazet
@ 2017-10-27 14:47 ` Eric Dumazet
2017-10-27 14:47 ` [PATCH net-next 02/12] tcp: Namespace-ify sysctl_tcp_moderate_rcvbuf Eric Dumazet
` (11 subsequent siblings)
12 siblings, 0 replies; 14+ messages in thread
From: Eric Dumazet @ 2017-10-27 14:47 UTC (permalink / raw)
To: David S . Miller; +Cc: netdev, Eric Dumazet, Eric Dumazet
Signed-off-by: Eric Dumazet <edumazet@google.com>
---
include/net/netns/ipv4.h | 1 +
include/net/tcp.h | 1 -
net/ipv4/sysctl_net_ipv4.c | 14 +++++++-------
net/ipv4/tcp_metrics.c | 4 +---
4 files changed, 9 insertions(+), 11 deletions(-)
diff --git a/include/net/netns/ipv4.h b/include/net/netns/ipv4.h
index f4622e28db3a1484553f51709b144ee769766a28..9606e2ea1f1428744c7a734bf6cc58944889a8dd 100644
--- a/include/net/netns/ipv4.h
+++ b/include/net/netns/ipv4.h
@@ -142,6 +142,7 @@ struct netns_ipv4 {
int sysctl_tcp_app_win;
int sysctl_tcp_adv_win_scale;
int sysctl_tcp_frto;
+ int sysctl_tcp_nometrics_save;
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 18f047501f53be3780bd41a5c8234adf9683cebf..6ab7fa4154b27aad8d70d02f48d1c0aa9302839a 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_nometrics_save;
extern int sysctl_tcp_moderate_rcvbuf;
extern int sysctl_tcp_tso_win_divisor;
extern int sysctl_tcp_workaround_signed_windows;
diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c
index f1bcb9b7e082c6688fad12e15be9b872ebed8151..b742a5e26a9d9c5a2ad730fda649bdd152b23988 100644
--- a/net/ipv4/sysctl_net_ipv4.c
+++ b/net/ipv4/sysctl_net_ipv4.c
@@ -451,13 +451,6 @@ static struct ctl_table ipv4_table[] = {
.mode = 0644,
.proc_handler = proc_dointvec
},
- {
- .procname = "tcp_no_metrics_save",
- .data = &sysctl_tcp_nometrics_save,
- .maxlen = sizeof(int),
- .mode = 0644,
- .proc_handler = proc_dointvec,
- },
{
.procname = "tcp_moderate_rcvbuf",
.data = &sysctl_tcp_moderate_rcvbuf,
@@ -1145,6 +1138,13 @@ static struct ctl_table ipv4_net_table[] = {
.mode = 0644,
.proc_handler = proc_dointvec
},
+ {
+ .procname = "tcp_no_metrics_save",
+ .data = &init_net.ipv4.sysctl_tcp_nometrics_save,
+ .maxlen = sizeof(int),
+ .mode = 0644,
+ .proc_handler = proc_dointvec,
+ },
{ }
};
diff --git a/net/ipv4/tcp_metrics.c b/net/ipv4/tcp_metrics.c
index 0ab78abc811bef0388089befed672e3d4ee9d881..0507b56b6d4b2721ebce42d2a89a6638350c3b74 100644
--- a/net/ipv4/tcp_metrics.c
+++ b/net/ipv4/tcp_metrics.c
@@ -20,8 +20,6 @@
#include <net/tcp.h>
#include <net/genetlink.h>
-int sysctl_tcp_nometrics_save __read_mostly;
-
static struct tcp_metrics_block *__tcp_get_metrics(const struct inetpeer_addr *saddr,
const struct inetpeer_addr *daddr,
struct net *net, unsigned int hash);
@@ -330,7 +328,7 @@ void tcp_update_metrics(struct sock *sk)
int m;
sk_dst_confirm(sk);
- if (sysctl_tcp_nometrics_save || !dst)
+ if (net->ipv4.sysctl_tcp_nometrics_save || !dst)
return;
rcu_read_lock();
--
2.15.0.rc2.357.g7e34df9404-goog
^ permalink raw reply related [flat|nested] 14+ messages in thread
* [PATCH net-next 02/12] tcp: Namespace-ify sysctl_tcp_moderate_rcvbuf
2017-10-27 14:47 [PATCH net-next 00/12] tcp: move 12 sysctls to namespaces Eric Dumazet
2017-10-27 14:47 ` [PATCH net-next 01/12] tcp: Namespace-ify sysctl_tcp_nometrics_save Eric Dumazet
@ 2017-10-27 14:47 ` Eric Dumazet
2017-10-27 14:47 ` [PATCH net-next 03/12] tcp: Namespace-ify sysctl_tcp_tso_win_divisor Eric Dumazet
` (10 subsequent siblings)
12 siblings, 0 replies; 14+ messages in thread
From: Eric Dumazet @ 2017-10-27 14:47 UTC (permalink / raw)
To: David S . Miller; +Cc: netdev, Eric Dumazet, Eric Dumazet
Signed-off-by: Eric Dumazet <edumazet@google.com>
---
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 9606e2ea1f1428744c7a734bf6cc58944889a8dd..4458a54fe3f4e9dc79561039dba0f4d19ee8271e 100644
--- a/include/net/netns/ipv4.h
+++ b/include/net/netns/ipv4.h
@@ -143,6 +143,7 @@ struct netns_ipv4 {
int sysctl_tcp_adv_win_scale;
int sysctl_tcp_frto;
int sysctl_tcp_nometrics_save;
+ int sysctl_tcp_moderate_rcvbuf;
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 6ab7fa4154b27aad8d70d02f48d1c0aa9302839a..f954e74578fff2e52cf89521132c72fe5a047d13 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_moderate_rcvbuf;
extern int sysctl_tcp_tso_win_divisor;
extern int sysctl_tcp_workaround_signed_windows;
diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c
index b742a5e26a9d9c5a2ad730fda649bdd152b23988..2ebe87fd11696948ab87349ba233a16eccb98b44 100644
--- a/net/ipv4/sysctl_net_ipv4.c
+++ b/net/ipv4/sysctl_net_ipv4.c
@@ -451,13 +451,6 @@ static struct ctl_table ipv4_table[] = {
.mode = 0644,
.proc_handler = proc_dointvec
},
- {
- .procname = "tcp_moderate_rcvbuf",
- .data = &sysctl_tcp_moderate_rcvbuf,
- .maxlen = sizeof(int),
- .mode = 0644,
- .proc_handler = proc_dointvec,
- },
{
.procname = "tcp_tso_win_divisor",
.data = &sysctl_tcp_tso_win_divisor,
@@ -1145,6 +1138,13 @@ static struct ctl_table ipv4_net_table[] = {
.mode = 0644,
.proc_handler = proc_dointvec,
},
+ {
+ .procname = "tcp_moderate_rcvbuf",
+ .data = &init_net.ipv4.sysctl_tcp_moderate_rcvbuf,
+ .maxlen = sizeof(int),
+ .mode = 0644,
+ .proc_handler = proc_dointvec,
+ },
{ }
};
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 90d76f1c8f96bc89618ddc59ae237a34cd25db7c..ce481325115fe7cb8a3b007276f2ce2a519a7383 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -84,7 +84,6 @@ int sysctl_tcp_challenge_ack_limit = 1000;
int sysctl_tcp_max_orphans __read_mostly = NR_FILE;
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;
#define FLAG_DATA 0x01 /* Incoming frame contained data. */
@@ -411,7 +410,7 @@ static void tcp_fixup_rcvbuf(struct sock *sk)
/* Dynamic Right Sizing (DRS) has 2 to 3 RTT latency
* Allow enough cushion so that sender is not limited by our window
*/
- if (sysctl_tcp_moderate_rcvbuf)
+ if (sock_net(sk)->ipv4.sysctl_tcp_moderate_rcvbuf)
rcvmem <<= 2;
if (sk->sk_rcvbuf < rcvmem)
@@ -602,7 +601,7 @@ void tcp_rcv_space_adjust(struct sock *sk)
* <prev RTT . ><current RTT .. ><next RTT .... >
*/
- if (sysctl_tcp_moderate_rcvbuf &&
+ if (sock_net(sk)->ipv4.sysctl_tcp_moderate_rcvbuf &&
!(sk->sk_userlocks & SOCK_RCVBUF_LOCK)) {
int rcvwin, rcvmem, rcvbuf;
diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c
index 49757c75821111c6d2cf413415be2c1b58482659..27f376b90913519493e78e1f3a198a3a840892f1 100644
--- a/net/ipv4/tcp_ipv4.c
+++ b/net/ipv4/tcp_ipv4.c
@@ -2493,6 +2493,7 @@ static int __net_init tcp_sk_init(struct net *net)
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_moderate_rcvbuf = 1;
net->ipv4.sysctl_tcp_fastopen = TFO_CLIENT_ENABLE;
spin_lock_init(&net->ipv4.tcp_fastopen_ctx_lock);
--
2.15.0.rc2.357.g7e34df9404-goog
^ permalink raw reply related [flat|nested] 14+ messages in thread
* [PATCH net-next 03/12] tcp: Namespace-ify sysctl_tcp_tso_win_divisor
2017-10-27 14:47 [PATCH net-next 00/12] tcp: move 12 sysctls to namespaces Eric Dumazet
2017-10-27 14:47 ` [PATCH net-next 01/12] tcp: Namespace-ify sysctl_tcp_nometrics_save Eric Dumazet
2017-10-27 14:47 ` [PATCH net-next 02/12] tcp: Namespace-ify sysctl_tcp_moderate_rcvbuf Eric Dumazet
@ 2017-10-27 14:47 ` Eric Dumazet
2017-10-27 14:47 ` [PATCH net-next 04/12] tcp: Namespace-ify sysctl_tcp_workaround_signed_windows Eric Dumazet
` (9 subsequent siblings)
12 siblings, 0 replies; 14+ messages in thread
From: Eric Dumazet @ 2017-10-27 14:47 UTC (permalink / raw)
To: David S . Miller; +Cc: netdev, Eric Dumazet, Eric Dumazet
Signed-off-by: Eric Dumazet <edumazet@google.com>
---
include/net/netns/ipv4.h | 1 +
include/net/tcp.h | 1 -
net/ipv4/sysctl_net_ipv4.c | 14 +++++++-------
net/ipv4/tcp_ipv4.c | 5 +++++
net/ipv4/tcp_output.c | 8 +-------
5 files changed, 14 insertions(+), 15 deletions(-)
diff --git a/include/net/netns/ipv4.h b/include/net/netns/ipv4.h
index 4458a54fe3f4e9dc79561039dba0f4d19ee8271e..60bccda046dba6c20bcc8bee097525e9455846b2 100644
--- a/include/net/netns/ipv4.h
+++ b/include/net/netns/ipv4.h
@@ -144,6 +144,7 @@ struct netns_ipv4 {
int sysctl_tcp_frto;
int sysctl_tcp_nometrics_save;
int sysctl_tcp_moderate_rcvbuf;
+ int sysctl_tcp_tso_win_divisor;
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 f954e74578fff2e52cf89521132c72fe5a047d13..ed0828dc82f1b8fd686cd68404e2abe31c6c357d 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_tso_win_divisor;
extern int sysctl_tcp_workaround_signed_windows;
#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 2ebe87fd11696948ab87349ba233a16eccb98b44..a053cacb82905bb12da60f8555a135ea11f23d28 100644
--- a/net/ipv4/sysctl_net_ipv4.c
+++ b/net/ipv4/sysctl_net_ipv4.c
@@ -451,13 +451,6 @@ static struct ctl_table ipv4_table[] = {
.mode = 0644,
.proc_handler = proc_dointvec
},
- {
- .procname = "tcp_tso_win_divisor",
- .data = &sysctl_tcp_tso_win_divisor,
- .maxlen = sizeof(int),
- .mode = 0644,
- .proc_handler = proc_dointvec,
- },
{
.procname = "tcp_congestion_control",
.mode = 0644,
@@ -1145,6 +1138,13 @@ static struct ctl_table ipv4_net_table[] = {
.mode = 0644,
.proc_handler = proc_dointvec,
},
+ {
+ .procname = "tcp_tso_win_divisor",
+ .data = &init_net.ipv4.sysctl_tcp_tso_win_divisor,
+ .maxlen = sizeof(int),
+ .mode = 0644,
+ .proc_handler = proc_dointvec,
+ },
{ }
};
diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c
index 27f376b90913519493e78e1f3a198a3a840892f1..284ff16148df908b8fdb0a0ec70e6c18513d35fc 100644
--- a/net/ipv4/tcp_ipv4.c
+++ b/net/ipv4/tcp_ipv4.c
@@ -2494,6 +2494,11 @@ static int __net_init tcp_sk_init(struct net *net)
net->ipv4.sysctl_tcp_adv_win_scale = 1;
net->ipv4.sysctl_tcp_frto = 2;
net->ipv4.sysctl_tcp_moderate_rcvbuf = 1;
+ /* This limits the percentage of the congestion window which we
+ * will allow a single TSO frame to consume. Building TSO frames
+ * which are too large can cause TCP streams to be bursty.
+ */
+ net->ipv4.sysctl_tcp_tso_win_divisor = 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 55a0aa4b96dfc7cd8f703ad42b932bae23ea5660..60df3ab52166ac20941d540ad075d60ab3426d25 100644
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -53,12 +53,6 @@ int sysctl_tcp_workaround_signed_windows __read_mostly = 0;
/* Default TSQ limit of four TSO segments */
int sysctl_tcp_limit_output_bytes __read_mostly = 262144;
-/* This limits the percentage of the congestion window which we
- * will allow a single TSO frame to consume. Building TSO frames
- * which are too large can cause TCP streams to be bursty.
- */
-int sysctl_tcp_tso_win_divisor __read_mostly = 3;
-
static bool tcp_write_xmit(struct sock *sk, unsigned int mss_now, int nonagle,
int push_one, gfp_t gfp);
@@ -1988,7 +1982,7 @@ static bool tcp_tso_should_defer(struct sock *sk, struct sk_buff *skb,
if ((skb != tcp_write_queue_tail(sk)) && (limit >= skb->len))
goto send_now;
- win_divisor = ACCESS_ONCE(sysctl_tcp_tso_win_divisor);
+ win_divisor = ACCESS_ONCE(sock_net(sk)->ipv4.sysctl_tcp_tso_win_divisor);
if (win_divisor) {
u32 chunk = min(tp->snd_wnd, tp->snd_cwnd * tp->mss_cache);
--
2.15.0.rc2.357.g7e34df9404-goog
^ permalink raw reply related [flat|nested] 14+ messages in thread
* [PATCH net-next 04/12] tcp: Namespace-ify sysctl_tcp_workaround_signed_windows
2017-10-27 14:47 [PATCH net-next 00/12] tcp: move 12 sysctls to namespaces Eric Dumazet
` (2 preceding siblings ...)
2017-10-27 14:47 ` [PATCH net-next 03/12] tcp: Namespace-ify sysctl_tcp_tso_win_divisor Eric Dumazet
@ 2017-10-27 14:47 ` Eric Dumazet
2017-10-27 14:47 ` [PATCH net-next 05/12] tcp: Namespace-ify sysctl_tcp_limit_output_bytes Eric Dumazet
` (8 subsequent siblings)
12 siblings, 0 replies; 14+ messages in thread
From: Eric Dumazet @ 2017-10-27 14:47 UTC (permalink / raw)
To: David S . Miller; +Cc: netdev, Eric Dumazet, Eric Dumazet
Signed-off-by: Eric Dumazet <edumazet@google.com>
---
include/net/netns/ipv4.h | 1 +
include/net/tcp.h | 4 ++--
net/ipv4/syncookies.c | 2 +-
net/ipv4/sysctl_net_ipv4.c | 14 +++++++-------
net/ipv4/tcp_minisocks.c | 2 +-
net/ipv4/tcp_output.c | 14 +++++---------
net/ipv6/syncookies.c | 2 +-
7 files changed, 18 insertions(+), 21 deletions(-)
diff --git a/include/net/netns/ipv4.h b/include/net/netns/ipv4.h
index 60bccda046dba6c20bcc8bee097525e9455846b2..e74c7c1b0d18e68a4b6caeeb65f6ec09148a6549 100644
--- a/include/net/netns/ipv4.h
+++ b/include/net/netns/ipv4.h
@@ -145,6 +145,7 @@ struct netns_ipv4 {
int sysctl_tcp_nometrics_save;
int sysctl_tcp_moderate_rcvbuf;
int sysctl_tcp_tso_win_divisor;
+ int sysctl_tcp_workaround_signed_windows;
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 ed0828dc82f1b8fd686cd68404e2abe31c6c357d..e338e16178dd8c46a96e81ff4d4aa1ccaaf9c937 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_workaround_signed_windows;
#define TCP_RACK_LOSS_DETECTION 0x1 /* Use RACK to detect losses */
@@ -1302,7 +1301,8 @@ static inline void tcp_slow_start_after_idle_check(struct sock *sk)
}
/* Determine a window scaling and initial window to offer. */
-void tcp_select_initial_window(int __space, __u32 mss, __u32 *rcv_wnd,
+void tcp_select_initial_window(const struct sock *sk, int __space,
+ __u32 mss, __u32 *rcv_wnd,
__u32 *window_clamp, int wscale_ok,
__u8 *rcv_wscale, __u32 init_rcv_wnd);
diff --git a/net/ipv4/syncookies.c b/net/ipv4/syncookies.c
index 77cf32a80952fcf3ceff4ada946cc2d0df2411d9..fda37f2862c923eb8b6d0b49d9442950fc6a7446 100644
--- a/net/ipv4/syncookies.c
+++ b/net/ipv4/syncookies.c
@@ -385,7 +385,7 @@ struct sock *cookie_v4_check(struct sock *sk, struct sk_buff *skb)
/* Try to redo what tcp_v4_send_synack did. */
req->rsk_window_clamp = tp->window_clamp ? :dst_metric(&rt->dst, RTAX_WINDOW);
- tcp_select_initial_window(tcp_full_space(sk), req->mss,
+ tcp_select_initial_window(sk, tcp_full_space(sk), req->mss,
&req->rsk_rcv_wnd, &req->rsk_window_clamp,
ireq->wscale_ok, &rcv_wscale,
dst_metric(&rt->dst, RTAX_INITRWND));
diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c
index a053cacb82905bb12da60f8555a135ea11f23d28..3ae9012a497997f542fa1d743ba78a6c61beaf95 100644
--- a/net/ipv4/sysctl_net_ipv4.c
+++ b/net/ipv4/sysctl_net_ipv4.c
@@ -457,13 +457,6 @@ static struct ctl_table ipv4_table[] = {
.maxlen = TCP_CA_NAME_MAX,
.proc_handler = proc_tcp_congestion_control,
},
- {
- .procname = "tcp_workaround_signed_windows",
- .data = &sysctl_tcp_workaround_signed_windows,
- .maxlen = sizeof(int),
- .mode = 0644,
- .proc_handler = proc_dointvec
- },
{
.procname = "tcp_limit_output_bytes",
.data = &sysctl_tcp_limit_output_bytes,
@@ -1145,6 +1138,13 @@ static struct ctl_table ipv4_net_table[] = {
.mode = 0644,
.proc_handler = proc_dointvec,
},
+ {
+ .procname = "tcp_workaround_signed_windows",
+ .data = &init_net.ipv4.sysctl_tcp_workaround_signed_windows,
+ .maxlen = sizeof(int),
+ .mode = 0644,
+ .proc_handler = proc_dointvec
+ },
{ }
};
diff --git a/net/ipv4/tcp_minisocks.c b/net/ipv4/tcp_minisocks.c
index 3270ab8416ce8691cbb1c3a25533142fe1029bed..3c65c1a3f94485e4099989273d7211e37243494e 100644
--- a/net/ipv4/tcp_minisocks.c
+++ b/net/ipv4/tcp_minisocks.c
@@ -369,7 +369,7 @@ void tcp_openreq_init_rwin(struct request_sock *req,
full_space = rcv_wnd * mss;
/* tcp_full_space because it is guaranteed to be the first packet */
- tcp_select_initial_window(full_space,
+ tcp_select_initial_window(sk_listener, full_space,
mss - (ireq->tstamp_ok ? TCPOLEN_TSTAMP_ALIGNED : 0),
&req->rsk_rcv_wnd,
&req->rsk_window_clamp,
diff --git a/net/ipv4/tcp_output.c b/net/ipv4/tcp_output.c
index 60df3ab52166ac20941d540ad075d60ab3426d25..5bbed67c27e9c163ede3065e57ceb0d04a4925d2 100644
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -45,11 +45,6 @@
#include <trace/events/tcp.h>
-/* People can turn this on to work with those rare, broken TCPs that
- * interpret the window field as a signed quantity.
- */
-int sysctl_tcp_workaround_signed_windows __read_mostly = 0;
-
/* Default TSQ limit of four TSO segments */
int sysctl_tcp_limit_output_bytes __read_mostly = 262144;
@@ -196,7 +191,7 @@ u32 tcp_default_init_rwnd(u32 mss)
* be a multiple of mss if possible. We assume here that mss >= 1.
* This MUST be enforced by all callers.
*/
-void tcp_select_initial_window(int __space, __u32 mss,
+void tcp_select_initial_window(const struct sock *sk, int __space, __u32 mss,
__u32 *rcv_wnd, __u32 *window_clamp,
int wscale_ok, __u8 *rcv_wscale,
__u32 init_rcv_wnd)
@@ -220,7 +215,7 @@ void tcp_select_initial_window(int __space, __u32 mss,
* which we interpret as a sign the remote TCP is not
* misinterpreting the window field as a signed quantity.
*/
- if (sysctl_tcp_workaround_signed_windows)
+ if (sock_net(sk)->ipv4.sysctl_tcp_workaround_signed_windows)
(*rcv_wnd) = min(space, MAX_TCP_WINDOW);
else
(*rcv_wnd) = space;
@@ -280,7 +275,8 @@ static u16 tcp_select_window(struct sock *sk)
/* Make sure we do not exceed the maximum possible
* scaled window.
*/
- if (!tp->rx_opt.rcv_wscale && sysctl_tcp_workaround_signed_windows)
+ if (!tp->rx_opt.rcv_wscale &&
+ sock_net(sk)->ipv4.sysctl_tcp_workaround_signed_windows)
new_win = min(new_win, MAX_TCP_WINDOW);
else
new_win = min(new_win, (65535U << tp->rx_opt.rcv_wscale));
@@ -3349,7 +3345,7 @@ static void tcp_connect_init(struct sock *sk)
if (rcv_wnd == 0)
rcv_wnd = dst_metric(dst, RTAX_INITRWND);
- tcp_select_initial_window(tcp_full_space(sk),
+ tcp_select_initial_window(sk, tcp_full_space(sk),
tp->advmss - (tp->rx_opt.ts_recent_stamp ? tp->tcp_header_len - sizeof(struct tcphdr) : 0),
&tp->rcv_wnd,
&tp->window_clamp,
diff --git a/net/ipv6/syncookies.c b/net/ipv6/syncookies.c
index 4e7817abc0b934fbff21ba481c3f6773475c7a63..e7a3a6b6cf5650f1036b875688b48f3ee2f9c967 100644
--- a/net/ipv6/syncookies.c
+++ b/net/ipv6/syncookies.c
@@ -244,7 +244,7 @@ struct sock *cookie_v6_check(struct sock *sk, struct sk_buff *skb)
}
req->rsk_window_clamp = tp->window_clamp ? :dst_metric(dst, RTAX_WINDOW);
- tcp_select_initial_window(tcp_full_space(sk), req->mss,
+ tcp_select_initial_window(sk, tcp_full_space(sk), req->mss,
&req->rsk_rcv_wnd, &req->rsk_window_clamp,
ireq->wscale_ok, &rcv_wscale,
dst_metric(dst, RTAX_INITRWND));
--
2.15.0.rc2.357.g7e34df9404-goog
^ permalink raw reply related [flat|nested] 14+ messages in thread
* [PATCH net-next 05/12] tcp: Namespace-ify sysctl_tcp_limit_output_bytes
2017-10-27 14:47 [PATCH net-next 00/12] tcp: move 12 sysctls to namespaces Eric Dumazet
` (3 preceding siblings ...)
2017-10-27 14:47 ` [PATCH net-next 04/12] tcp: Namespace-ify sysctl_tcp_workaround_signed_windows Eric Dumazet
@ 2017-10-27 14:47 ` Eric Dumazet
2017-10-27 14:47 ` [PATCH net-next 06/12] tcp: Namespace-ify sysctl_tcp_challenge_ack_limit Eric Dumazet
` (7 subsequent siblings)
12 siblings, 0 replies; 14+ messages in thread
From: Eric Dumazet @ 2017-10-27 14:47 UTC (permalink / raw)
To: David S . Miller; +Cc: netdev, Eric Dumazet, Eric Dumazet
Signed-off-by: Eric Dumazet <edumazet@google.com>
---
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 | 6 ++----
5 files changed, 12 insertions(+), 12 deletions(-)
diff --git a/include/net/netns/ipv4.h b/include/net/netns/ipv4.h
index e74c7c1b0d18e68a4b6caeeb65f6ec09148a6549..e98f473bab13cd9b8ff5cc3a62a75134dd84371a 100644
--- a/include/net/netns/ipv4.h
+++ b/include/net/netns/ipv4.h
@@ -146,6 +146,7 @@ struct netns_ipv4 {
int sysctl_tcp_moderate_rcvbuf;
int sysctl_tcp_tso_win_divisor;
int sysctl_tcp_workaround_signed_windows;
+ int sysctl_tcp_limit_output_bytes;
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 e338e16178dd8c46a96e81ff4d4aa1ccaaf9c937..33f9d30a69050e4d3eaf4e1f4869d148442902fc 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -250,7 +250,6 @@ extern int sysctl_tcp_rmem[3];
#define TCP_RACK_LOSS_DETECTION 0x1 /* Use RACK to detect losses */
-extern int sysctl_tcp_limit_output_bytes;
extern int sysctl_tcp_challenge_ack_limit;
extern int sysctl_tcp_min_tso_segs;
extern int sysctl_tcp_min_rtt_wlen;
diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c
index 3ae9012a497997f542fa1d743ba78a6c61beaf95..6caf5c40730fa10f14a35e1f3219a69f5365a2ce 100644
--- a/net/ipv4/sysctl_net_ipv4.c
+++ b/net/ipv4/sysctl_net_ipv4.c
@@ -457,13 +457,6 @@ static struct ctl_table ipv4_table[] = {
.maxlen = TCP_CA_NAME_MAX,
.proc_handler = proc_tcp_congestion_control,
},
- {
- .procname = "tcp_limit_output_bytes",
- .data = &sysctl_tcp_limit_output_bytes,
- .maxlen = sizeof(int),
- .mode = 0644,
- .proc_handler = proc_dointvec
- },
{
.procname = "tcp_challenge_ack_limit",
.data = &sysctl_tcp_challenge_ack_limit,
@@ -1145,6 +1138,13 @@ static struct ctl_table ipv4_net_table[] = {
.mode = 0644,
.proc_handler = proc_dointvec
},
+ {
+ .procname = "tcp_limit_output_bytes",
+ .data = &init_net.ipv4.sysctl_tcp_limit_output_bytes,
+ .maxlen = sizeof(int),
+ .mode = 0644,
+ .proc_handler = proc_dointvec
+ },
{ }
};
diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c
index 284ff16148df908b8fdb0a0ec70e6c18513d35fc..713b80261e4fdbafa9031a7e3bfc06f0700c2279 100644
--- a/net/ipv4/tcp_ipv4.c
+++ b/net/ipv4/tcp_ipv4.c
@@ -2499,6 +2499,8 @@ static int __net_init tcp_sk_init(struct net *net)
* which are too large can cause TCP streams to be bursty.
*/
net->ipv4.sysctl_tcp_tso_win_divisor = 3;
+ /* Default TSQ limit of four TSO segments */
+ net->ipv4.sysctl_tcp_limit_output_bytes = 262144;
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 5bbed67c27e9c163ede3065e57ceb0d04a4925d2..f018892c6a98ca7b806570995474bd394d9ab427 100644
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -45,9 +45,6 @@
#include <trace/events/tcp.h>
-/* Default TSQ limit of four TSO segments */
-int sysctl_tcp_limit_output_bytes __read_mostly = 262144;
-
static bool tcp_write_xmit(struct sock *sk, unsigned int mss_now, int nonagle,
int push_one, gfp_t gfp);
@@ -2215,7 +2212,8 @@ static bool tcp_small_queue_check(struct sock *sk, const struct sk_buff *skb,
unsigned int limit;
limit = max(2 * skb->truesize, sk->sk_pacing_rate >> 10);
- limit = min_t(u32, limit, sysctl_tcp_limit_output_bytes);
+ limit = min_t(u32, limit,
+ sock_net(sk)->ipv4.sysctl_tcp_limit_output_bytes);
limit <<= factor;
if (refcount_read(&sk->sk_wmem_alloc) > limit) {
--
2.15.0.rc2.357.g7e34df9404-goog
^ permalink raw reply related [flat|nested] 14+ messages in thread
* [PATCH net-next 06/12] tcp: Namespace-ify sysctl_tcp_challenge_ack_limit
2017-10-27 14:47 [PATCH net-next 00/12] tcp: move 12 sysctls to namespaces Eric Dumazet
` (4 preceding siblings ...)
2017-10-27 14:47 ` [PATCH net-next 05/12] tcp: Namespace-ify sysctl_tcp_limit_output_bytes Eric Dumazet
@ 2017-10-27 14:47 ` Eric Dumazet
2017-10-27 14:47 ` [PATCH net-next 07/12] tcp: Namespace-ify sysctl_tcp_min_tso_segs Eric Dumazet
` (6 subsequent siblings)
12 siblings, 0 replies; 14+ messages in thread
From: Eric Dumazet @ 2017-10-27 14:47 UTC (permalink / raw)
To: David S . Miller; +Cc: netdev, Eric Dumazet, Eric Dumazet
Signed-off-by: Eric Dumazet <edumazet@google.com>
---
include/net/netns/ipv4.h | 1 +
include/net/tcp.h | 1 -
net/ipv4/sysctl_net_ipv4.c | 14 +++++++-------
net/ipv4/tcp_input.c | 14 ++++++--------
net/ipv4/tcp_ipv4.c | 2 ++
5 files changed, 16 insertions(+), 16 deletions(-)
diff --git a/include/net/netns/ipv4.h b/include/net/netns/ipv4.h
index e98f473bab13cd9b8ff5cc3a62a75134dd84371a..e9895d40868e609dc460755b13a85cdcc82c053e 100644
--- a/include/net/netns/ipv4.h
+++ b/include/net/netns/ipv4.h
@@ -147,6 +147,7 @@ struct netns_ipv4 {
int sysctl_tcp_tso_win_divisor;
int sysctl_tcp_workaround_signed_windows;
int sysctl_tcp_limit_output_bytes;
+ int sysctl_tcp_challenge_ack_limit;
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 33f9d30a69050e4d3eaf4e1f4869d148442902fc..afc23596e9aa30fe6b4ed64049b3f55b6760be2b 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -250,7 +250,6 @@ extern int sysctl_tcp_rmem[3];
#define TCP_RACK_LOSS_DETECTION 0x1 /* Use RACK to detect losses */
-extern int sysctl_tcp_challenge_ack_limit;
extern int sysctl_tcp_min_tso_segs;
extern int sysctl_tcp_min_rtt_wlen;
extern int sysctl_tcp_autocorking;
diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c
index 6caf5c40730fa10f14a35e1f3219a69f5365a2ce..e28b3b7a7bbc75d3eb20718caef18d03bea79f95 100644
--- a/net/ipv4/sysctl_net_ipv4.c
+++ b/net/ipv4/sysctl_net_ipv4.c
@@ -457,13 +457,6 @@ static struct ctl_table ipv4_table[] = {
.maxlen = TCP_CA_NAME_MAX,
.proc_handler = proc_tcp_congestion_control,
},
- {
- .procname = "tcp_challenge_ack_limit",
- .data = &sysctl_tcp_challenge_ack_limit,
- .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_challenge_ack_limit",
+ .data = &init_net.ipv4.sysctl_tcp_challenge_ack_limit,
+ .maxlen = sizeof(int),
+ .mode = 0644,
+ .proc_handler = proc_dointvec
+ },
{ }
};
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index ce481325115fe7cb8a3b007276f2ce2a519a7383..928048a4e2c5bf5f7875adfe019af88af8244533 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -79,9 +79,6 @@
#include <linux/unaligned/access_ok.h>
#include <linux/static_key.h>
-/* rfc5961 challenge ack rate limiting */
-int sysctl_tcp_challenge_ack_limit = 1000;
-
int sysctl_tcp_max_orphans __read_mostly = NR_FILE;
int sysctl_tcp_min_rtt_wlen __read_mostly = 300;
int sysctl_tcp_invalid_ratelimit __read_mostly = HZ/2;
@@ -3443,10 +3440,11 @@ static void tcp_send_challenge_ack(struct sock *sk, const struct sk_buff *skb)
static u32 challenge_timestamp;
static unsigned int challenge_count;
struct tcp_sock *tp = tcp_sk(sk);
+ struct net *net = sock_net(sk);
u32 count, now;
/* First check our per-socket dupack rate limit. */
- if (__tcp_oow_rate_limited(sock_net(sk),
+ if (__tcp_oow_rate_limited(net,
LINUX_MIB_TCPACKSKIPPEDCHALLENGE,
&tp->last_oow_ack_time))
return;
@@ -3454,16 +3452,16 @@ static void tcp_send_challenge_ack(struct sock *sk, const struct sk_buff *skb)
/* Then check host-wide RFC 5961 rate limit. */
now = jiffies / HZ;
if (now != challenge_timestamp) {
- u32 half = (sysctl_tcp_challenge_ack_limit + 1) >> 1;
+ u32 ack_limit = net->ipv4.sysctl_tcp_challenge_ack_limit;
+ u32 half = (ack_limit + 1) >> 1;
challenge_timestamp = now;
- WRITE_ONCE(challenge_count, half +
- prandom_u32_max(sysctl_tcp_challenge_ack_limit));
+ WRITE_ONCE(challenge_count, half + prandom_u32_max(ack_limit));
}
count = READ_ONCE(challenge_count);
if (count > 0) {
WRITE_ONCE(challenge_count, count - 1);
- NET_INC_STATS(sock_net(sk), LINUX_MIB_TCPCHALLENGEACK);
+ NET_INC_STATS(net, LINUX_MIB_TCPCHALLENGEACK);
tcp_send_ack(sk);
}
}
diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c
index 713b80261e4fdbafa9031a7e3bfc06f0700c2279..50ab3a3eced342846f57a93f0ec8c7d5ae0d7e32 100644
--- a/net/ipv4/tcp_ipv4.c
+++ b/net/ipv4/tcp_ipv4.c
@@ -2501,6 +2501,8 @@ static int __net_init tcp_sk_init(struct net *net)
net->ipv4.sysctl_tcp_tso_win_divisor = 3;
/* Default TSQ limit of four TSO segments */
net->ipv4.sysctl_tcp_limit_output_bytes = 262144;
+ /* rfc5961 challenge ack rate limiting */
+ net->ipv4.sysctl_tcp_challenge_ack_limit = 1000;
net->ipv4.sysctl_tcp_fastopen = TFO_CLIENT_ENABLE;
spin_lock_init(&net->ipv4.tcp_fastopen_ctx_lock);
--
2.15.0.rc2.357.g7e34df9404-goog
^ permalink raw reply related [flat|nested] 14+ messages in thread
* [PATCH net-next 07/12] tcp: Namespace-ify sysctl_tcp_min_tso_segs
2017-10-27 14:47 [PATCH net-next 00/12] tcp: move 12 sysctls to namespaces Eric Dumazet
` (5 preceding siblings ...)
2017-10-27 14:47 ` [PATCH net-next 06/12] tcp: Namespace-ify sysctl_tcp_challenge_ack_limit Eric Dumazet
@ 2017-10-27 14:47 ` Eric Dumazet
2017-10-27 14:47 ` [PATCH net-next 08/12] tcp: Namespace-ify sysctl_tcp_min_rtt_wlen Eric Dumazet
` (5 subsequent siblings)
12 siblings, 0 replies; 14+ messages in thread
From: Eric Dumazet @ 2017-10-27 14:47 UTC (permalink / raw)
To: David S . Miller; +Cc: netdev, Eric Dumazet, Eric Dumazet
Signed-off-by: Eric Dumazet <edumazet@google.com>
---
include/net/netns/ipv4.h | 1 +
include/net/tcp.h | 1 -
net/ipv4/sysctl_net_ipv4.c | 18 +++++++++---------
net/ipv4/tcp.c | 2 --
net/ipv4/tcp_ipv4.c | 1 +
net/ipv4/tcp_output.c | 3 ++-
6 files changed, 13 insertions(+), 13 deletions(-)
diff --git a/include/net/netns/ipv4.h b/include/net/netns/ipv4.h
index e9895d40868e609dc460755b13a85cdcc82c053e..a2da3e19a977bdf4447339e6b71937eaf7e24c08 100644
--- a/include/net/netns/ipv4.h
+++ b/include/net/netns/ipv4.h
@@ -148,6 +148,7 @@ struct netns_ipv4 {
int sysctl_tcp_workaround_signed_windows;
int sysctl_tcp_limit_output_bytes;
int sysctl_tcp_challenge_ack_limit;
+ int sysctl_tcp_min_tso_segs;
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 afc23596e9aa30fe6b4ed64049b3f55b6760be2b..0735303a6575670babef06b23964ab9572aa2a07 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -250,7 +250,6 @@ extern int sysctl_tcp_rmem[3];
#define TCP_RACK_LOSS_DETECTION 0x1 /* Use RACK to detect losses */
-extern int sysctl_tcp_min_tso_segs;
extern int sysctl_tcp_min_rtt_wlen;
extern int sysctl_tcp_autocorking;
extern int sysctl_tcp_invalid_ratelimit;
diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c
index e28b3b7a7bbc75d3eb20718caef18d03bea79f95..00b4aea3705b21e552ca482eb816ac7023c13565 100644
--- a/net/ipv4/sysctl_net_ipv4.c
+++ b/net/ipv4/sysctl_net_ipv4.c
@@ -499,15 +499,6 @@ static struct ctl_table ipv4_table[] = {
.mode = 0644,
.proc_handler = proc_allowed_congestion_control,
},
- {
- .procname = "tcp_min_tso_segs",
- .data = &sysctl_tcp_min_tso_segs,
- .maxlen = sizeof(int),
- .mode = 0644,
- .proc_handler = proc_dointvec_minmax,
- .extra1 = &one,
- .extra2 = &gso_max_segs,
- },
{
.procname = "tcp_pacing_ss_ratio",
.data = &sysctl_tcp_pacing_ss_ratio,
@@ -1145,6 +1136,15 @@ static struct ctl_table ipv4_net_table[] = {
.mode = 0644,
.proc_handler = proc_dointvec
},
+ {
+ .procname = "tcp_min_tso_segs",
+ .data = &init_net.ipv4.sysctl_tcp_min_tso_segs,
+ .maxlen = sizeof(int),
+ .mode = 0644,
+ .proc_handler = proc_dointvec_minmax,
+ .extra1 = &one,
+ .extra2 = &gso_max_segs,
+ },
{ }
};
diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
index c7c983f0f817c639e68f6fb1a70916cb604de90b..a01c97708d8388cdb44577597f76837b35ea7047 100644
--- a/net/ipv4/tcp.c
+++ b/net/ipv4/tcp.c
@@ -285,8 +285,6 @@
#include <trace/events/tcp.h>
-int sysctl_tcp_min_tso_segs __read_mostly = 2;
-
int sysctl_tcp_autocorking __read_mostly = 1;
struct percpu_counter tcp_orphan_count;
diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c
index 50ab3a3eced342846f57a93f0ec8c7d5ae0d7e32..6192f26145d36d478ee5115a241c19c3a36a7642 100644
--- a/net/ipv4/tcp_ipv4.c
+++ b/net/ipv4/tcp_ipv4.c
@@ -2503,6 +2503,7 @@ static int __net_init tcp_sk_init(struct net *net)
net->ipv4.sysctl_tcp_limit_output_bytes = 262144;
/* rfc5961 challenge ack rate limiting */
net->ipv4.sysctl_tcp_challenge_ack_limit = 1000;
+ net->ipv4.sysctl_tcp_min_tso_segs = 2;
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 f018892c6a98ca7b806570995474bd394d9ab427..aab6e7145013e5a77fa989c66e8b5f6937158bbf 100644
--- a/net/ipv4/tcp_output.c
+++ b/net/ipv4/tcp_output.c
@@ -1758,7 +1758,8 @@ static u32 tcp_tso_segs(struct sock *sk, unsigned int mss_now)
u32 tso_segs = ca_ops->tso_segs_goal ? ca_ops->tso_segs_goal(sk) : 0;
return tso_segs ? :
- tcp_tso_autosize(sk, mss_now, sysctl_tcp_min_tso_segs);
+ tcp_tso_autosize(sk, mss_now,
+ sock_net(sk)->ipv4.sysctl_tcp_min_tso_segs);
}
/* Returns the portion of skb which can be sent right away */
--
2.15.0.rc2.357.g7e34df9404-goog
^ permalink raw reply related [flat|nested] 14+ messages in thread
* [PATCH net-next 08/12] tcp: Namespace-ify sysctl_tcp_min_rtt_wlen
2017-10-27 14:47 [PATCH net-next 00/12] tcp: move 12 sysctls to namespaces Eric Dumazet
` (6 preceding siblings ...)
2017-10-27 14:47 ` [PATCH net-next 07/12] tcp: Namespace-ify sysctl_tcp_min_tso_segs Eric Dumazet
@ 2017-10-27 14:47 ` Eric Dumazet
2017-10-27 14:47 ` [PATCH net-next 09/12] tcp: Namespace-ify sysctl_tcp_autocorking Eric Dumazet
` (4 subsequent siblings)
12 siblings, 0 replies; 14+ messages in thread
From: Eric Dumazet @ 2017-10-27 14:47 UTC (permalink / raw)
To: David S . Miller; +Cc: netdev, Eric Dumazet, Eric Dumazet
Signed-off-by: Eric Dumazet <edumazet@google.com>
---
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 a2da3e19a977bdf4447339e6b71937eaf7e24c08..1a66af8a0d32424cc80213001f131230d117b61d 100644
--- a/include/net/netns/ipv4.h
+++ b/include/net/netns/ipv4.h
@@ -149,6 +149,7 @@ struct netns_ipv4 {
int sysctl_tcp_limit_output_bytes;
int sysctl_tcp_challenge_ack_limit;
int sysctl_tcp_min_tso_segs;
+ int sysctl_tcp_min_rtt_wlen;
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 0735303a6575670babef06b23964ab9572aa2a07..56f50c9a3e6a8dd54d44d48bde678a144b10abbf 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -250,7 +250,6 @@ extern int sysctl_tcp_rmem[3];
#define TCP_RACK_LOSS_DETECTION 0x1 /* Use RACK to detect losses */
-extern int sysctl_tcp_min_rtt_wlen;
extern int sysctl_tcp_autocorking;
extern int sysctl_tcp_invalid_ratelimit;
extern int sysctl_tcp_pacing_ss_ratio;
diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c
index 00b4aea3705b21e552ca482eb816ac7023c13565..029692d2e4ae3a4dbf039ac6e5cb916383a833be 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_min_rtt_wlen",
- .data = &sysctl_tcp_min_rtt_wlen,
- .maxlen = sizeof(int),
- .mode = 0644,
- .proc_handler = proc_dointvec
- },
{
.procname = "tcp_low_latency",
.data = &sysctl_tcp_low_latency,
@@ -1145,6 +1138,13 @@ static struct ctl_table ipv4_net_table[] = {
.extra1 = &one,
.extra2 = &gso_max_segs,
},
+ {
+ .procname = "tcp_min_rtt_wlen",
+ .data = &init_net.ipv4.sysctl_tcp_min_rtt_wlen,
+ .maxlen = sizeof(int),
+ .mode = 0644,
+ .proc_handler = proc_dointvec
+ },
{ }
};
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 928048a4e2c5bf5f7875adfe019af88af8244533..da1ef666d1f9bb008fec3fe49852ba273802bd1a 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -80,7 +80,6 @@
#include <linux/static_key.h>
int sysctl_tcp_max_orphans __read_mostly = NR_FILE;
-int sysctl_tcp_min_rtt_wlen __read_mostly = 300;
int sysctl_tcp_invalid_ratelimit __read_mostly = HZ/2;
#define FLAG_DATA 0x01 /* Incoming frame contained data. */
@@ -2915,8 +2914,8 @@ static void tcp_fastretrans_alert(struct sock *sk, const int acked,
static void tcp_update_rtt_min(struct sock *sk, u32 rtt_us)
{
+ u32 wlen = sock_net(sk)->ipv4.sysctl_tcp_min_rtt_wlen * HZ;
struct tcp_sock *tp = tcp_sk(sk);
- u32 wlen = sysctl_tcp_min_rtt_wlen * HZ;
minmax_running_min(&tp->rtt_min, wlen, tcp_jiffies32,
rtt_us ? : jiffies_to_usecs(1));
diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c
index 6192f26145d36d478ee5115a241c19c3a36a7642..ced35af5737a041aa7aa19336056863da84f989e 100644
--- a/net/ipv4/tcp_ipv4.c
+++ b/net/ipv4/tcp_ipv4.c
@@ -2504,6 +2504,7 @@ static int __net_init tcp_sk_init(struct net *net)
/* rfc5961 challenge ack rate limiting */
net->ipv4.sysctl_tcp_challenge_ack_limit = 1000;
net->ipv4.sysctl_tcp_min_tso_segs = 2;
+ net->ipv4.sysctl_tcp_min_rtt_wlen = 300;
net->ipv4.sysctl_tcp_fastopen = TFO_CLIENT_ENABLE;
spin_lock_init(&net->ipv4.tcp_fastopen_ctx_lock);
--
2.15.0.rc2.357.g7e34df9404-goog
^ permalink raw reply related [flat|nested] 14+ messages in thread
* [PATCH net-next 09/12] tcp: Namespace-ify sysctl_tcp_autocorking
2017-10-27 14:47 [PATCH net-next 00/12] tcp: move 12 sysctls to namespaces Eric Dumazet
` (7 preceding siblings ...)
2017-10-27 14:47 ` [PATCH net-next 08/12] tcp: Namespace-ify sysctl_tcp_min_rtt_wlen Eric Dumazet
@ 2017-10-27 14:47 ` Eric Dumazet
2017-10-27 14:47 ` [PATCH net-next 10/12] tcp: Namespace-ify sysctl_tcp_invalid_ratelimit Eric Dumazet
` (3 subsequent siblings)
12 siblings, 0 replies; 14+ messages in thread
From: Eric Dumazet @ 2017-10-27 14:47 UTC (permalink / raw)
To: David S . Miller; +Cc: netdev, Eric Dumazet, Eric Dumazet
Signed-off-by: Eric Dumazet <edumazet@google.com>
---
include/net/netns/ipv4.h | 1 +
include/net/tcp.h | 1 -
net/ipv4/sysctl_net_ipv4.c | 18 +++++++++---------
net/ipv4/tcp.c | 4 +---
net/ipv4/tcp_ipv4.c | 1 +
5 files changed, 12 insertions(+), 13 deletions(-)
diff --git a/include/net/netns/ipv4.h b/include/net/netns/ipv4.h
index 1a66af8a0d32424cc80213001f131230d117b61d..537830882149c3c18c8c0b5b50b8ecded46fad09 100644
--- a/include/net/netns/ipv4.h
+++ b/include/net/netns/ipv4.h
@@ -150,6 +150,7 @@ struct netns_ipv4 {
int sysctl_tcp_challenge_ack_limit;
int sysctl_tcp_min_tso_segs;
int sysctl_tcp_min_rtt_wlen;
+ int sysctl_tcp_autocorking;
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 56f50c9a3e6a8dd54d44d48bde678a144b10abbf..0268f1025d9d5fa24046b383f4fde5fb660712fd 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -250,7 +250,6 @@ extern int sysctl_tcp_rmem[3];
#define TCP_RACK_LOSS_DETECTION 0x1 /* Use RACK to detect losses */
-extern int sysctl_tcp_autocorking;
extern int sysctl_tcp_invalid_ratelimit;
extern int sysctl_tcp_pacing_ss_ratio;
extern int sysctl_tcp_pacing_ca_ratio;
diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c
index 029692d2e4ae3a4dbf039ac6e5cb916383a833be..43a18a3170531ca9ea490fc5bb251cb9c9b8b889 100644
--- a/net/ipv4/sysctl_net_ipv4.c
+++ b/net/ipv4/sysctl_net_ipv4.c
@@ -510,15 +510,6 @@ static struct ctl_table ipv4_table[] = {
.extra1 = &zero,
.extra2 = &thousand,
},
- {
- .procname = "tcp_autocorking",
- .data = &sysctl_tcp_autocorking,
- .maxlen = sizeof(int),
- .mode = 0644,
- .proc_handler = proc_dointvec_minmax,
- .extra1 = &zero,
- .extra2 = &one,
- },
{
.procname = "tcp_invalid_ratelimit",
.data = &sysctl_tcp_invalid_ratelimit,
@@ -1145,6 +1136,15 @@ static struct ctl_table ipv4_net_table[] = {
.mode = 0644,
.proc_handler = proc_dointvec
},
+ {
+ .procname = "tcp_autocorking",
+ .data = &init_net.ipv4.sysctl_tcp_autocorking,
+ .maxlen = sizeof(int),
+ .mode = 0644,
+ .proc_handler = proc_dointvec_minmax,
+ .extra1 = &zero,
+ .extra2 = &one,
+ },
{ }
};
diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
index a01c97708d8388cdb44577597f76837b35ea7047..a7a0f316eb86f491f648187c051ea0fe7875c207 100644
--- a/net/ipv4/tcp.c
+++ b/net/ipv4/tcp.c
@@ -285,8 +285,6 @@
#include <trace/events/tcp.h>
-int sysctl_tcp_autocorking __read_mostly = 1;
-
struct percpu_counter tcp_orphan_count;
EXPORT_SYMBOL_GPL(tcp_orphan_count);
@@ -697,7 +695,7 @@ static bool tcp_should_autocork(struct sock *sk, struct sk_buff *skb,
int size_goal)
{
return skb->len < size_goal &&
- sysctl_tcp_autocorking &&
+ sock_net(sk)->ipv4.sysctl_tcp_autocorking &&
skb != tcp_write_queue_head(sk) &&
refcount_read(&sk->sk_wmem_alloc) > skb->truesize;
}
diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c
index ced35af5737a041aa7aa19336056863da84f989e..351e3497c8f3ae565114c80465f149c7341b8272 100644
--- a/net/ipv4/tcp_ipv4.c
+++ b/net/ipv4/tcp_ipv4.c
@@ -2505,6 +2505,7 @@ static int __net_init tcp_sk_init(struct net *net)
net->ipv4.sysctl_tcp_challenge_ack_limit = 1000;
net->ipv4.sysctl_tcp_min_tso_segs = 2;
net->ipv4.sysctl_tcp_min_rtt_wlen = 300;
+ net->ipv4.sysctl_tcp_autocorking = 1;
net->ipv4.sysctl_tcp_fastopen = TFO_CLIENT_ENABLE;
spin_lock_init(&net->ipv4.tcp_fastopen_ctx_lock);
--
2.15.0.rc2.357.g7e34df9404-goog
^ permalink raw reply related [flat|nested] 14+ messages in thread
* [PATCH net-next 10/12] tcp: Namespace-ify sysctl_tcp_invalid_ratelimit
2017-10-27 14:47 [PATCH net-next 00/12] tcp: move 12 sysctls to namespaces Eric Dumazet
` (8 preceding siblings ...)
2017-10-27 14:47 ` [PATCH net-next 09/12] tcp: Namespace-ify sysctl_tcp_autocorking Eric Dumazet
@ 2017-10-27 14:47 ` Eric Dumazet
2017-10-27 14:47 ` [PATCH net-next 11/12] tcp: Namespace-ify sysctl_tcp_pacing_ss_ratio Eric Dumazet
` (2 subsequent siblings)
12 siblings, 0 replies; 14+ messages in thread
From: Eric Dumazet @ 2017-10-27 14:47 UTC (permalink / raw)
To: David S . Miller; +Cc: netdev, Eric Dumazet, Eric Dumazet
Signed-off-by: Eric Dumazet <edumazet@google.com>
---
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 537830882149c3c18c8c0b5b50b8ecded46fad09..e52c2124b32e799f32943621cc0cc55dc7d8413d 100644
--- a/include/net/netns/ipv4.h
+++ b/include/net/netns/ipv4.h
@@ -151,6 +151,7 @@ struct netns_ipv4 {
int sysctl_tcp_min_tso_segs;
int sysctl_tcp_min_rtt_wlen;
int sysctl_tcp_autocorking;
+ int sysctl_tcp_invalid_ratelimit;
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 0268f1025d9d5fa24046b383f4fde5fb660712fd..5869a822ecb1b8f350cb393b4b43343ccf206eb5 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -250,7 +250,6 @@ extern int sysctl_tcp_rmem[3];
#define TCP_RACK_LOSS_DETECTION 0x1 /* Use RACK to detect losses */
-extern int sysctl_tcp_invalid_ratelimit;
extern int sysctl_tcp_pacing_ss_ratio;
extern int sysctl_tcp_pacing_ca_ratio;
diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c
index 43a18a3170531ca9ea490fc5bb251cb9c9b8b889..6a9349c27f00b1a665b969fe0291d195ec473e3e 100644
--- a/net/ipv4/sysctl_net_ipv4.c
+++ b/net/ipv4/sysctl_net_ipv4.c
@@ -510,13 +510,6 @@ static struct ctl_table ipv4_table[] = {
.extra1 = &zero,
.extra2 = &thousand,
},
- {
- .procname = "tcp_invalid_ratelimit",
- .data = &sysctl_tcp_invalid_ratelimit,
- .maxlen = sizeof(int),
- .mode = 0644,
- .proc_handler = proc_dointvec_ms_jiffies,
- },
{
.procname = "tcp_available_ulp",
.maxlen = TCP_ULP_BUF_MAX,
@@ -1145,6 +1138,13 @@ static struct ctl_table ipv4_net_table[] = {
.extra1 = &zero,
.extra2 = &one,
},
+ {
+ .procname = "tcp_invalid_ratelimit",
+ .data = &init_net.ipv4.sysctl_tcp_invalid_ratelimit,
+ .maxlen = sizeof(int),
+ .mode = 0644,
+ .proc_handler = proc_dointvec_ms_jiffies,
+ },
{ }
};
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index da1ef666d1f9bb008fec3fe49852ba273802bd1a..db4d458d0205027681f96498019a5fd0bb0303b8 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -80,7 +80,6 @@
#include <linux/static_key.h>
int sysctl_tcp_max_orphans __read_mostly = NR_FILE;
-int sysctl_tcp_invalid_ratelimit __read_mostly = HZ/2;
#define FLAG_DATA 0x01 /* Incoming frame contained data. */
#define FLAG_WIN_UPDATE 0x02 /* Incoming ACK was a window update. */
@@ -3403,7 +3402,7 @@ static bool __tcp_oow_rate_limited(struct net *net, int mib_idx,
if (*last_oow_ack_time) {
s32 elapsed = (s32)(tcp_jiffies32 - *last_oow_ack_time);
- if (0 <= elapsed && elapsed < sysctl_tcp_invalid_ratelimit) {
+ if (0 <= elapsed && elapsed < net->ipv4.sysctl_tcp_invalid_ratelimit) {
NET_INC_STATS(net, mib_idx);
return true; /* rate-limited: don't send yet! */
}
diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c
index 351e3497c8f3ae565114c80465f149c7341b8272..6617aae18ba2e2ad23663708a39311b0e987522d 100644
--- a/net/ipv4/tcp_ipv4.c
+++ b/net/ipv4/tcp_ipv4.c
@@ -2506,6 +2506,7 @@ static int __net_init tcp_sk_init(struct net *net)
net->ipv4.sysctl_tcp_min_tso_segs = 2;
net->ipv4.sysctl_tcp_min_rtt_wlen = 300;
net->ipv4.sysctl_tcp_autocorking = 1;
+ net->ipv4.sysctl_tcp_invalid_ratelimit = HZ/2;
net->ipv4.sysctl_tcp_fastopen = TFO_CLIENT_ENABLE;
spin_lock_init(&net->ipv4.tcp_fastopen_ctx_lock);
--
2.15.0.rc2.357.g7e34df9404-goog
^ permalink raw reply related [flat|nested] 14+ messages in thread
* [PATCH net-next 11/12] tcp: Namespace-ify sysctl_tcp_pacing_ss_ratio
2017-10-27 14:47 [PATCH net-next 00/12] tcp: move 12 sysctls to namespaces Eric Dumazet
` (9 preceding siblings ...)
2017-10-27 14:47 ` [PATCH net-next 10/12] tcp: Namespace-ify sysctl_tcp_invalid_ratelimit Eric Dumazet
@ 2017-10-27 14:47 ` Eric Dumazet
2017-10-27 14:47 ` [PATCH net-next 12/12] tcp: Namespace-ify sysctl_tcp_pacing_ca_ratio Eric Dumazet
2017-10-28 10:25 ` [PATCH net-next 00/12] tcp: move 12 sysctls to namespaces David Miller
12 siblings, 0 replies; 14+ messages in thread
From: Eric Dumazet @ 2017-10-27 14:47 UTC (permalink / raw)
To: David S . Miller; +Cc: netdev, Eric Dumazet, Eric Dumazet
Also remove an obsolete comment about TCP pacing.
Signed-off-by: Eric Dumazet <edumazet@google.com>
---
include/net/netns/ipv4.h | 1 +
include/net/tcp.h | 1 -
net/ipv4/sysctl_net_ipv4.c | 18 +++++++++---------
net/ipv4/tcp_input.c | 9 +--------
net/ipv4/tcp_ipv4.c | 1 +
5 files changed, 12 insertions(+), 18 deletions(-)
diff --git a/include/net/netns/ipv4.h b/include/net/netns/ipv4.h
index e52c2124b32e799f32943621cc0cc55dc7d8413d..eb2dcf1cbe61bf392e1c186e6fd1a11b82d8bad2 100644
--- a/include/net/netns/ipv4.h
+++ b/include/net/netns/ipv4.h
@@ -152,6 +152,7 @@ struct netns_ipv4 {
int sysctl_tcp_min_rtt_wlen;
int sysctl_tcp_autocorking;
int sysctl_tcp_invalid_ratelimit;
+ int sysctl_tcp_pacing_ss_ratio;
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 5869a822ecb1b8f350cb393b4b43343ccf206eb5..2a5f8261ca03b76061dfbe85bbcf0fa15e5f8b40 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -250,7 +250,6 @@ extern int sysctl_tcp_rmem[3];
#define TCP_RACK_LOSS_DETECTION 0x1 /* Use RACK to detect losses */
-extern int sysctl_tcp_pacing_ss_ratio;
extern int sysctl_tcp_pacing_ca_ratio;
extern atomic_long_t tcp_memory_allocated;
diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c
index 6a9349c27f00b1a665b969fe0291d195ec473e3e..7f0dba852d471a6e78d508d0b87966e26d4ca621 100644
--- a/net/ipv4/sysctl_net_ipv4.c
+++ b/net/ipv4/sysctl_net_ipv4.c
@@ -492,15 +492,6 @@ static struct ctl_table ipv4_table[] = {
.mode = 0644,
.proc_handler = proc_allowed_congestion_control,
},
- {
- .procname = "tcp_pacing_ss_ratio",
- .data = &sysctl_tcp_pacing_ss_ratio,
- .maxlen = sizeof(int),
- .mode = 0644,
- .proc_handler = proc_dointvec_minmax,
- .extra1 = &zero,
- .extra2 = &thousand,
- },
{
.procname = "tcp_pacing_ca_ratio",
.data = &sysctl_tcp_pacing_ca_ratio,
@@ -1145,6 +1136,15 @@ static struct ctl_table ipv4_net_table[] = {
.mode = 0644,
.proc_handler = proc_dointvec_ms_jiffies,
},
+ {
+ .procname = "tcp_pacing_ss_ratio",
+ .data = &init_net.ipv4.sysctl_tcp_pacing_ss_ratio,
+ .maxlen = sizeof(int),
+ .mode = 0644,
+ .proc_handler = proc_dointvec_minmax,
+ .extra1 = &zero,
+ .extra2 = &thousand,
+ },
{ }
};
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index db4d458d0205027681f96498019a5fd0bb0303b8..29539d39e61a0b99267e6c9b8f63b8acf6ae6200 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -767,13 +767,6 @@ static void tcp_rtt_estimator(struct sock *sk, long mrtt_us)
tp->srtt_us = max(1U, srtt);
}
-/* Set the sk_pacing_rate to allow proper sizing of TSO packets.
- * Note: TCP stack does not yet implement pacing.
- * FQ packet scheduler can be used to implement cheap but effective
- * TCP pacing, to smooth the burst on large writes when packets
- * in flight is significantly lower than cwnd (or rwin)
- */
-int sysctl_tcp_pacing_ss_ratio __read_mostly = 200;
int sysctl_tcp_pacing_ca_ratio __read_mostly = 120;
static void tcp_update_pacing_rate(struct sock *sk)
@@ -793,7 +786,7 @@ static void tcp_update_pacing_rate(struct sock *sk)
* end of slow start and should slow down.
*/
if (tp->snd_cwnd < tp->snd_ssthresh / 2)
- rate *= sysctl_tcp_pacing_ss_ratio;
+ rate *= sock_net(sk)->ipv4.sysctl_tcp_pacing_ss_ratio;
else
rate *= sysctl_tcp_pacing_ca_ratio;
diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c
index 6617aae18ba2e2ad23663708a39311b0e987522d..1d8fc663af51f4d248de73ee30e28db44ffc940a 100644
--- a/net/ipv4/tcp_ipv4.c
+++ b/net/ipv4/tcp_ipv4.c
@@ -2507,6 +2507,7 @@ static int __net_init tcp_sk_init(struct net *net)
net->ipv4.sysctl_tcp_min_rtt_wlen = 300;
net->ipv4.sysctl_tcp_autocorking = 1;
net->ipv4.sysctl_tcp_invalid_ratelimit = HZ/2;
+ net->ipv4.sysctl_tcp_pacing_ss_ratio = 200;
net->ipv4.sysctl_tcp_fastopen = TFO_CLIENT_ENABLE;
spin_lock_init(&net->ipv4.tcp_fastopen_ctx_lock);
--
2.15.0.rc2.357.g7e34df9404-goog
^ permalink raw reply related [flat|nested] 14+ messages in thread
* [PATCH net-next 12/12] tcp: Namespace-ify sysctl_tcp_pacing_ca_ratio
2017-10-27 14:47 [PATCH net-next 00/12] tcp: move 12 sysctls to namespaces Eric Dumazet
` (10 preceding siblings ...)
2017-10-27 14:47 ` [PATCH net-next 11/12] tcp: Namespace-ify sysctl_tcp_pacing_ss_ratio Eric Dumazet
@ 2017-10-27 14:47 ` Eric Dumazet
2017-10-28 10:25 ` [PATCH net-next 00/12] tcp: move 12 sysctls to namespaces David Miller
12 siblings, 0 replies; 14+ messages in thread
From: Eric Dumazet @ 2017-10-27 14:47 UTC (permalink / raw)
To: David S . Miller; +Cc: netdev, Eric Dumazet, Eric Dumazet
Signed-off-by: Eric Dumazet <edumazet@google.com>
---
include/net/netns/ipv4.h | 1 +
include/net/tcp.h | 2 --
net/ipv4/sysctl_net_ipv4.c | 18 +++++++++---------
net/ipv4/tcp_input.c | 4 +---
net/ipv4/tcp_ipv4.c | 1 +
5 files changed, 12 insertions(+), 14 deletions(-)
diff --git a/include/net/netns/ipv4.h b/include/net/netns/ipv4.h
index eb2dcf1cbe61bf392e1c186e6fd1a11b82d8bad2..141ba82b5efbda4df09ea022679096f97c36df93 100644
--- a/include/net/netns/ipv4.h
+++ b/include/net/netns/ipv4.h
@@ -153,6 +153,7 @@ struct netns_ipv4 {
int sysctl_tcp_autocorking;
int sysctl_tcp_invalid_ratelimit;
int sysctl_tcp_pacing_ss_ratio;
+ int sysctl_tcp_pacing_ca_ratio;
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 2a5f8261ca03b76061dfbe85bbcf0fa15e5f8b40..092d606fcc1658d9e0b5dd451022d1b5ad1f20c1 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -250,8 +250,6 @@ extern int sysctl_tcp_rmem[3];
#define TCP_RACK_LOSS_DETECTION 0x1 /* Use RACK to detect losses */
-extern int sysctl_tcp_pacing_ca_ratio;
-
extern atomic_long_t tcp_memory_allocated;
extern struct percpu_counter tcp_sockets_allocated;
extern unsigned long tcp_memory_pressure;
diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c
index 7f0dba852d471a6e78d508d0b87966e26d4ca621..4602af6d5358d414df68a0e74ad2dae7fed9bfb6 100644
--- a/net/ipv4/sysctl_net_ipv4.c
+++ b/net/ipv4/sysctl_net_ipv4.c
@@ -492,15 +492,6 @@ static struct ctl_table ipv4_table[] = {
.mode = 0644,
.proc_handler = proc_allowed_congestion_control,
},
- {
- .procname = "tcp_pacing_ca_ratio",
- .data = &sysctl_tcp_pacing_ca_ratio,
- .maxlen = sizeof(int),
- .mode = 0644,
- .proc_handler = proc_dointvec_minmax,
- .extra1 = &zero,
- .extra2 = &thousand,
- },
{
.procname = "tcp_available_ulp",
.maxlen = TCP_ULP_BUF_MAX,
@@ -1145,6 +1136,15 @@ static struct ctl_table ipv4_net_table[] = {
.extra1 = &zero,
.extra2 = &thousand,
},
+ {
+ .procname = "tcp_pacing_ca_ratio",
+ .data = &init_net.ipv4.sysctl_tcp_pacing_ca_ratio,
+ .maxlen = sizeof(int),
+ .mode = 0644,
+ .proc_handler = proc_dointvec_minmax,
+ .extra1 = &zero,
+ .extra2 = &thousand,
+ },
{ }
};
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 29539d39e61a0b99267e6c9b8f63b8acf6ae6200..21c358c0cf2ea6c4a7bcf22b0490e043a57dc680 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -767,8 +767,6 @@ static void tcp_rtt_estimator(struct sock *sk, long mrtt_us)
tp->srtt_us = max(1U, srtt);
}
-int sysctl_tcp_pacing_ca_ratio __read_mostly = 120;
-
static void tcp_update_pacing_rate(struct sock *sk)
{
const struct tcp_sock *tp = tcp_sk(sk);
@@ -788,7 +786,7 @@ static void tcp_update_pacing_rate(struct sock *sk)
if (tp->snd_cwnd < tp->snd_ssthresh / 2)
rate *= sock_net(sk)->ipv4.sysctl_tcp_pacing_ss_ratio;
else
- rate *= sysctl_tcp_pacing_ca_ratio;
+ rate *= sock_net(sk)->ipv4.sysctl_tcp_pacing_ca_ratio;
rate *= max(tp->snd_cwnd, tp->packets_out);
diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c
index 1d8fc663af51f4d248de73ee30e28db44ffc940a..7c1dae6493c3fa36f34f38b7d5fc070d628ddee9 100644
--- a/net/ipv4/tcp_ipv4.c
+++ b/net/ipv4/tcp_ipv4.c
@@ -2508,6 +2508,7 @@ static int __net_init tcp_sk_init(struct net *net)
net->ipv4.sysctl_tcp_autocorking = 1;
net->ipv4.sysctl_tcp_invalid_ratelimit = HZ/2;
net->ipv4.sysctl_tcp_pacing_ss_ratio = 200;
+ net->ipv4.sysctl_tcp_pacing_ca_ratio = 120;
net->ipv4.sysctl_tcp_fastopen = TFO_CLIENT_ENABLE;
spin_lock_init(&net->ipv4.tcp_fastopen_ctx_lock);
--
2.15.0.rc2.357.g7e34df9404-goog
^ permalink raw reply related [flat|nested] 14+ messages in thread
* Re: [PATCH net-next 00/12] tcp: move 12 sysctls to namespaces
2017-10-27 14:47 [PATCH net-next 00/12] tcp: move 12 sysctls to namespaces Eric Dumazet
` (11 preceding siblings ...)
2017-10-27 14:47 ` [PATCH net-next 12/12] tcp: Namespace-ify sysctl_tcp_pacing_ca_ratio Eric Dumazet
@ 2017-10-28 10:25 ` David Miller
12 siblings, 0 replies; 14+ messages in thread
From: David Miller @ 2017-10-28 10:25 UTC (permalink / raw)
To: edumazet; +Cc: netdev, eric.dumazet
From: Eric Dumazet <edumazet@google.com>
Date: Fri, 27 Oct 2017 07:47:20 -0700
> Ideally all TCP sysctls should be per netns.
> This patch series takes care of 12 sysctls.
Series applied, thanks Eric.
> Remains the ones that need discussion :
>
> sysctl_tcp_mem, sysctl_tcp_rmem, sysctl_tcp_wmem, and sysctl_tcp_max_orphans
Yeah those will be tricky.
^ permalink raw reply [flat|nested] 14+ messages in thread
end of thread, other threads:[~2017-10-28 10:25 UTC | newest]
Thread overview: 14+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2017-10-27 14:47 [PATCH net-next 00/12] tcp: move 12 sysctls to namespaces Eric Dumazet
2017-10-27 14:47 ` [PATCH net-next 01/12] tcp: Namespace-ify sysctl_tcp_nometrics_save Eric Dumazet
2017-10-27 14:47 ` [PATCH net-next 02/12] tcp: Namespace-ify sysctl_tcp_moderate_rcvbuf Eric Dumazet
2017-10-27 14:47 ` [PATCH net-next 03/12] tcp: Namespace-ify sysctl_tcp_tso_win_divisor Eric Dumazet
2017-10-27 14:47 ` [PATCH net-next 04/12] tcp: Namespace-ify sysctl_tcp_workaround_signed_windows Eric Dumazet
2017-10-27 14:47 ` [PATCH net-next 05/12] tcp: Namespace-ify sysctl_tcp_limit_output_bytes Eric Dumazet
2017-10-27 14:47 ` [PATCH net-next 06/12] tcp: Namespace-ify sysctl_tcp_challenge_ack_limit Eric Dumazet
2017-10-27 14:47 ` [PATCH net-next 07/12] tcp: Namespace-ify sysctl_tcp_min_tso_segs Eric Dumazet
2017-10-27 14:47 ` [PATCH net-next 08/12] tcp: Namespace-ify sysctl_tcp_min_rtt_wlen Eric Dumazet
2017-10-27 14:47 ` [PATCH net-next 09/12] tcp: Namespace-ify sysctl_tcp_autocorking Eric Dumazet
2017-10-27 14:47 ` [PATCH net-next 10/12] tcp: Namespace-ify sysctl_tcp_invalid_ratelimit Eric Dumazet
2017-10-27 14:47 ` [PATCH net-next 11/12] tcp: Namespace-ify sysctl_tcp_pacing_ss_ratio Eric Dumazet
2017-10-27 14:47 ` [PATCH net-next 12/12] tcp: Namespace-ify sysctl_tcp_pacing_ca_ratio Eric Dumazet
2017-10-28 10:25 ` [PATCH net-next 00/12] tcp: move 12 sysctls to namespaces 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).