netdev - [PATCH net-next] tcp: introduce rto_max

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [thread-next>] [day] [month] [year] [list]

Message-Id: <20240715033118.32322-1-kerneljasonxing@gmail.com>
Date: Mon, 15 Jul 2024 11:31:18 +0800
From: Jason Xing <kerneljasonxing@...il.com>
To: edumazet@...gle.com,
	kuba@...nel.org,
	pabeni@...hat.com,
	davem@...emloft.net,
	dsahern@...nel.org,
	ncardwell@...gle.com,
	corbet@....net
Cc: netdev@...r.kernel.org,
	kerneljasonxing@...il.com,
	Jason Xing <kernelxing@...cent.com>
Subject: [PATCH net-next] tcp: introduce rto_max_us sysctl knob

From: Jason Xing <kernelxing@...cent.com>

As we all know, the algorithm of rto is exponential backoff as RFC
defined long time ago. After several rounds of repeatedly transmitting
a lost skb, the expiry of rto timer could reach above 1 second within
the upper bound (120s).

Waiting more than one second to retransmit for some latency-sensitive
application is a little bit unacceptable nowadays, so I decided to
introduce a sysctl knob to allow users to tune it. Still, the maximum
value is 120 seconds.

Signed-off-by: Jason Xing <kernelxing@...cent.com>
---
 Documentation/networking/ip-sysctl.rst | 10 ++++++++++
 include/net/inet_connection_sock.h     | 10 +++++++++-
 include/net/netns/ipv4.h               |  1 +
 net/ipv4/sysctl_net_ipv4.c             |  8 ++++++++
 net/ipv4/tcp_ipv4.c                    |  1 +
 5 files changed, 29 insertions(+), 1 deletion(-)

diff --git a/Documentation/networking/ip-sysctl.rst b/Documentation/networking/ip-sysctl.rst
index 3616389c8c2d..32a1907ca95d 100644
--- a/Documentation/networking/ip-sysctl.rst
+++ b/Documentation/networking/ip-sysctl.rst
@@ -1223,6 +1223,16 @@ tcp_rto_min_us - INTEGER
 
 	Default: 200000
 
+tcp_rto_max_us - INTEGER
+	Maximum TCP retransmission timeout (in microseconds).
+
+	The recommended practice is to use a value less or equal to 120000000
+	microseconds.
+
+	Possible Values: 1 - INT_MAX
+
+	Default: 120000000
+
 UDP variables
 =============
 
diff --git a/include/net/inet_connection_sock.h b/include/net/inet_connection_sock.h
index c0deaafebfdc..a0abbafcab9e 100644
--- a/include/net/inet_connection_sock.h
+++ b/include/net/inet_connection_sock.h
@@ -217,10 +217,18 @@ static inline void inet_csk_clear_xmit_timer(struct sock *sk, const int what)
  */
 static inline void inet_csk_reset_xmit_timer(struct sock *sk, const int what,
 					     unsigned long when,
-					     const unsigned long max_when)
+					     unsigned long max_when)
 {
 	struct inet_connection_sock *icsk = inet_csk(sk);
 
+	if (what == ICSK_TIME_RETRANS) {
+		int rto_max_us = READ_ONCE(sock_net(sk)->ipv4.sysctl_tcp_rto_max_us);
+		unsigned int rto_max = usecs_to_jiffies(rto_max_us);
+
+		if (rto_max < max_when)
+			max_when = rto_max;
+	}
+
 	if (when > max_when) {
 		pr_debug("reset_xmit_timer: sk=%p %d when=0x%lx, caller=%p\n",
 			 sk, what, when, (void *)_THIS_IP_);
diff --git a/include/net/netns/ipv4.h b/include/net/netns/ipv4.h
index 5fcd61ada622..09a28a5c94d2 100644
--- a/include/net/netns/ipv4.h
+++ b/include/net/netns/ipv4.h
@@ -178,6 +178,7 @@ struct netns_ipv4 {
 	u8 sysctl_tcp_window_scaling;
 	u8 sysctl_tcp_timestamps;
 	int sysctl_tcp_rto_min_us;
+	int sysctl_tcp_rto_max_us;
 	u8 sysctl_tcp_recovery;
 	u8 sysctl_tcp_thin_linear_timeouts;
 	u8 sysctl_tcp_slow_start_after_idle;
diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c
index 9140d20eb2d4..304f173837bc 100644
--- a/net/ipv4/sysctl_net_ipv4.c
+++ b/net/ipv4/sysctl_net_ipv4.c
@@ -1573,6 +1573,14 @@ static struct ctl_table ipv4_net_table[] = {
 		.proc_handler	= proc_dointvec_minmax,
 		.extra1		= SYSCTL_ONE,
 	},
+	{
+		.procname	= "tcp_rto_max_us",
+		.data		= &init_net.ipv4.sysctl_tcp_rto_max_us,
+		.maxlen		= sizeof(int),
+		.mode		= 0644,
+		.proc_handler	= proc_dointvec_minmax,
+		.extra1		= SYSCTL_ONE,
+	},
 };
 
 static __net_init int ipv4_sysctl_init_net(struct net *net)
diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c
index fd17f25ff288..f06859be5942 100644
--- a/net/ipv4/tcp_ipv4.c
+++ b/net/ipv4/tcp_ipv4.c
@@ -3506,6 +3506,7 @@ static int __net_init tcp_sk_init(struct net *net)
 
 	net->ipv4.sysctl_tcp_pingpong_thresh = 1;
 	net->ipv4.sysctl_tcp_rto_min_us = jiffies_to_usecs(TCP_RTO_MIN);
+	net->ipv4.sysctl_tcp_rto_max_us = jiffies_to_usecs(TCP_RTO_MAX);
 
 	return 0;
 }
-- 
2.37.3