[PATCH 1/2] ipv4: Namespaceify tcp_tw_recycle and tcp_max_tw_buckets knob

From: Haishuang Yan
Date: Wed Dec 28 2016 - 04:53:49 EST


Different namespace application might require fast recycling
TIME-WAIT sockets independently of the host.

Signed-off-by: Haishuang Yan <yanhaishuang@xxxxxxxxxxxxxxxxxxxx>
---
include/net/inet_timewait_sock.h | 13 +------------
include/net/netns/ipv4.h | 11 +++++++++++
include/net/tcp.h | 1 -
net/ipv4/af_inet.c | 2 --
net/ipv4/inet_timewait_sock.c | 3 +--
net/ipv4/proc.c | 2 +-
net/ipv4/sysctl_net_ipv4.c | 28 ++++++++++++++--------------
net/ipv4/tcp.c | 3 ++-
net/ipv4/tcp_input.c | 2 +-
net/ipv4/tcp_ipv4.c | 12 ++++++++----
net/ipv4/tcp_minisocks.c | 14 +++++---------
net/ipv6/tcp_ipv6.c | 7 ++++---
12 files changed, 48 insertions(+), 50 deletions(-)

diff --git a/include/net/inet_timewait_sock.h b/include/net/inet_timewait_sock.h
index c9b3eb7..6a75d67 100644
--- a/include/net/inet_timewait_sock.h
+++ b/include/net/inet_timewait_sock.h
@@ -29,16 +29,6 @@

#include <linux/atomic.h>

-struct inet_hashinfo;
-
-struct inet_timewait_death_row {
- atomic_t tw_count;
-
- struct inet_hashinfo *hashinfo ____cacheline_aligned_in_smp;
- int sysctl_tw_recycle;
- int sysctl_max_tw_buckets;
-};
-
struct inet_bind_bucket;

/*
@@ -125,8 +115,7 @@ static inline void inet_twsk_reschedule(struct inet_timewait_sock *tw, int timeo

void inet_twsk_deschedule_put(struct inet_timewait_sock *tw);

-void inet_twsk_purge(struct inet_hashinfo *hashinfo,
- struct inet_timewait_death_row *twdr, int family);
+void inet_twsk_purge(struct inet_hashinfo *hashinfo, int family);

static inline
struct net *twsk_net(const struct inet_timewait_sock *twsk)
diff --git a/include/net/netns/ipv4.h b/include/net/netns/ipv4.h
index 0378e88..99becaf 100644
--- a/include/net/netns/ipv4.h
+++ b/include/net/netns/ipv4.h
@@ -27,6 +27,16 @@ struct ping_group_range {
kgid_t range[2];
};

+struct inet_hashinfo;
+
+struct inet_timewait_death_row {
+ atomic_t tw_count;
+
+ struct inet_hashinfo *hashinfo ____cacheline_aligned_in_smp;
+ int sysctl_tw_recycle;
+ int sysctl_max_tw_buckets;
+};
+
struct netns_ipv4 {
#ifdef CONFIG_SYSCTL
struct ctl_table_header *forw_hdr;
@@ -111,6 +121,7 @@ struct netns_ipv4 {
int sysctl_tcp_fin_timeout;
unsigned int sysctl_tcp_notsent_lowat;
int sysctl_tcp_tw_reuse;
+ struct inet_timewait_death_row tcp_death_row;

int sysctl_igmp_max_memberships;
int sysctl_igmp_max_msf;
diff --git a/include/net/tcp.h b/include/net/tcp.h
index 6061963..1da0aa7 100644
--- a/include/net/tcp.h
+++ b/include/net/tcp.h
@@ -231,7 +231,6 @@ void tcp_time_wait(struct sock *sk, int state, int timeo);
*/
#define TFO_SERVER_WO_SOCKOPT1 0x400

-extern struct inet_timewait_death_row tcp_death_row;

/* sysctl variables for tcp */
extern int sysctl_tcp_timestamps;
diff --git a/net/ipv4/af_inet.c b/net/ipv4/af_inet.c
index 1830e6f..29b1dd9 100644
--- a/net/ipv4/af_inet.c
+++ b/net/ipv4/af_inet.c
@@ -1831,8 +1831,6 @@ static int __init inet_init(void)

ip_init();

- tcp_v4_init();
-
/* Setup TCP slab cache for open requests. */
tcp_init();

diff --git a/net/ipv4/inet_timewait_sock.c b/net/ipv4/inet_timewait_sock.c
index ddcd56c..f8aff2c 100644
--- a/net/ipv4/inet_timewait_sock.c
+++ b/net/ipv4/inet_timewait_sock.c
@@ -257,8 +257,7 @@ void __inet_twsk_schedule(struct inet_timewait_sock *tw, int timeo, bool rearm)
}
EXPORT_SYMBOL_GPL(__inet_twsk_schedule);

-void inet_twsk_purge(struct inet_hashinfo *hashinfo,
- struct inet_timewait_death_row *twdr, int family)
+void inet_twsk_purge(struct inet_hashinfo *hashinfo, int family)
{
struct inet_timewait_sock *tw;
struct sock *sk;
diff --git a/net/ipv4/proc.c b/net/ipv4/proc.c
index 7143ca1..0247ca0 100644
--- a/net/ipv4/proc.c
+++ b/net/ipv4/proc.c
@@ -65,7 +65,7 @@ static int sockstat_seq_show(struct seq_file *seq, void *v)
socket_seq_show(seq);
seq_printf(seq, "TCP: inuse %d orphan %d tw %d alloc %d mem %ld\n",
sock_prot_inuse_get(net, &tcp_prot), orphans,
- atomic_read(&tcp_death_row.tw_count), sockets,
+ atomic_read(&net->ipv4.tcp_death_row.tw_count), sockets,
proto_memory_allocated(&tcp_prot));
seq_printf(seq, "UDP: inuse %d mem %ld\n",
sock_prot_inuse_get(net, &udp_prot),
diff --git a/net/ipv4/sysctl_net_ipv4.c b/net/ipv4/sysctl_net_ipv4.c
index 22cbd61..66f8f1b 100644
--- a/net/ipv4/sysctl_net_ipv4.c
+++ b/net/ipv4/sysctl_net_ipv4.c
@@ -290,13 +290,6 @@ static struct ctl_table ipv4_table[] = {
.proc_handler = proc_dointvec
},
{
- .procname = "tcp_max_tw_buckets",
- .data = &tcp_death_row.sysctl_max_tw_buckets,
- .maxlen = sizeof(int),
- .mode = 0644,
- .proc_handler = proc_dointvec
- },
- {
.procname = "tcp_fastopen",
.data = &sysctl_tcp_fastopen,
.maxlen = sizeof(int),
@@ -310,13 +303,6 @@ static struct ctl_table ipv4_table[] = {
.proc_handler = proc_tcp_fastopen_key,
},
{
- .procname = "tcp_tw_recycle",
- .data = &tcp_death_row.sysctl_tw_recycle,
- .maxlen = sizeof(int),
- .mode = 0644,
- .proc_handler = proc_dointvec
- },
- {
.procname = "tcp_abort_on_overflow",
.data = &sysctl_tcp_abort_on_overflow,
.maxlen = sizeof(int),
@@ -960,6 +946,20 @@ static struct ctl_table ipv4_net_table[] = {
.mode = 0644,
.proc_handler = proc_dointvec
},
+ {
+ .procname = "tcp_max_tw_buckets",
+ .data = &init_net.ipv4.tcp_death_row.sysctl_max_tw_buckets,
+ .maxlen = sizeof(int),
+ .mode = 0644,
+ .proc_handler = proc_dointvec
+ },
+ {
+ .procname = "tcp_tw_recycle",
+ .data = &init_net.ipv4.tcp_death_row.sysctl_tw_recycle,
+ .maxlen = sizeof(int),
+ .mode = 0644,
+ .proc_handler = proc_dointvec
+ },
#ifdef CONFIG_IP_ROUTE_MULTIPATH
{
.procname = "fib_multipath_use_neigh",
diff --git a/net/ipv4/tcp.c b/net/ipv4/tcp.c
index 1ef3165..91938c9 100644
--- a/net/ipv4/tcp.c
+++ b/net/ipv4/tcp.c
@@ -3334,6 +3334,7 @@ void __init tcp_init(void)

percpu_counter_init(&tcp_sockets_allocated, 0, GFP_KERNEL);
percpu_counter_init(&tcp_orphan_count, 0, GFP_KERNEL);
+ inet_hashinfo_init(&tcp_hashinfo);
tcp_hashinfo.bind_bucket_cachep =
kmem_cache_create("tcp_bind_bucket",
sizeof(struct inet_bind_bucket), 0,
@@ -3378,7 +3379,6 @@ void __init tcp_init(void)

cnt = tcp_hashinfo.ehash_mask + 1;

- tcp_death_row.sysctl_max_tw_buckets = cnt / 2;
sysctl_tcp_max_orphans = cnt / 2;
sysctl_max_syn_backlog = max(128, cnt / 256);

@@ -3399,6 +3399,7 @@ void __init tcp_init(void)
pr_info("Hash tables configured (established %u bind %u)\n",
tcp_hashinfo.ehash_mask + 1, tcp_hashinfo.bhash_size);

+ tcp_v4_init();
tcp_metrics_init();
BUG_ON(tcp_register_congestion_control(&tcp_reno) != 0);
tcp_tasklet_init();
diff --git a/net/ipv4/tcp_input.c b/net/ipv4/tcp_input.c
index 6c79075..c614802 100644
--- a/net/ipv4/tcp_input.c
+++ b/net/ipv4/tcp_input.c
@@ -6363,7 +6363,7 @@ int tcp_conn_request(struct request_sock_ops *rsk_ops,
* timewait bucket, so that all the necessary checks
* are made in the function processing timewait state.
*/
- if (tcp_death_row.sysctl_tw_recycle) {
+ if (net->ipv4.tcp_death_row.sysctl_tw_recycle) {
bool strict;

dst = af_ops->route_req(sk, &fl, req, &strict);
diff --git a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c
index fe9da4f..56b5f49 100644
--- a/net/ipv4/tcp_ipv4.c
+++ b/net/ipv4/tcp_ipv4.c
@@ -146,6 +146,7 @@ int tcp_v4_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len)
struct rtable *rt;
int err;
struct ip_options_rcu *inet_opt;
+ struct inet_timewait_death_row *tcp_death_row = &sock_net(sk)->ipv4.tcp_death_row;

if (addr_len < sizeof(struct sockaddr_in))
return -EINVAL;
@@ -196,7 +197,7 @@ int tcp_v4_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len)
tp->write_seq = 0;
}

- if (tcp_death_row.sysctl_tw_recycle &&
+ if (tcp_death_row->sysctl_tw_recycle &&
!tp->rx_opt.ts_recent_stamp && fl4->daddr == daddr)
tcp_fetch_timewait_stamp(sk, &rt->dst);

@@ -215,7 +216,7 @@ int tcp_v4_connect(struct sock *sk, struct sockaddr *uaddr, int addr_len)
* complete initialization after this.
*/
tcp_set_state(sk, TCP_SYN_SENT);
- err = inet_hash_connect(&tcp_death_row, sk);
+ err = inet_hash_connect(tcp_death_row, sk);
if (err)
goto failure;

@@ -2457,6 +2458,10 @@ static int __net_init tcp_sk_init(struct net *net)
net->ipv4.sysctl_tcp_notsent_lowat = UINT_MAX;
net->ipv4.sysctl_tcp_tw_reuse = 0;

+ net->ipv4.tcp_death_row.sysctl_tw_recycle = 0;
+ net->ipv4.tcp_death_row.sysctl_max_tw_buckets = (tcp_hashinfo.ehash_mask + 1) / 2;
+ net->ipv4.tcp_death_row.hashinfo = &tcp_hashinfo;
+
return 0;
fail:
tcp_sk_exit(net);
@@ -2466,7 +2471,7 @@ static int __net_init tcp_sk_init(struct net *net)

static void __net_exit tcp_sk_exit_batch(struct list_head *net_exit_list)
{
- inet_twsk_purge(&tcp_hashinfo, &tcp_death_row, AF_INET);
+ inet_twsk_purge(&tcp_hashinfo, AF_INET);
}

static struct pernet_operations __net_initdata tcp_sk_ops = {
@@ -2477,7 +2482,6 @@ static struct pernet_operations __net_initdata tcp_sk_ops = {

void __init tcp_v4_init(void)
{
- inet_hashinfo_init(&tcp_hashinfo);
if (register_pernet_subsys(&tcp_sk_ops))
panic("Failed to create the TCP control socket.\n");
}
diff --git a/net/ipv4/tcp_minisocks.c b/net/ipv4/tcp_minisocks.c
index 28ce5ee..06fde26 100644
--- a/net/ipv4/tcp_minisocks.c
+++ b/net/ipv4/tcp_minisocks.c
@@ -29,12 +29,6 @@

int sysctl_tcp_abort_on_overflow __read_mostly;

-struct inet_timewait_death_row tcp_death_row = {
- .sysctl_max_tw_buckets = NR_FILE * 2,
- .hashinfo = &tcp_hashinfo,
-};
-EXPORT_SYMBOL_GPL(tcp_death_row);
-
static bool tcp_in_window(u32 seq, u32 end_seq, u32 s_win, u32 e_win)
{
if (seq == s_win)
@@ -100,6 +94,7 @@ tcp_timewait_state_process(struct inet_timewait_sock *tw, struct sk_buff *skb,
struct tcp_options_received tmp_opt;
struct tcp_timewait_sock *tcptw = tcp_twsk((struct sock *)tw);
bool paws_reject = false;
+ struct inet_timewait_death_row *tcp_death_row = &sock_net((struct sock*)tw)->ipv4.tcp_death_row;

tmp_opt.saw_tstamp = 0;
if (th->doff > (sizeof(*th) >> 2) && tcptw->tw_ts_recent_stamp) {
@@ -153,7 +148,7 @@ tcp_timewait_state_process(struct inet_timewait_sock *tw, struct sk_buff *skb,
tcptw->tw_ts_recent = tmp_opt.rcv_tsval;
}

- if (tcp_death_row.sysctl_tw_recycle &&
+ if (tcp_death_row->sysctl_tw_recycle &&
tcptw->tw_ts_recent_stamp &&
tcp_tw_remember_stamp(tw))
inet_twsk_reschedule(tw, tw->tw_timeout);
@@ -264,11 +259,12 @@ void tcp_time_wait(struct sock *sk, int state, int timeo)
const struct tcp_sock *tp = tcp_sk(sk);
struct inet_timewait_sock *tw;
bool recycle_ok = false;
+ struct inet_timewait_death_row *tcp_death_row = &sock_net(sk)->ipv4.tcp_death_row;

- if (tcp_death_row.sysctl_tw_recycle && tp->rx_opt.ts_recent_stamp)
+ if (tcp_death_row->sysctl_tw_recycle && tp->rx_opt.ts_recent_stamp)
recycle_ok = tcp_remember_stamp(sk);

- tw = inet_twsk_alloc(sk, &tcp_death_row, state);
+ tw = inet_twsk_alloc(sk, tcp_death_row, state);

if (tw) {
struct tcp_timewait_sock *tcptw = tcp_twsk((struct sock *)tw);
diff --git a/net/ipv6/tcp_ipv6.c b/net/ipv6/tcp_ipv6.c
index 73bc8fc6..a4cdf6a 100644
--- a/net/ipv6/tcp_ipv6.c
+++ b/net/ipv6/tcp_ipv6.c
@@ -123,6 +123,7 @@ static int tcp_v6_connect(struct sock *sk, struct sockaddr *uaddr,
struct dst_entry *dst;
int addr_type;
int err;
+ struct inet_timewait_death_row *tcp_death_row = &sock_net(sk)->ipv4.tcp_death_row;

if (addr_len < SIN6_LEN_RFC2133)
return -EINVAL;
@@ -258,7 +259,7 @@ static int tcp_v6_connect(struct sock *sk, struct sockaddr *uaddr,
sk->sk_gso_type = SKB_GSO_TCPV6;
ip6_dst_store(sk, dst, NULL, NULL);

- if (tcp_death_row.sysctl_tw_recycle &&
+ if (tcp_death_row->sysctl_tw_recycle &&
!tp->rx_opt.ts_recent_stamp &&
ipv6_addr_equal(&fl6.daddr, &sk->sk_v6_daddr))
tcp_fetch_timewait_stamp(sk, dst);
@@ -273,7 +274,7 @@ static int tcp_v6_connect(struct sock *sk, struct sockaddr *uaddr,
inet->inet_dport = usin->sin6_port;

tcp_set_state(sk, TCP_SYN_SENT);
- err = inet6_hash_connect(&tcp_death_row, sk);
+ err = inet6_hash_connect(tcp_death_row, sk);
if (err)
goto late_failure;

@@ -1948,7 +1949,7 @@ static void __net_exit tcpv6_net_exit(struct net *net)

static void __net_exit tcpv6_net_exit_batch(struct list_head *net_exit_list)
{
- inet_twsk_purge(&tcp_hashinfo, &tcp_death_row, AF_INET6);
+ inet_twsk_purge(&tcp_hashinfo, AF_INET6);
}

static struct pernet_operations tcpv6_net_ops = {
--
2.5.0