[PATCH 4/4] Per IP network statistics: usage

From: Menny Hamburger
Date: Tue Sep 03 2013 - 03:52:14 EST

diff -r -U 4 a/include/net/ip.h b/include/net/ip.h
--- a/include/net/ip.h 2013-08-25 15:51:01.609307946 +0300
+++ b/include/net/ip.h 2013-08-25 15:51:01.692293432 +0300
@@ -202,8 +202,10 @@
#define NET_INC_STATS_USER(net, field) __NET_INC_STATS_USER(net, field)
#define NET_ADD_STATS_BH(net, field, adnd) __NET_ADD_STATS_BH(net, field, adnd)
#define NET_ADD_STATS_USER(net, field, adnd) __NET_ADD_STATS_USER(net, field, adnd)

+#include <net/stat_sk_hashtable_net_overrides.h>
+
extern unsigned long snmp_fold_field(void *mib[], int offt);
extern int snmp_mib_init(void *ptr[2], size_t mibsize);
extern void snmp_mib_free(void *ptr[2]);

diff -r -U 4 a/include/net/sock.h b/include/net/sock.h
--- a/include/net/sock.h 2013-08-25 15:51:01.615295421 +0300
+++ b/include/net/sock.h 2013-08-25 15:51:01.708293434 +0300
@@ -59,8 +59,12 @@
#include <asm/atomic.h>
#include <net/dst.h>
#include <net/checksum.h>

+#ifdef CONFIG_NET_IPV4_SOCK_STATS
+#include <net/stat_hashtable_cookie.h>
+#endif
+
/*
* This structure really needs to be cleaned up.
* Most of it is for TCP, and not used by any of
* the other protocols.
@@ -351,8 +355,12 @@
* sk_rcvqueues_full(), sk_set_min_ttl(), etc. would break for
* existing modules. */
__u8 rcv_tos;
u32 icsk_user_timeout;
+
+#ifdef CONFIG_NET_IPV4_SOCK_STATS
+ struct stat_hash_cookie hash_cookie;
+#endif
};

#define __sk_tx_queue_mapping(sk) \
sk_extended(sk)->__sk_common_extended.skc_tx_queue_mapping
diff -r -U 4 a/include/net/tcp.h b/include/net/tcp.h
--- a/include/net/tcp.h 2013-08-25 15:51:01.621293957 +0300
+++ b/include/net/tcp.h 2013-08-25 15:51:01.727293408 +0300
@@ -307,8 +307,10 @@
#define TCP_INC_STATS_BH(net, field) __TCP_INC_STATS_BH(net, field)
#define TCP_DEC_STATS(net, field) __TCP_DEC_STATS(net, field)
#define TCP_ADD_STATS_USER(net, field, val) __TCP_ADD_STATS_USER(net, field, val)

+#include <net/stat_sk_hashtable_tcp_overrides.h>
+
extern void tcp_v4_err(struct sk_buff *skb, u32);

extern void tcp_shutdown (struct sock *sk, int how);

diff -r -U 4 a/net/core/sock.c b/net/core/sock.c
--- a/net/core/sock.c 2013-08-25 15:51:01.627303626 +0300
+++ b/net/core/sock.c 2013-08-25 15:51:01.744299451 +0300
@@ -133,8 +133,10 @@
#ifdef CONFIG_INET
#include <net/tcp.h>
#endif

+#include <net/stat_sk_hashtable.h>
+
/*
* Each address family might have different locking rules, so we have
* one slock key per address family:
*/
@@ -1105,8 +1107,10 @@
atomic_set(&sk->sk_wmem_alloc, 1);

sock_update_classid(sk);
sock_update_netprioidx(sk);
+
+ STAT_HASH_SK_INIT(sk);
}

return sk;
}
@@ -1255,8 +1259,10 @@

if (sock_flag(newsk, SOCK_TIMESTAMP) ||
sock_flag(newsk, SOCK_TIMESTAMPING_RX_SOFTWARE))
net_enable_timestamp();
+
+ STAT_HASH_SK_INIT(newsk);
}
out:
return newsk;
}
diff -r -U 4 a/net/ipv4/arp.c b/net/ipv4/arp.c
--- a/net/ipv4/arp.c 2013-08-25 15:51:01.633309737 +0300
+++ b/net/ipv4/arp.c 2013-08-25 15:51:01.761359596 +0300
@@ -426,9 +426,9 @@

if (ip_route_output_key(net, &rt, &fl) < 0)
return 1;
if (rt->u.dst.dev != dev) {
- NET_INC_STATS_BH(net, LINUX_MIB_ARPFILTER);
+ __NET_INC_STATS_BH(net, LINUX_MIB_ARPFILTER);
flag = 1;
}
ip_rt_put(rt);
return flag;
diff -r -U 4 a/net/ipv4/inet_timewait_sock.c b/net/ipv4/inet_timewait_sock.c
--- a/net/ipv4/inet_timewait_sock.c 2013-08-25 15:51:01.639297975 +0300
+++ b/net/ipv4/inet_timewait_sock.c 2013-08-25 15:51:01.774487246 +0300
@@ -171,9 +171,9 @@
__inet_twsk_del_dead_node(tw);
spin_unlock(&twdr->death_lock);
__inet_twsk_kill(tw, twdr->hashinfo);
#ifdef CONFIG_NET_NS
- NET_INC_STATS_BH(twsk_net(tw), LINUX_MIB_TIMEWAITED);
+ __NET_INC_STATS_BH(twsk_net(tw), LINUX_MIB_TIMEWAITED);
#endif
inet_twsk_put(tw);
killed++;
spin_lock(&twdr->death_lock);
@@ -192,9 +192,9 @@
}

twdr->tw_count -= killed;
#ifndef CONFIG_NET_NS
- NET_ADD_STATS_BH(&init_net, LINUX_MIB_TIMEWAITED, killed);
+ __NET_ADD_STATS_BH(&init_net, LINUX_MIB_TIMEWAITED, killed);
#endif
return ret;
}

@@ -387,9 +387,9 @@
&twdr->twcal_row[slot]) {
__inet_twsk_del_dead_node(tw);
__inet_twsk_kill(tw, twdr->hashinfo);
#ifdef CONFIG_NET_NS
- NET_INC_STATS_BH(twsk_net(tw), LINUX_MIB_TIMEWAITKILLED);
+ __NET_INC_STATS_BH(twsk_net(tw), LINUX_MIB_TIMEWAITKILLED);
#endif
inet_twsk_put(tw);
killed++;
}
diff -r -U 4 a/net/ipv4/tcp.c b/net/ipv4/tcp.c
--- a/net/ipv4/tcp.c 2013-08-25 15:51:01.645293122 +0300
+++ b/net/ipv4/tcp.c 2013-08-25 15:51:01.787393766 +0300
@@ -1756,8 +1756,10 @@
void tcp_set_state(struct sock *sk, int state)
{
int oldstate = sk->sk_state;

+ STAT_HASH_SK_INSERT_EXISTING_NOALLOC(sk);
+
switch (state) {
case TCP_ESTABLISHED:
if (oldstate != TCP_ESTABLISHED)
TCP_INC_STATS(sock_net(sk), TCP_MIB_CURRESTAB);
diff -r -U 4 a/net/ipv4/tcp_ipv4.c b/net/ipv4/tcp_ipv4.c
--- a/net/ipv4/tcp_ipv4.c 2013-08-25 15:51:01.651303464 +0300
+++ b/net/ipv4/tcp_ipv4.c 2013-08-25 15:51:01.804255105 +0300
@@ -248,8 +248,10 @@
usin->sin_port);

inet->id = tp->write_seq ^ jiffies;

+ STAT_HASH_SK_INSERT_ATOMIC(sk, STAT_HASH_SK_OUT);
+
err = tcp_connect(sk);
rt = NULL;
if (err)
goto failure;
@@ -633,10 +635,10 @@
net = dev_net(skb_dst(skb)->dev);
ip_send_reply(net->ipv4.tcp_sock, skb,
&arg, arg.iov[0].iov_len);

- TCP_INC_STATS_BH(net, TCP_MIB_OUTSEGS);
- TCP_INC_STATS_BH(net, TCP_MIB_OUTRSTS);
+ TCP_INC_SOCK_STATS_BH(net->ipv4.tcp_sock, net, TCP_MIB_OUTSEGS);
+ TCP_INC_SOCK_STATS_BH(net->ipv4.tcp_sock, net, TCP_MIB_OUTRSTS);
}

/* The code following below sending ACKs in SYN-RECV and TIME-WAIT states
outside socket context is ugly, certainly. What can I do?
@@ -708,9 +710,9 @@

ip_send_reply(net->ipv4.tcp_sock, skb,
&arg, arg.iov[0].iov_len);

- TCP_INC_STATS_BH(net, TCP_MIB_OUTSEGS);
+ TCP_INC_SOCK_STATS_BH(net->ipv4.tcp_sock, net, TCP_MIB_OUTSEGS);
}

static void tcp_v4_timewait_ack(struct sock *sk, struct sk_buff *skb)
{
@@ -1272,8 +1274,10 @@
tmp_opt.tstamp_ok = tmp_opt.saw_tstamp;

tcp_openreq_init(req, &tmp_opt, skb);

+ STAT_HASH_SK_INSERT_ATOMIC(sk, STAT_HASH_SK_IN);
+
ireq = inet_rsk(req);
ireq->loc_addr = daddr;
ireq->rmt_addr = saddr;
ireq->no_srccheck = inet_sk(sk)->transparent;
@@ -1435,8 +1439,10 @@
goto exit;
}
__inet_hash_nolisten(newsk);

+ STAT_HASH_SK_INSERT_ATOMIC(newsk, STAT_HASH_SK_IN);
+
return newsk;

exit_overflow:
NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_LISTENOVERFLOWS);
@@ -1580,17 +1586,17 @@
int tcp_v4_rcv(struct sk_buff *skb)
{
const struct iphdr *iph;
struct tcphdr *th;
- struct sock *sk;
+ struct sock *sk = NULL;
int ret;
struct net *net = dev_net(skb->dev);

if (skb->pkt_type != PACKET_HOST)
goto discard_it;

/* Count it even if it's bad */
- TCP_INC_STATS_BH(net, TCP_MIB_INSEGS);
+ TCP_INC_SKB_STATS_BH(net, TCP_MIB_INSEGS);

if (!pskb_may_pull(skb, sizeof(struct tcphdr)))
goto discard_it;

diff -r -U 4 a/net/ipv4/tcp_minisocks.c b/net/ipv4/tcp_minisocks.c
--- a/net/ipv4/tcp_minisocks.c 2013-08-25 15:51:01.656293235 +0300
+++ b/net/ipv4/tcp_minisocks.c 2013-08-25 15:51:01.817254458 +0300
@@ -244,9 +244,9 @@
return TCP_TW_SYN;
}

if (paws_reject)
- NET_INC_STATS_BH(twsk_net(tw), LINUX_MIB_PAWSESTABREJECTED);
+ __NET_INC_STATS_BH(twsk_net(tw), LINUX_MIB_PAWSESTABREJECTED);

if (!th->rst) {
/* In this case we must reset the TIMEWAIT timer.
*
diff -r -U 4 a/net/ipv6/inet6_hashtables.c b/net/ipv6/inet6_hashtables.c
--- a/net/ipv6/inet6_hashtables.c 2013-08-25 15:51:01.662303572 +0300
+++ b/net/ipv6/inet6_hashtables.c 2013-08-25 15:51:01.830264020 +0300
@@ -22,8 +22,12 @@
#include <net/inet6_hashtables.h>
#include <net/secure_seq.h>
#include <net/ip.h>

+#ifndef CONFIG_IPV6_STAT_HASHTABLES
+#include <net/stat_sk_hashtable_undo_overrides.h>
+#endif
+
void __inet6_hash(struct sock *sk)
{
struct inet_hashinfo *hashinfo = sk->sk_prot->h.hashinfo;

diff -r -U 4 a/net/ipv6/syncookies.c b/net/ipv6/syncookies.c
--- a/net/ipv6/syncookies.c 2013-08-25 15:51:01.668316079 +0300
+++ b/net/ipv6/syncookies.c 2013-08-25 15:51:01.843250995 +0300
@@ -20,8 +20,12 @@
#include <linux/kernel.h>
#include <net/ipv6.h>
#include <net/tcp.h>

+#ifndef CONFIG_IPV6_STAT_HASHTABLES
+#include <net/stat_sk_hashtable_undo_overrides.h>
+#endif
+
extern int sysctl_tcp_syncookies;
extern __u32 syncookie_secret[2][16-4+SHA_DIGEST_WORDS];

#define COOKIEBITS 24 /* Upper bits store count */
diff -r -U 4 a/net/ipv6/tcp_ipv6.c b/net/ipv6/tcp_ipv6.c
--- a/net/ipv6/tcp_ipv6.c 2013-08-25 15:51:01.674293247 +0300
+++ b/net/ipv6/tcp_ipv6.c 2013-08-25 15:51:01.859250310 +0300
@@ -69,8 +69,12 @@

#include <linux/crypto.h>
#include <linux/scatterlist.h>

+#ifndef CONFIG_IPV6_STAT_HASHTABLES
+#include <net/stat_sk_hashtable_undo_overrides.h>
+#endif
+
static void tcp_v6_send_reset(struct sock *sk, struct sk_buff *skb);
static void tcp_v6_reqsk_send_ack(struct sock *sk, struct sk_buff *skb,
struct request_sock *req);

@@ -301,8 +305,10 @@
np->daddr.s6_addr32,
inet->sport,
inet->dport);

+ STAT_HASH_SK_INSERT_ATOMIC(sk, STAT_HASH_SK_OUT);
+
err = tcp_connect(sk);
if (err)
goto late_failure;

@@ -1064,11 +1070,11 @@
if (!ip6_dst_lookup(ctl_sk, &dst, &fl)) {
if (xfrm_lookup(net, &dst, &fl, NULL, 0) >= 0) {
skb_dst_set(buff, dst);
ip6_xmit(ctl_sk, buff, &fl, NULL, 0);
- TCP_INC_STATS_BH(net, TCP_MIB_OUTSEGS);
+ TCP_INC_SKB_STATS_BH(net, TCP_MIB_OUTSEGS);
if (rst)
- TCP_INC_STATS_BH(net, TCP_MIB_OUTRSTS);
+ TCP_INC_SKB_STATS_BH(net, TCP_MIB_OUTRSTS);
return;
}
}

@@ -1210,8 +1216,10 @@
tmp_opt.user_mss = tp->rx_opt.user_mss;

tcp_parse_options(skb, &tmp_opt, 0);

+ STAT_HASH_SK_INSERT_ATOMIC(sk, STAT_HASH_SK_IN);
+
if (want_cookie && !tmp_opt.saw_tstamp)
tcp_clear_options(&tmp_opt);

tmp_opt.tstamp_ok = tmp_opt.saw_tstamp;
@@ -1329,8 +1337,10 @@
Sync it now.
*/
tcp_sync_mss(newsk, inet_csk(newsk)->icsk_pmtu_cookie);

+ STAT_HASH_SK_INSERT_ATOMIC(newsk, STAT_HASH_SK_IN);
+
return newsk;
}

treq = inet6_rsk(req);
@@ -1640,9 +1650,9 @@
static int tcp_v6_rcv(struct sk_buff *skb)
{
struct tcphdr *th;
struct ipv6hdr *hdr;
- struct sock *sk;
+ struct sock *sk = NULL;
int ret;
struct net *net = dev_net(skb->dev);

if (skb->pkt_type != PACKET_HOST)
@@ -1650,9 +1660,9 @@

/*
* Count it even if it's bad.
*/
- TCP_INC_STATS_BH(net, TCP_MIB_INSEGS);
+ TCP_INC_SKB_STATS_BH(net, TCP_MIB_INSEGS);

if (!pskb_may_pull(skb, sizeof(struct tcphdr)))
goto discard_it;

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/