[net-next-2.6.git] / net / ipv6 / inet6_hashtables.c

/*
 * INET		An implementation of the TCP/IP protocol suite for the LINUX
 *		operating system.  INET is implemented using the BSD Socket
 *		interface as the means of communication with the user level.
 *
 *		Generic INET6 transport hashtables
 *
 * Authors:	Lotsa people, from code originally in tcp, generalised here
 * 		by Arnaldo Carvalho de Melo <acme@mandriva.com>
 *
 *	This program is free software; you can redistribute it and/or
 *      modify it under the terms of the GNU General Public License
 *      as published by the Free Software Foundation; either version
 *      2 of the License, or (at your option) any later version.
 */

#include <linux/module.h>
#include <linux/random.h>

#include <net/inet_connection_sock.h>
#include <net/inet_hashtables.h>
#include <net/inet6_hashtables.h>
#include <net/ip.h>

void __inet6_hash(struct sock *sk)
{
	struct inet_hashinfo *hashinfo = sk->sk_prot->h.hashinfo;
	struct hlist_head *list;
	rwlock_t *lock;

	WARN_ON(!sk_unhashed(sk));

	if (sk->sk_state == TCP_LISTEN) {
		list = &hashinfo->listening_hash[inet_sk_listen_hashfn(sk)];
		lock = &hashinfo->lhash_lock;
		inet_listen_wlock(hashinfo);
	} else {
		unsigned int hash;
		sk->sk_hash = hash = inet6_sk_ehashfn(sk);
		list = &inet_ehash_bucket(hashinfo, hash)->chain;
		lock = inet_ehash_lockp(hashinfo, hash);
		write_lock(lock);
	}

	__sk_add_node(sk, list);
	sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1);
	write_unlock(lock);
}
EXPORT_SYMBOL(__inet6_hash);

/*
 * Sockets in TCP_CLOSE state are _always_ taken out of the hash, so
 * we need not check it for TCP lookups anymore, thanks Alexey. -DaveM
 *
 * The sockhash lock must be held as a reader here.
 */
struct sock *__inet6_lookup_established(struct net *net,
					struct inet_hashinfo *hashinfo,
					   const struct in6_addr *saddr,
					   const __be16 sport,
					   const struct in6_addr *daddr,
					   const u16 hnum,
					   const int dif)
{
	struct sock *sk;
	const struct hlist_node *node;
	const __portpair ports = INET_COMBINED_PORTS(sport, hnum);
	/* Optimize here for direct hit, only listening connections can
	 * have wildcards anyways.
	 */
	unsigned int hash = inet6_ehashfn(net, daddr, hnum, saddr, sport);
	struct inet_ehash_bucket *head = inet_ehash_bucket(hashinfo, hash);
	rwlock_t *lock = inet_ehash_lockp(hashinfo, hash);

	prefetch(head->chain.first);
	read_lock(lock);
	sk_for_each(sk, node, &head->chain) {
		/* For IPV6 do the cheaper port and family tests first. */
		if (INET6_MATCH(sk, net, hash, saddr, daddr, ports, dif))
			goto hit; /* You sunk my battleship! */
	}
	/* Must check for a TIME_WAIT'er before going to listener hash. */
	sk_for_each(sk, node, &head->twchain) {
		if (INET6_TW_MATCH(sk, net, hash, saddr, daddr, ports, dif))
			goto hit;
	}
	read_unlock(lock);
	return NULL;

hit:
	sock_hold(sk);
	read_unlock(lock);
	return sk;
}
EXPORT_SYMBOL(__inet6_lookup_established);

struct sock *inet6_lookup_listener(struct net *net,
		struct inet_hashinfo *hashinfo, const struct in6_addr *daddr,
		const unsigned short hnum, const int dif)
{
	struct sock *sk;
	const struct hlist_node *node;
	struct sock *result = NULL;
	int score, hiscore = 0;

	read_lock(&hashinfo->lhash_lock);
	sk_for_each(sk, node,
			&hashinfo->listening_hash[inet_lhashfn(net, hnum)]) {
		if (net_eq(sock_net(sk), net) && inet_sk(sk)->num == hnum &&
				sk->sk_family == PF_INET6) {
			const struct ipv6_pinfo *np = inet6_sk(sk);

			score = 1;
			if (!ipv6_addr_any(&np->rcv_saddr)) {
				if (!ipv6_addr_equal(&np->rcv_saddr, daddr))
					continue;
				score++;
			}
			if (sk->sk_bound_dev_if) {
				if (sk->sk_bound_dev_if != dif)
					continue;
				score++;
			}
			if (score == 3) {
				result = sk;
				break;
			}
			if (score > hiscore) {
				hiscore = score;
				result = sk;
			}
		}
	}
	if (result)
		sock_hold(result);
	read_unlock(&hashinfo->lhash_lock);
	return result;
}

EXPORT_SYMBOL_GPL(inet6_lookup_listener);

struct sock *inet6_lookup(struct net *net, struct inet_hashinfo *hashinfo,
			  const struct in6_addr *saddr, const __be16 sport,
			  const struct in6_addr *daddr, const __be16 dport,
			  const int dif)
{
	struct sock *sk;

	local_bh_disable();
	sk = __inet6_lookup(net, hashinfo, saddr, sport, daddr, ntohs(dport), dif);
	local_bh_enable();

	return sk;
}

EXPORT_SYMBOL_GPL(inet6_lookup);

static int __inet6_check_established(struct inet_timewait_death_row *death_row,
				     struct sock *sk, const __u16 lport,
				     struct inet_timewait_sock **twp)
{
	struct inet_hashinfo *hinfo = death_row->hashinfo;
	struct inet_sock *inet = inet_sk(sk);
	const struct ipv6_pinfo *np = inet6_sk(sk);
	const struct in6_addr *daddr = &np->rcv_saddr;
	const struct in6_addr *saddr = &np->daddr;
	const int dif = sk->sk_bound_dev_if;
	const __portpair ports = INET_COMBINED_PORTS(inet->dport, lport);
	struct net *net = sock_net(sk);
	const unsigned int hash = inet6_ehashfn(net, daddr, lport, saddr,
						inet->dport);
	struct inet_ehash_bucket *head = inet_ehash_bucket(hinfo, hash);
	rwlock_t *lock = inet_ehash_lockp(hinfo, hash);
	struct sock *sk2;
	const struct hlist_node *node;
	struct inet_timewait_sock *tw;

	prefetch(head->chain.first);
	write_lock(lock);

	/* Check TIME-WAIT sockets first. */
	sk_for_each(sk2, node, &head->twchain) {
		tw = inet_twsk(sk2);

		if (INET6_TW_MATCH(sk2, net, hash, saddr, daddr, ports, dif)) {
			if (twsk_unique(sk, sk2, twp))
				goto unique;
			else
				goto not_unique;
		}
	}
	tw = NULL;

	/* And established part... */
	sk_for_each(sk2, node, &head->chain) {
		if (INET6_MATCH(sk2, net, hash, saddr, daddr, ports, dif))
			goto not_unique;
	}

unique:
	/* Must record num and sport now. Otherwise we will see
	 * in hash table socket with a funny identity. */
	inet->num = lport;
	inet->sport = htons(lport);
	WARN_ON(!sk_unhashed(sk));
	__sk_add_node(sk, &head->chain);
	sk->sk_hash = hash;
	sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1);
	write_unlock(lock);

	if (twp != NULL) {
		*twp = tw;
		NET_INC_STATS_BH(twsk_net(tw), LINUX_MIB_TIMEWAITRECYCLED);
	} else if (tw != NULL) {
		/* Silly. Should hash-dance instead... */
		inet_twsk_deschedule(tw, death_row);
		NET_INC_STATS_BH(twsk_net(tw), LINUX_MIB_TIMEWAITRECYCLED);

		inet_twsk_put(tw);
	}
	return 0;

not_unique:
	write_unlock(lock);
	return -EADDRNOTAVAIL;
}

static inline u32 inet6_sk_port_offset(const struct sock *sk)
{
	const struct inet_sock *inet = inet_sk(sk);
	const struct ipv6_pinfo *np = inet6_sk(sk);
	return secure_ipv6_port_ephemeral(np->rcv_saddr.s6_addr32,
					  np->daddr.s6_addr32,
					  inet->dport);
}

int inet6_hash_connect(struct inet_timewait_death_row *death_row,
		       struct sock *sk)
{
	return __inet_hash_connect(death_row, sk, inet6_sk_port_offset(sk),
			__inet6_check_established, __inet6_hash);
}

EXPORT_SYMBOL_GPL(inet6_hash_connect);
Commit	Line	Data
5324a040 ACM	1	/*
	2	* INET An implementation of the TCP/IP protocol suite for the LINUX
	3	* operating system. INET is implemented using the BSD Socket
	4	* interface as the means of communication with the user level.
	5	*
	6	* Generic INET6 transport hashtables
	7	*
d8313f5c ACM	8	* Authors: Lotsa people, from code originally in tcp, generalised here
d8313f5c ACM	9	* by Arnaldo Carvalho de Melo <acme@mandriva.com>
5324a040 ACM	10	*
	11	* This program is free software; you can redistribute it and/or
	12	* modify it under the terms of the GNU General Public License
	13	* as published by the Free Software Foundation; either version
	14	* 2 of the License, or (at your option) any later version.
	15	*/
	16
5324a040	17	#include <linux/module.h>
d8313f5c	18	#include <linux/random.h>
5324a040 ACM	19
	20	#include <net/inet_connection_sock.h>
	21	#include <net/inet_hashtables.h>
	22	#include <net/inet6_hashtables.h>
d8313f5c	23	#include <net/ip.h>
5324a040	24
ab1e0a13	25	void __inet6_hash(struct sock *sk)
b1a7ffcb	26	{
39d8cda7	27	struct inet_hashinfo *hashinfo = sk->sk_prot->h.hashinfo;
b1a7ffcb DV	28	struct hlist_head *list;
	29	rwlock_t *lock;
	30
547b792c	31	WARN_ON(!sk_unhashed(sk));
b1a7ffcb DV	32
	33	if (sk->sk_state == TCP_LISTEN) {
	34	list = &hashinfo->listening_hash[inet_sk_listen_hashfn(sk)];
	35	lock = &hashinfo->lhash_lock;
	36	inet_listen_wlock(hashinfo);
	37	} else {
	38	unsigned int hash;
	39	sk->sk_hash = hash = inet6_sk_ehashfn(sk);
230140cf ED	40	list = &inet_ehash_bucket(hashinfo, hash)->chain;
230140cf ED	41	lock = inet_ehash_lockp(hashinfo, hash);
b1a7ffcb DV	42	write_lock(lock);
	43	}
	44
	45	__sk_add_node(sk, list);
c29a0bc4	46	sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1);
b1a7ffcb DV	47	write_unlock(lock);
	48	}
	49	EXPORT_SYMBOL(__inet6_hash);
	50
	51	/*
	52	* Sockets in TCP_CLOSE state are _always_ taken out of the hash, so
	53	* we need not check it for TCP lookups anymore, thanks Alexey. -DaveM
	54	*
	55	* The sockhash lock must be held as a reader here.
	56	*/
d86e0dac PE	57	struct sock __inet6_lookup_established(struct net net,
d86e0dac PE	58	struct inet_hashinfo *hashinfo,
b1a7ffcb	59	const struct in6_addr *saddr,
d2ecd9cc	60	const __be16 sport,
b1a7ffcb DV	61	const struct in6_addr *daddr,
	62	const u16 hnum,
	63	const int dif)
	64	{
	65	struct sock *sk;
	66	const struct hlist_node *node;
4f765d84	67	const __portpair ports = INET_COMBINED_PORTS(sport, hnum);
b1a7ffcb DV	68	/* Optimize here for direct hit, only listening connections can
	69	* have wildcards anyways.
	70	*/
33de014c	71	unsigned int hash = inet6_ehashfn(net, daddr, hnum, saddr, sport);
b1a7ffcb	72	struct inet_ehash_bucket *head = inet_ehash_bucket(hashinfo, hash);
230140cf	73	rwlock_t *lock = inet_ehash_lockp(hashinfo, hash);
b1a7ffcb DV	74
b1a7ffcb DV	75	prefetch(head->chain.first);
230140cf	76	read_lock(lock);
b1a7ffcb DV	77	sk_for_each(sk, node, &head->chain) {
b1a7ffcb DV	78	/* For IPV6 do the cheaper port and family tests first. */
d86e0dac	79	if (INET6_MATCH(sk, net, hash, saddr, daddr, ports, dif))
b1a7ffcb DV	80	goto hit; /* You sunk my battleship! */
	81	}
	82	/* Must check for a TIME_WAIT'er before going to listener hash. */
dbca9b27	83	sk_for_each(sk, node, &head->twchain) {
d86e0dac	84	if (INET6_TW_MATCH(sk, net, hash, saddr, daddr, ports, dif))
535174ef	85	goto hit;
b1a7ffcb	86	}
230140cf	87	read_unlock(lock);
b1a7ffcb DV	88	return NULL;
	89
	90	hit:
	91	sock_hold(sk);
230140cf	92	read_unlock(lock);
b1a7ffcb DV	93	return sk;
	94	}
	95	EXPORT_SYMBOL(__inet6_lookup_established);
	96
d86e0dac PE	97	struct sock inet6_lookup_listener(struct net net,
	98	struct inet_hashinfo hashinfo, const struct in6_addr daddr,
	99	const unsigned short hnum, const int dif)
5324a040 ACM	100	{
	101	struct sock *sk;
	102	const struct hlist_node *node;
	103	struct sock *result = NULL;
	104	int score, hiscore = 0;
	105
	106	read_lock(&hashinfo->lhash_lock);
2086a650 PE	107	sk_for_each(sk, node,
2086a650 PE	108	&hashinfo->listening_hash[inet_lhashfn(net, hnum)]) {
878628fb	109	if (net_eq(sock_net(sk), net) && inet_sk(sk)->num == hnum &&
d86e0dac	110	sk->sk_family == PF_INET6) {
5324a040	111	const struct ipv6_pinfo *np = inet6_sk(sk);
1ab1457c	112
5324a040 ACM	113	score = 1;
	114	if (!ipv6_addr_any(&np->rcv_saddr)) {
	115	if (!ipv6_addr_equal(&np->rcv_saddr, daddr))
	116	continue;
	117	score++;
	118	}
	119	if (sk->sk_bound_dev_if) {
	120	if (sk->sk_bound_dev_if != dif)
	121	continue;
	122	score++;
	123	}
	124	if (score == 3) {
	125	result = sk;
	126	break;
	127	}
	128	if (score > hiscore) {
	129	hiscore = score;
	130	result = sk;
	131	}
	132	}
	133	}
	134	if (result)
	135	sock_hold(result);
	136	read_unlock(&hashinfo->lhash_lock);
	137	return result;
	138	}
	139
	140	EXPORT_SYMBOL_GPL(inet6_lookup_listener);
	141
d86e0dac	142	struct sock inet6_lookup(struct net net, struct inet_hashinfo *hashinfo,
d2ecd9cc AV	143	const struct in6_addr *saddr, const __be16 sport,
d2ecd9cc AV	144	const struct in6_addr *daddr, const __be16 dport,
5324a040 ACM	145	const int dif)
	146	{
	147	struct sock *sk;
	148
	149	local_bh_disable();
d86e0dac	150	sk = __inet6_lookup(net, hashinfo, saddr, sport, daddr, ntohs(dport), dif);
5324a040 ACM	151	local_bh_enable();
	152
	153	return sk;
	154	}
	155
	156	EXPORT_SYMBOL_GPL(inet6_lookup);
d8313f5c ACM	157
	158	static int __inet6_check_established(struct inet_timewait_death_row *death_row,
	159	struct sock *sk, const __u16 lport,
	160	struct inet_timewait_sock **twp)
	161	{
	162	struct inet_hashinfo *hinfo = death_row->hashinfo;
3759fa9c	163	struct inet_sock *inet = inet_sk(sk);
d8313f5c ACM	164	const struct ipv6_pinfo *np = inet6_sk(sk);
	165	const struct in6_addr *daddr = &np->rcv_saddr;
	166	const struct in6_addr *saddr = &np->daddr;
	167	const int dif = sk->sk_bound_dev_if;
4f765d84	168	const __portpair ports = INET_COMBINED_PORTS(inet->dport, lport);
33de014c PE	169	struct net *net = sock_net(sk);
33de014c PE	170	const unsigned int hash = inet6_ehashfn(net, daddr, lport, saddr,
d8313f5c ACM	171	inet->dport);
d8313f5c ACM	172	struct inet_ehash_bucket *head = inet_ehash_bucket(hinfo, hash);
230140cf	173	rwlock_t *lock = inet_ehash_lockp(hinfo, hash);
d8313f5c ACM	174	struct sock *sk2;
	175	const struct hlist_node *node;
	176	struct inet_timewait_sock *tw;
	177
	178	prefetch(head->chain.first);
230140cf	179	write_lock(lock);
d8313f5c ACM	180
d8313f5c ACM	181	/* Check TIME-WAIT sockets first. */
dbca9b27	182	sk_for_each(sk2, node, &head->twchain) {
d8313f5c ACM	183	tw = inet_twsk(sk2);
d8313f5c ACM	184
d86e0dac	185	if (INET6_TW_MATCH(sk2, net, hash, saddr, daddr, ports, dif)) {
d8313f5c ACM	186	if (twsk_unique(sk, sk2, twp))
	187	goto unique;
	188	else
	189	goto not_unique;
	190	}
	191	}
	192	tw = NULL;
	193
	194	/* And established part... */
	195	sk_for_each(sk2, node, &head->chain) {
d86e0dac	196	if (INET6_MATCH(sk2, net, hash, saddr, daddr, ports, dif))
d8313f5c ACM	197	goto not_unique;
	198	}
	199
	200	unique:
3759fa9c HX	201	/* Must record num and sport now. Otherwise we will see
	202	* in hash table socket with a funny identity. */
	203	inet->num = lport;
	204	inet->sport = htons(lport);
547b792c	205	WARN_ON(!sk_unhashed(sk));
d8313f5c ACM	206	__sk_add_node(sk, &head->chain);
d8313f5c ACM	207	sk->sk_hash = hash;
c29a0bc4	208	sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1);
230140cf	209	write_unlock(lock);
d8313f5c ACM	210
	211	if (twp != NULL) {
	212	*twp = tw;
de0744af	213	NET_INC_STATS_BH(twsk_net(tw), LINUX_MIB_TIMEWAITRECYCLED);
d8313f5c ACM	214	} else if (tw != NULL) {
	215	/* Silly. Should hash-dance instead... */
	216	inet_twsk_deschedule(tw, death_row);
de0744af	217	NET_INC_STATS_BH(twsk_net(tw), LINUX_MIB_TIMEWAITRECYCLED);
d8313f5c ACM	218
	219	inet_twsk_put(tw);
	220	}
	221	return 0;
	222
	223	not_unique:
230140cf	224	write_unlock(lock);
d8313f5c ACM	225	return -EADDRNOTAVAIL;
	226	}
	227
	228	static inline u32 inet6_sk_port_offset(const struct sock *sk)
	229	{
	230	const struct inet_sock *inet = inet_sk(sk);
	231	const struct ipv6_pinfo *np = inet6_sk(sk);
	232	return secure_ipv6_port_ephemeral(np->rcv_saddr.s6_addr32,
	233	np->daddr.s6_addr32,
	234	inet->dport);
	235	}
	236
	237	int inet6_hash_connect(struct inet_timewait_death_row *death_row,
	238	struct sock *sk)
	239	{
5d8c0aa9	240	return __inet_hash_connect(death_row, sk, inet6_sk_port_offset(sk),
5ee31fc1	241	__inet6_check_established, __inet6_hash);
d8313f5c ACM	242	}
	243
	244	EXPORT_SYMBOL_GPL(inet6_hash_connect);