inet6_hashtables.c 8.0 KB
Newer Older
1 2 3 4 5 6 7
/*
 * 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
 *
8 9
 * Authors:	Lotsa people, from code originally in tcp, generalised here
 * 		by Arnaldo Carvalho de Melo <acme@mandriva.com>
10 11 12 13 14 15 16 17
 *
 *	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>
18
#include <linux/random.h>
19 20 21 22

#include <net/inet_connection_sock.h>
#include <net/inet_hashtables.h>
#include <net/inet6_hashtables.h>
23
#include <net/ip.h>
24

25
int __inet6_hash(struct sock *sk, struct inet_timewait_sock *tw)
26
{
27
	struct inet_hashinfo *hashinfo = sk->sk_prot->h.hashinfo;
28
	int twrefcnt = 0;
29

30
	WARN_ON(!sk_unhashed(sk));
31 32

	if (sk->sk_state == TCP_LISTEN) {
33
		struct inet_listen_hashbucket *ilb;
34

35 36
		ilb = &hashinfo->listening_hash[inet_sk_listen_hashfn(sk)];
		spin_lock(&ilb->lock);
37
		__sk_nulls_add_node_rcu(sk, &ilb->head);
38
		spin_unlock(&ilb->lock);
39 40
	} else {
		unsigned int hash;
41
		struct hlist_nulls_head *list;
42
		spinlock_t *lock;
43

44
		sk->sk_hash = hash = inet6_sk_ehashfn(sk);
45 46
		list = &inet_ehash_bucket(hashinfo, hash)->chain;
		lock = inet_ehash_lockp(hashinfo, hash);
47
		spin_lock(lock);
48
		__sk_nulls_add_node_rcu(sk, list);
49 50 51 52
		if (tw) {
			WARN_ON(sk->sk_hash != tw->tw_hash);
			twrefcnt = inet_twsk_unhash(tw);
		}
53
		spin_unlock(lock);
54 55
	}

56
	sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1);
57
	return twrefcnt;
58 59 60 61 62 63 64 65 66
}
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.
 */
67 68
struct sock *__inet6_lookup_established(struct net *net,
					struct inet_hashinfo *hashinfo,
69
					   const struct in6_addr *saddr,
A
Al Viro 已提交
70
					   const __be16 sport,
71 72 73 74 75
					   const struct in6_addr *daddr,
					   const u16 hnum,
					   const int dif)
{
	struct sock *sk;
76
	const struct hlist_nulls_node *node;
A
Al Viro 已提交
77
	const __portpair ports = INET_COMBINED_PORTS(sport, hnum);
78 79 80
	/* Optimize here for direct hit, only listening connections can
	 * have wildcards anyways.
	 */
81
	unsigned int hash = inet6_ehashfn(net, daddr, hnum, saddr, sport);
82
	unsigned int slot = hash & hashinfo->ehash_mask;
83
	struct inet_ehash_bucket *head = &hashinfo->ehash[slot];
84

85 86 87 88

	rcu_read_lock();
begin:
	sk_nulls_for_each_rcu(sk, node, &head->chain) {
89
		/* For IPV6 do the cheaper port and family tests first. */
90 91 92 93 94 95 96 97 98
		if (INET6_MATCH(sk, net, hash, saddr, daddr, ports, dif)) {
			if (unlikely(!atomic_inc_not_zero(&sk->sk_refcnt)))
				goto begintw;
			if (!INET6_MATCH(sk, net, hash, saddr, daddr, ports, dif)) {
				sock_put(sk);
				goto begin;
			}
		goto out;
		}
99
	}
100 101 102 103
	if (get_nulls_value(node) != slot)
		goto begin;

begintw:
104
	/* Must check for a TIME_WAIT'er before going to listener hash. */
105 106 107 108 109 110 111 112 113 114 115 116
	sk_nulls_for_each_rcu(sk, node, &head->twchain) {
		if (INET6_TW_MATCH(sk, net, hash, saddr, daddr, ports, dif)) {
			if (unlikely(!atomic_inc_not_zero(&sk->sk_refcnt))) {
				sk = NULL;
				goto out;
			}
			if (!INET6_TW_MATCH(sk, net, hash, saddr, daddr, ports, dif)) {
				sock_put(sk);
				goto begintw;
			}
			goto out;
		}
117
	}
118 119 120 121 122
	if (get_nulls_value(node) != slot)
		goto begintw;
	sk = NULL;
out:
	rcu_read_unlock();
123 124 125 126
	return sk;
}
EXPORT_SYMBOL(__inet6_lookup_established);

127
static inline int compute_score(struct sock *sk, struct net *net,
128 129 130 131 132 133
				const unsigned short hnum,
				const struct in6_addr *daddr,
				const int dif)
{
	int score = -1;

E
Eric Dumazet 已提交
134
	if (net_eq(sock_net(sk), net) && inet_sk(sk)->inet_num == hnum &&
135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152
	    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))
				return -1;
			score++;
		}
		if (sk->sk_bound_dev_if) {
			if (sk->sk_bound_dev_if != dif)
				return -1;
			score++;
		}
	}
	return score;
}

153 154 155
struct sock *inet6_lookup_listener(struct net *net,
		struct inet_hashinfo *hashinfo, const struct in6_addr *daddr,
		const unsigned short hnum, const int dif)
156 157
{
	struct sock *sk;
158 159 160 161 162 163 164 165 166 167 168 169 170 171 172
	const struct hlist_nulls_node *node;
	struct sock *result;
	int score, hiscore;
	unsigned int hash = inet_lhashfn(net, hnum);
	struct inet_listen_hashbucket *ilb = &hashinfo->listening_hash[hash];

	rcu_read_lock();
begin:
	result = NULL;
	hiscore = -1;
	sk_nulls_for_each(sk, node, &ilb->head) {
		score = compute_score(sk, net, hnum, daddr, dif);
		if (score > hiscore) {
			hiscore = score;
			result = sk;
173 174
		}
	}
175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191
	/*
	 * if the nulls value we got at the end of this lookup is
	 * not the expected one, we must restart lookup.
	 * We probably met an item that was moved to another chain.
	 */
	if (get_nulls_value(node) != hash + LISTENING_NULLS_BASE)
		goto begin;
	if (result) {
		if (unlikely(!atomic_inc_not_zero(&result->sk_refcnt)))
			result = NULL;
		else if (unlikely(compute_score(result, net, hnum, daddr,
				  dif) < hiscore)) {
			sock_put(result);
			goto begin;
		}
	}
	rcu_read_unlock();
192 193 194 195 196
	return result;
}

EXPORT_SYMBOL_GPL(inet6_lookup_listener);

197
struct sock *inet6_lookup(struct net *net, struct inet_hashinfo *hashinfo,
A
Al Viro 已提交
198 199
			  const struct in6_addr *saddr, const __be16 sport,
			  const struct in6_addr *daddr, const __be16 dport,
200 201 202 203 204
			  const int dif)
{
	struct sock *sk;

	local_bh_disable();
205
	sk = __inet6_lookup(net, hashinfo, saddr, sport, daddr, ntohs(dport), dif);
206 207 208 209 210 211
	local_bh_enable();

	return sk;
}

EXPORT_SYMBOL_GPL(inet6_lookup);
212 213 214 215 216 217

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;
218
	struct inet_sock *inet = inet_sk(sk);
219 220 221 222
	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;
E
Eric Dumazet 已提交
223
	const __portpair ports = INET_COMBINED_PORTS(inet->inet_dport, lport);
224 225
	struct net *net = sock_net(sk);
	const unsigned int hash = inet6_ehashfn(net, daddr, lport, saddr,
E
Eric Dumazet 已提交
226
						inet->inet_dport);
227
	struct inet_ehash_bucket *head = inet_ehash_bucket(hinfo, hash);
228
	spinlock_t *lock = inet_ehash_lockp(hinfo, hash);
229
	struct sock *sk2;
230
	const struct hlist_nulls_node *node;
231
	struct inet_timewait_sock *tw;
232
	int twrefcnt = 0;
233

234
	spin_lock(lock);
235 236

	/* Check TIME-WAIT sockets first. */
237
	sk_nulls_for_each(sk2, node, &head->twchain) {
238 239
		tw = inet_twsk(sk2);

240
		if (INET6_TW_MATCH(sk2, net, hash, saddr, daddr, ports, dif)) {
241 242 243 244 245 246 247 248 249
			if (twsk_unique(sk, sk2, twp))
				goto unique;
			else
				goto not_unique;
		}
	}
	tw = NULL;

	/* And established part... */
250
	sk_nulls_for_each(sk2, node, &head->chain) {
251
		if (INET6_MATCH(sk2, net, hash, saddr, daddr, ports, dif))
252 253 254 255
			goto not_unique;
	}

unique:
256 257
	/* Must record num and sport now. Otherwise we will see
	 * in hash table socket with a funny identity. */
E
Eric Dumazet 已提交
258 259
	inet->inet_num = lport;
	inet->inet_sport = htons(lport);
260
	sk->sk_hash = hash;
261
	WARN_ON(!sk_unhashed(sk));
262
	__sk_nulls_add_node_rcu(sk, &head->chain);
263 264 265 266
	if (tw) {
		twrefcnt = inet_twsk_unhash(tw);
		NET_INC_STATS_BH(net, LINUX_MIB_TIMEWAITRECYCLED);
	}
267
	spin_unlock(lock);
268 269
	if (twrefcnt)
		inet_twsk_put(tw);
270
	sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1);
271

272
	if (twp) {
273
		*twp = tw;
274
	} else if (tw) {
275 276 277 278 279 280 281 282
		/* Silly. Should hash-dance instead... */
		inet_twsk_deschedule(tw, death_row);

		inet_twsk_put(tw);
	}
	return 0;

not_unique:
283
	spin_unlock(lock);
284 285 286 287 288 289 290 291 292
	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,
E
Eric Dumazet 已提交
293
					  inet->inet_dport);
294 295 296 297 298
}

int inet6_hash_connect(struct inet_timewait_death_row *death_row,
		       struct sock *sk)
{
299
	return __inet_hash_connect(death_row, sk, inet6_sk_port_offset(sk),
300
			__inet6_check_established, __inet6_hash);
301 302 303
}

EXPORT_SYMBOL_GPL(inet6_hash_connect);