“240d5d482b69415b92ffce4b8dfca10799c9890b”上不存在“README.md”
inet6_hashtables.c 8.6 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/secure_seq.h>
24
#include <net/ip.h>
25

26
int __inet6_hash(struct sock *sk, struct inet_timewait_sock *tw)
27
{
28
	struct inet_hashinfo *hashinfo = sk->sk_prot->h.hashinfo;
29
	int twrefcnt = 0;
30

31
	WARN_ON(!sk_unhashed(sk));
32 33

	if (sk->sk_state == TCP_LISTEN) {
34
		struct inet_listen_hashbucket *ilb;
35

36 37
		ilb = &hashinfo->listening_hash[inet_sk_listen_hashfn(sk)];
		spin_lock(&ilb->lock);
38
		__sk_nulls_add_node_rcu(sk, &ilb->head);
39
		spin_unlock(&ilb->lock);
40 41
	} else {
		unsigned int hash;
42
		struct hlist_nulls_head *list;
43
		spinlock_t *lock;
44

45
		sk->sk_hash = hash = inet6_sk_ehashfn(sk);
46 47
		list = &inet_ehash_bucket(hashinfo, hash)->chain;
		lock = inet_ehash_lockp(hashinfo, hash);
48
		spin_lock(lock);
49
		__sk_nulls_add_node_rcu(sk, list);
50 51 52 53
		if (tw) {
			WARN_ON(sk->sk_hash != tw->tw_hash);
			twrefcnt = inet_twsk_unhash(tw);
		}
54
		spin_unlock(lock);
55 56
	}

57
	sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1);
58
	return twrefcnt;
59 60 61 62 63 64 65 66 67
}
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.
 */
68 69
struct sock *__inet6_lookup_established(struct net *net,
					struct inet_hashinfo *hashinfo,
70
					   const struct in6_addr *saddr,
71
					   const __be16 sport,
72 73 74 75 76
					   const struct in6_addr *daddr,
					   const u16 hnum,
					   const int dif)
{
	struct sock *sk;
77
	const struct hlist_nulls_node *node;
78
	const __portpair ports = INET_COMBINED_PORTS(sport, hnum);
79 80 81
	/* Optimize here for direct hit, only listening connections can
	 * have wildcards anyways.
	 */
82
	unsigned int hash = inet6_ehashfn(net, daddr, hnum, saddr, sport);
83
	unsigned int slot = hash & hashinfo->ehash_mask;
84
	struct inet_ehash_bucket *head = &hashinfo->ehash[slot];
85

86 87 88 89

	rcu_read_lock();
begin:
	sk_nulls_for_each_rcu(sk, node, &head->chain) {
90 91 92
		if (sk->sk_hash != hash)
			continue;
		if (likely(INET6_MATCH(sk, net, saddr, daddr, ports, dif))) {
93 94
			if (unlikely(!atomic_inc_not_zero(&sk->sk_refcnt)))
				goto begintw;
95 96
			if (unlikely(!INET6_MATCH(sk, net, saddr, daddr,
						  ports, dif))) {
97 98 99 100 101
				sock_put(sk);
				goto begin;
			}
		goto out;
		}
102
	}
103 104 105 106
	if (get_nulls_value(node) != slot)
		goto begin;

begintw:
107
	/* Must check for a TIME_WAIT'er before going to listener hash. */
108
	sk_nulls_for_each_rcu(sk, node, &head->twchain) {
109 110 111 112
		if (sk->sk_hash != hash)
			continue;
		if (likely(INET6_TW_MATCH(sk, net, saddr, daddr,
					  ports, dif))) {
113 114 115 116
			if (unlikely(!atomic_inc_not_zero(&sk->sk_refcnt))) {
				sk = NULL;
				goto out;
			}
117 118
			if (unlikely(!INET6_TW_MATCH(sk, net, saddr, daddr,
						     ports, dif))) {
119
				inet_twsk_put(inet_twsk(sk));
120 121 122 123
				goto begintw;
			}
			goto out;
		}
124
	}
125 126 127 128 129
	if (get_nulls_value(node) != slot)
		goto begintw;
	sk = NULL;
out:
	rcu_read_unlock();
130 131 132 133
	return sk;
}
EXPORT_SYMBOL(__inet6_lookup_established);

134
static inline int compute_score(struct sock *sk, struct net *net,
135 136 137 138 139 140
				const unsigned short hnum,
				const struct in6_addr *daddr,
				const int dif)
{
	int score = -1;

141
	if (net_eq(sock_net(sk), net) && inet_sk(sk)->inet_num == hnum &&
142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159
	    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;
}

160
struct sock *inet6_lookup_listener(struct net *net,
161 162
		struct inet_hashinfo *hashinfo, const struct in6_addr *saddr,
		const __be16 sport, const struct in6_addr *daddr,
163
		const unsigned short hnum, const int dif)
164 165
{
	struct sock *sk;
166 167
	const struct hlist_nulls_node *node;
	struct sock *result;
168 169
	int score, hiscore, matches = 0, reuseport = 0;
	u32 phash = 0;
170 171 172 173 174 175
	unsigned int hash = inet_lhashfn(net, hnum);
	struct inet_listen_hashbucket *ilb = &hashinfo->listening_hash[hash];

	rcu_read_lock();
begin:
	result = NULL;
176
	hiscore = 0;
177 178 179 180 181
	sk_nulls_for_each(sk, node, &ilb->head) {
		score = compute_score(sk, net, hnum, daddr, dif);
		if (score > hiscore) {
			hiscore = score;
			result = sk;
182 183 184 185 186 187 188 189 190 191 192
			reuseport = sk->sk_reuseport;
			if (reuseport) {
				phash = inet6_ehashfn(net, daddr, hnum,
						      saddr, sport);
				matches = 1;
			}
		} else if (score == hiscore && reuseport) {
			matches++;
			if (((u64)phash * matches) >> 32 == 0)
				result = sk;
			phash = next_pseudo_random32(phash);
193 194
		}
	}
195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211
	/*
	 * 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();
212 213 214 215 216
	return result;
}

EXPORT_SYMBOL_GPL(inet6_lookup_listener);

217
struct sock *inet6_lookup(struct net *net, struct inet_hashinfo *hashinfo,
218 219
			  const struct in6_addr *saddr, const __be16 sport,
			  const struct in6_addr *daddr, const __be16 dport,
220 221 222 223 224
			  const int dif)
{
	struct sock *sk;

	local_bh_disable();
225
	sk = __inet6_lookup(net, hashinfo, saddr, sport, daddr, ntohs(dport), dif);
226 227 228 229 230 231
	local_bh_enable();

	return sk;
}

EXPORT_SYMBOL_GPL(inet6_lookup);
232 233 234 235 236 237

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;
238
	struct inet_sock *inet = inet_sk(sk);
239 240 241 242
	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;
243
	const __portpair ports = INET_COMBINED_PORTS(inet->inet_dport, lport);
244 245
	struct net *net = sock_net(sk);
	const unsigned int hash = inet6_ehashfn(net, daddr, lport, saddr,
246
						inet->inet_dport);
247
	struct inet_ehash_bucket *head = inet_ehash_bucket(hinfo, hash);
248
	spinlock_t *lock = inet_ehash_lockp(hinfo, hash);
249
	struct sock *sk2;
250
	const struct hlist_nulls_node *node;
251
	struct inet_timewait_sock *tw;
252
	int twrefcnt = 0;
253

254
	spin_lock(lock);
255 256

	/* Check TIME-WAIT sockets first. */
257
	sk_nulls_for_each(sk2, node, &head->twchain) {
258 259
		if (sk2->sk_hash != hash)
			continue;
260

261 262 263
		if (likely(INET6_TW_MATCH(sk2, net, saddr, daddr,
					  ports, dif))) {
			tw = inet_twsk(sk2);
264 265 266 267 268 269 270 271 272
			if (twsk_unique(sk, sk2, twp))
				goto unique;
			else
				goto not_unique;
		}
	}
	tw = NULL;

	/* And established part... */
273
	sk_nulls_for_each(sk2, node, &head->chain) {
274 275 276
		if (sk2->sk_hash != hash)
			continue;
		if (likely(INET6_MATCH(sk2, net, saddr, daddr, ports, dif)))
277 278 279 280
			goto not_unique;
	}

unique:
281 282
	/* Must record num and sport now. Otherwise we will see
	 * in hash table socket with a funny identity. */
283 284
	inet->inet_num = lport;
	inet->inet_sport = htons(lport);
285
	sk->sk_hash = hash;
286
	WARN_ON(!sk_unhashed(sk));
287
	__sk_nulls_add_node_rcu(sk, &head->chain);
288 289 290 291
	if (tw) {
		twrefcnt = inet_twsk_unhash(tw);
		NET_INC_STATS_BH(net, LINUX_MIB_TIMEWAITRECYCLED);
	}
292
	spin_unlock(lock);
293 294
	if (twrefcnt)
		inet_twsk_put(tw);
295
	sock_prot_inuse_add(sock_net(sk), sk->sk_prot, 1);
296

297
	if (twp) {
298
		*twp = tw;
299
	} else if (tw) {
300 301 302 303 304 305 306 307
		/* Silly. Should hash-dance instead... */
		inet_twsk_deschedule(tw, death_row);

		inet_twsk_put(tw);
	}
	return 0;

not_unique:
308
	spin_unlock(lock);
309 310 311 312 313 314 315 316 317
	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,
318
					  inet->inet_dport);
319 320 321 322 323
}

int inet6_hash_connect(struct inet_timewait_death_row *death_row,
		       struct sock *sk)
{
324
	return __inet_hash_connect(death_row, sk, inet6_sk_port_offset(sk),
325
			__inet6_check_established, __inet6_hash);
326 327 328
}

EXPORT_SYMBOL_GPL(inet6_hash_connect);
新手
引导
客服 返回
顶部