[deliverable/linux.git] / net / ipv4 / inet_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 INET transport hashtables
 *
 * Authors:	Lotsa people, from code originally in tcp
 *
 *	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 <linux/sched.h>
#include <linux/slab.h>
#include <linux/wait.h>

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

/*
 * Allocate and initialize a new local port bind bucket.
 * The bindhash mutex for snum's hash chain must be held here.
 */
struct inet_bind_bucket *inet_bind_bucket_create(struct kmem_cache *cachep,
						 struct net *net,
						 struct inet_bind_hashbucket *head,
						 const unsigned short snum)
{
	struct inet_bind_bucket *tb = kmem_cache_alloc(cachep, GFP_ATOMIC);

	if (tb != NULL) {
		tb->ib_net       = net;
		tb->port      = snum;
		tb->fastreuse = 0;
		INIT_HLIST_HEAD(&tb->owners);
		hlist_add_head(&tb->node, &head->chain);
	}
	return tb;
}

/*
 * Caller must hold hashbucket lock for this tb with local BH disabled
 */
void inet_bind_bucket_destroy(struct kmem_cache *cachep, struct inet_bind_bucket *tb)
{
	if (hlist_empty(&tb->owners)) {
		__hlist_del(&tb->node);
		kmem_cache_free(cachep, tb);
	}
}

void inet_bind_hash(struct sock *sk, struct inet_bind_bucket *tb,
		    const unsigned short snum)
{
	inet_sk(sk)->num = snum;
	sk_add_bind_node(sk, &tb->owners);
	inet_csk(sk)->icsk_bind_hash = tb;
}

/*
 * Get rid of any references to a local port held by the given sock.
 */
static void __inet_put_port(struct inet_hashinfo *hashinfo, struct sock *sk)
{
	const int bhash = inet_bhashfn(inet_sk(sk)->num, hashinfo->bhash_size);
	struct inet_bind_hashbucket *head = &hashinfo->bhash[bhash];
	struct inet_bind_bucket *tb;

	spin_lock(&head->lock);
	tb = inet_csk(sk)->icsk_bind_hash;
	__sk_del_bind_node(sk);
	inet_csk(sk)->icsk_bind_hash = NULL;
	inet_sk(sk)->num = 0;
	inet_bind_bucket_destroy(hashinfo->bind_bucket_cachep, tb);
	spin_unlock(&head->lock);
}

void inet_put_port(struct inet_hashinfo *hashinfo, struct sock *sk)
{
	local_bh_disable();
	__inet_put_port(hashinfo, sk);
	local_bh_enable();
}

EXPORT_SYMBOL(inet_put_port);

/*
 * This lock without WQ_FLAG_EXCLUSIVE is good on UP and it can be very bad on SMP.
 * Look, when several writers sleep and reader wakes them up, all but one
 * immediately hit write lock and grab all the cpus. Exclusive sleep solves
 * this, _but_ remember, it adds useless work on UP machines (wake up each
 * exclusive lock release). It should be ifdefed really.
 */
void inet_listen_wlock(struct inet_hashinfo *hashinfo)
	__acquires(hashinfo->lhash_lock)
{
	write_lock(&hashinfo->lhash_lock);

	if (atomic_read(&hashinfo->lhash_users)) {
		DEFINE_WAIT(wait);

		for (;;) {
			prepare_to_wait_exclusive(&hashinfo->lhash_wait,
						  &wait, TASK_UNINTERRUPTIBLE);
			if (!atomic_read(&hashinfo->lhash_users))
				break;
			write_unlock_bh(&hashinfo->lhash_lock);
			schedule();
			write_lock_bh(&hashinfo->lhash_lock);
		}

		finish_wait(&hashinfo->lhash_wait, &wait);
	}
}

EXPORT_SYMBOL(inet_listen_wlock);

/*
 * Don't inline this cruft. Here are some nice properties to exploit here. The
 * BSD API does not allow a listening sock to specify the remote port nor the
 * remote address for the connection. So always assume those are both
 * wildcarded during the search since they can never be otherwise.
 */
static struct sock *inet_lookup_listener_slow(struct net *net,
					      const struct hlist_head *head,
					      const __be32 daddr,
					      const unsigned short hnum,
					      const int dif)
{
	struct sock *result = NULL, *sk;
	const struct hlist_node *node;
	int hiscore = -1;

	sk_for_each(sk, node, head) {
		const struct inet_sock *inet = inet_sk(sk);

		if (sk->sk_net == net && inet->num == hnum &&
				!ipv6_only_sock(sk)) {
			const __be32 rcv_saddr = inet->rcv_saddr;
			int score = sk->sk_family == PF_INET ? 1 : 0;

			if (rcv_saddr) {
				if (rcv_saddr != daddr)
					continue;
				score += 2;
			}
			if (sk->sk_bound_dev_if) {
				if (sk->sk_bound_dev_if != dif)
					continue;
				score += 2;
			}
			if (score == 5)
				return sk;
			if (score > hiscore) {
				hiscore	= score;
				result	= sk;
			}
		}
	}
	return result;
}

/* Optimize the common listener case. */
struct sock *__inet_lookup_listener(struct net *net,
				    struct inet_hashinfo *hashinfo,
				    const __be32 daddr, const unsigned short hnum,
				    const int dif)
{
	struct sock *sk = NULL;
	const struct hlist_head *head;

	read_lock(&hashinfo->lhash_lock);
	head = &hashinfo->listening_hash[inet_lhashfn(hnum)];
	if (!hlist_empty(head)) {
		const struct inet_sock *inet = inet_sk((sk = __sk_head(head)));

		if (inet->num == hnum && !sk->sk_node.next &&
		    (!inet->rcv_saddr || inet->rcv_saddr == daddr) &&
		    (sk->sk_family == PF_INET || !ipv6_only_sock(sk)) &&
		    !sk->sk_bound_dev_if && sk->sk_net == net)
			goto sherry_cache;
		sk = inet_lookup_listener_slow(net, head, daddr, hnum, dif);
	}
	if (sk) {
sherry_cache:
		sock_hold(sk);
	}
	read_unlock(&hashinfo->lhash_lock);
	return sk;
}
EXPORT_SYMBOL_GPL(__inet_lookup_listener);

struct sock * __inet_lookup_established(struct net *net,
				  struct inet_hashinfo *hashinfo,
				  const __be32 saddr, const __be16 sport,
				  const __be32 daddr, const u16 hnum,
				  const int dif)
{
	INET_ADDR_COOKIE(acookie, saddr, daddr)
	const __portpair ports = INET_COMBINED_PORTS(sport, hnum);
	struct sock *sk;
	const struct hlist_node *node;
	/* Optimize here for direct hit, only listening connections can
	 * have wildcards anyways.
	 */
	unsigned int hash = inet_ehashfn(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) {
		if (INET_MATCH(sk, net, hash, acookie,
					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 (INET_TW_MATCH(sk, net, hash, acookie,
					saddr, daddr, ports, dif))
			goto hit;
	}
	sk = NULL;
out:
	read_unlock(lock);
	return sk;
hit:
	sock_hold(sk);
	goto out;
}
EXPORT_SYMBOL_GPL(__inet_lookup_established);

/* called with local bh disabled */
static int __inet_check_established(struct inet_timewait_death_row *death_row,
				    struct sock *sk, __u16 lport,
				    struct inet_timewait_sock **twp)
{
	struct inet_hashinfo *hinfo = death_row->hashinfo;
	struct inet_sock *inet = inet_sk(sk);
	__be32 daddr = inet->rcv_saddr;
	__be32 saddr = inet->daddr;
	int dif = sk->sk_bound_dev_if;
	INET_ADDR_COOKIE(acookie, saddr, daddr)
	const __portpair ports = INET_COMBINED_PORTS(inet->dport, lport);
	unsigned int hash = inet_ehashfn(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;
	struct net *net = sk->sk_net;

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

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

		if (INET_TW_MATCH(sk2, net, hash, acookie,
					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 (INET_MATCH(sk2, net, hash, acookie,
					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);
	sk->sk_hash = hash;
	BUG_TRAP(sk_unhashed(sk));
	__sk_add_node(sk, &head->chain);
	sock_prot_inuse_add(sk->sk_prot, 1);
	write_unlock(lock);

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

		inet_twsk_put(tw);
	}

	return 0;

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

static inline u32 inet_sk_port_offset(const struct sock *sk)
{
	const struct inet_sock *inet = inet_sk(sk);
	return secure_ipv4_port_ephemeral(inet->rcv_saddr, inet->daddr,
					  inet->dport);
}

void __inet_hash_nolisten(struct inet_hashinfo *hashinfo, struct sock *sk)
{
	struct hlist_head *list;
	rwlock_t *lock;
	struct inet_ehash_bucket *head;

	BUG_TRAP(sk_unhashed(sk));

	sk->sk_hash = inet_sk_ehashfn(sk);
	head = inet_ehash_bucket(hashinfo, sk->sk_hash);
	list = &head->chain;
	lock = inet_ehash_lockp(hashinfo, sk->sk_hash);

	write_lock(lock);
	__sk_add_node(sk, list);
	sock_prot_inuse_add(sk->sk_prot, 1);
	write_unlock(lock);
}
EXPORT_SYMBOL_GPL(__inet_hash_nolisten);

void __inet_hash(struct inet_hashinfo *hashinfo, struct sock *sk)
{
	struct hlist_head *list;
	rwlock_t *lock;

	if (sk->sk_state != TCP_LISTEN) {
		__inet_hash_nolisten(hashinfo, sk);
		return;
	}

	BUG_TRAP(sk_unhashed(sk));
	list = &hashinfo->listening_hash[inet_sk_listen_hashfn(sk)];
	lock = &hashinfo->lhash_lock;

	inet_listen_wlock(hashinfo);
	__sk_add_node(sk, list);
	sock_prot_inuse_add(sk->sk_prot, 1);
	write_unlock(lock);
	wake_up(&hashinfo->lhash_wait);
}
EXPORT_SYMBOL_GPL(__inet_hash);

int __inet_hash_connect(struct inet_timewait_death_row *death_row,
		struct sock *sk,
		int (*check_established)(struct inet_timewait_death_row *,
			struct sock *, __u16, struct inet_timewait_sock **),
		void (*hash)(struct inet_hashinfo *, struct sock *))
{
	struct inet_hashinfo *hinfo = death_row->hashinfo;
	const unsigned short snum = inet_sk(sk)->num;
	struct inet_bind_hashbucket *head;
	struct inet_bind_bucket *tb;
	int ret;
	struct net *net = sk->sk_net;

	if (!snum) {
		int i, remaining, low, high, port;
		static u32 hint;
		u32 offset = hint + inet_sk_port_offset(sk);
		struct hlist_node *node;
		struct inet_timewait_sock *tw = NULL;

		inet_get_local_port_range(&low, &high);
		remaining = (high - low) + 1;

		local_bh_disable();
		for (i = 1; i <= remaining; i++) {
			port = low + (i + offset) % remaining;
			head = &hinfo->bhash[inet_bhashfn(port, hinfo->bhash_size)];
			spin_lock(&head->lock);

			/* Does not bother with rcv_saddr checks,
			 * because the established check is already
			 * unique enough.
			 */
			inet_bind_bucket_for_each(tb, node, &head->chain) {
				if (tb->ib_net == net && tb->port == port) {
					BUG_TRAP(!hlist_empty(&tb->owners));
					if (tb->fastreuse >= 0)
						goto next_port;
					if (!check_established(death_row, sk,
								port, &tw))
						goto ok;
					goto next_port;
				}
			}

			tb = inet_bind_bucket_create(hinfo->bind_bucket_cachep,
					net, head, port);
			if (!tb) {
				spin_unlock(&head->lock);
				break;
			}
			tb->fastreuse = -1;
			goto ok;

		next_port:
			spin_unlock(&head->lock);
		}
		local_bh_enable();

		return -EADDRNOTAVAIL;

ok:
		hint += i;

		/* Head lock still held and bh's disabled */
		inet_bind_hash(sk, tb, port);
		if (sk_unhashed(sk)) {
			inet_sk(sk)->sport = htons(port);
			hash(hinfo, sk);
		}
		spin_unlock(&head->lock);

		if (tw) {
			inet_twsk_deschedule(tw, death_row);
			inet_twsk_put(tw);
		}

		ret = 0;
		goto out;
	}

	head = &hinfo->bhash[inet_bhashfn(snum, hinfo->bhash_size)];
	tb  = inet_csk(sk)->icsk_bind_hash;
	spin_lock_bh(&head->lock);
	if (sk_head(&tb->owners) == sk && !sk->sk_bind_node.next) {
		hash(hinfo, sk);
		spin_unlock_bh(&head->lock);
		return 0;
	} else {
		spin_unlock(&head->lock);
		/* No definite answer... Walk to established hash table */
		ret = check_established(death_row, sk, snum, NULL);
out:
		local_bh_enable();
		return ret;
	}
}
EXPORT_SYMBOL_GPL(__inet_hash_connect);

/*
 * Bind a port for a connect operation and hash it.
 */
int inet_hash_connect(struct inet_timewait_death_row *death_row,
		      struct sock *sk)
{
	return __inet_hash_connect(death_row, sk,
			__inet_check_established, __inet_hash_nolisten);
}

EXPORT_SYMBOL_GPL(inet_hash_connect);
Commit	Line	Data
77d8bf9c 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 INET transport hashtables
	7	*
	8	* Authors: Lotsa people, from code originally in tcp
	9	*
	10	* This program is free software; you can redistribute it and/or
	11	* modify it under the terms of the GNU General Public License
	12	* as published by the Free Software Foundation; either version
	13	* 2 of the License, or (at your option) any later version.
	14	*/
	15
2d8c4ce5	16	#include <linux/module.h>
a7f5e7f1	17	#include <linux/random.h>
f3f05f70	18	#include <linux/sched.h>
77d8bf9c	19	#include <linux/slab.h>
f3f05f70	20	#include <linux/wait.h>
77d8bf9c	21
463c84b9	22	#include <net/inet_connection_sock.h>
77d8bf9c	23	#include <net/inet_hashtables.h>
a7f5e7f1	24	#include <net/ip.h>
77d8bf9c ACM	25
	26	/*
	27	* Allocate and initialize a new local port bind bucket.
	28	* The bindhash mutex for snum's hash chain must be held here.
	29	*/
e18b890b	30	struct inet_bind_bucket inet_bind_bucket_create(struct kmem_cache cachep,
941b1d22	31	struct net *net,
77d8bf9c ACM	32	struct inet_bind_hashbucket *head,
	33	const unsigned short snum)
	34	{
54e6ecb2	35	struct inet_bind_bucket *tb = kmem_cache_alloc(cachep, GFP_ATOMIC);
77d8bf9c ACM	36
77d8bf9c ACM	37	if (tb != NULL) {
941b1d22	38	tb->ib_net = net;
77d8bf9c ACM	39	tb->port = snum;
	40	tb->fastreuse = 0;
	41	INIT_HLIST_HEAD(&tb->owners);
	42	hlist_add_head(&tb->node, &head->chain);
	43	}
	44	return tb;
	45	}
	46
77d8bf9c ACM	47	/*
	48	* Caller must hold hashbucket lock for this tb with local BH disabled
	49	*/
e18b890b	50	void inet_bind_bucket_destroy(struct kmem_cache cachep, struct inet_bind_bucket tb)
77d8bf9c ACM	51	{
	52	if (hlist_empty(&tb->owners)) {
	53	__hlist_del(&tb->node);
	54	kmem_cache_free(cachep, tb);
	55	}
	56	}
2d8c4ce5 ACM	57
	58	void inet_bind_hash(struct sock sk, struct inet_bind_bucket tb,
	59	const unsigned short snum)
	60	{
463c84b9	61	inet_sk(sk)->num = snum;
2d8c4ce5	62	sk_add_bind_node(sk, &tb->owners);
463c84b9	63	inet_csk(sk)->icsk_bind_hash = tb;
2d8c4ce5 ACM	64	}
2d8c4ce5 ACM	65
2d8c4ce5 ACM	66	/*
	67	* Get rid of any references to a local port held by the given sock.
	68	*/
	69	static void __inet_put_port(struct inet_hashinfo hashinfo, struct sock sk)
	70	{
463c84b9	71	const int bhash = inet_bhashfn(inet_sk(sk)->num, hashinfo->bhash_size);
2d8c4ce5 ACM	72	struct inet_bind_hashbucket *head = &hashinfo->bhash[bhash];
	73	struct inet_bind_bucket *tb;
	74
	75	spin_lock(&head->lock);
463c84b9	76	tb = inet_csk(sk)->icsk_bind_hash;
2d8c4ce5	77	__sk_del_bind_node(sk);
463c84b9 ACM	78	inet_csk(sk)->icsk_bind_hash = NULL;
463c84b9 ACM	79	inet_sk(sk)->num = 0;
2d8c4ce5 ACM	80	inet_bind_bucket_destroy(hashinfo->bind_bucket_cachep, tb);
	81	spin_unlock(&head->lock);
	82	}
	83
	84	void inet_put_port(struct inet_hashinfo hashinfo, struct sock sk)
	85	{
	86	local_bh_disable();
	87	__inet_put_port(hashinfo, sk);
	88	local_bh_enable();
	89	}
	90
	91	EXPORT_SYMBOL(inet_put_port);
f3f05f70 ACM	92
	93	/*
	94	* This lock without WQ_FLAG_EXCLUSIVE is good on UP and it can be very bad on SMP.
	95	* Look, when several writers sleep and reader wakes them up, all but one
	96	* immediately hit write lock and grab all the cpus. Exclusive sleep solves
	97	* this, _but_ remember, it adds useless work on UP machines (wake up each
	98	* exclusive lock release). It should be ifdefed really.
	99	*/
	100	void inet_listen_wlock(struct inet_hashinfo *hashinfo)
9a429c49	101	__acquires(hashinfo->lhash_lock)
f3f05f70 ACM	102	{
	103	write_lock(&hashinfo->lhash_lock);
	104
	105	if (atomic_read(&hashinfo->lhash_users)) {
	106	DEFINE_WAIT(wait);
	107
	108	for (;;) {
	109	prepare_to_wait_exclusive(&hashinfo->lhash_wait,
	110	&wait, TASK_UNINTERRUPTIBLE);
	111	if (!atomic_read(&hashinfo->lhash_users))
	112	break;
	113	write_unlock_bh(&hashinfo->lhash_lock);
	114	schedule();
	115	write_lock_bh(&hashinfo->lhash_lock);
	116	}
	117
	118	finish_wait(&hashinfo->lhash_wait, &wait);
	119	}
	120	}
	121
	122	EXPORT_SYMBOL(inet_listen_wlock);
33b62231 ACM	123
	124	/*
	125	* Don't inline this cruft. Here are some nice properties to exploit here. The
	126	* BSD API does not allow a listening sock to specify the remote port nor the
	127	* remote address for the connection. So always assume those are both
	128	* wildcarded during the search since they can never be otherwise.
	129	*/
c67499c0 PE	130	static struct sock inet_lookup_listener_slow(struct net net,
c67499c0 PE	131	const struct hlist_head *head,
fb99c848	132	const __be32 daddr,
8f491069 HX	133	const unsigned short hnum,
8f491069 HX	134	const int dif)
33b62231 ACM	135	{
	136	struct sock result = NULL, sk;
	137	const struct hlist_node *node;
	138	int hiscore = -1;
	139
	140	sk_for_each(sk, node, head) {
	141	const struct inet_sock *inet = inet_sk(sk);
	142
c67499c0 PE	143	if (sk->sk_net == net && inet->num == hnum &&
c67499c0 PE	144	!ipv6_only_sock(sk)) {
fb99c848	145	const __be32 rcv_saddr = inet->rcv_saddr;
33b62231 ACM	146	int score = sk->sk_family == PF_INET ? 1 : 0;
	147
	148	if (rcv_saddr) {
	149	if (rcv_saddr != daddr)
	150	continue;
	151	score += 2;
	152	}
	153	if (sk->sk_bound_dev_if) {
	154	if (sk->sk_bound_dev_if != dif)
	155	continue;
	156	score += 2;
	157	}
	158	if (score == 5)
	159	return sk;
	160	if (score > hiscore) {
	161	hiscore = score;
	162	result = sk;
	163	}
	164	}
	165	}
	166	return result;
	167	}
e48c414e	168
99a92ff5	169	/* Optimize the common listener case. */
c67499c0 PE	170	struct sock __inet_lookup_listener(struct net net,
c67499c0 PE	171	struct inet_hashinfo *hashinfo,
fb99c848	172	const __be32 daddr, const unsigned short hnum,
8f491069	173	const int dif)
99a92ff5 HX	174	{
	175	struct sock *sk = NULL;
	176	const struct hlist_head *head;
	177
	178	read_lock(&hashinfo->lhash_lock);
	179	head = &hashinfo->listening_hash[inet_lhashfn(hnum)];
	180	if (!hlist_empty(head)) {
	181	const struct inet_sock *inet = inet_sk((sk = __sk_head(head)));
	182
	183	if (inet->num == hnum && !sk->sk_node.next &&
	184	(!inet->rcv_saddr \|\| inet->rcv_saddr == daddr) &&
	185	(sk->sk_family == PF_INET \|\| !ipv6_only_sock(sk)) &&
c67499c0	186	!sk->sk_bound_dev_if && sk->sk_net == net)
99a92ff5	187	goto sherry_cache;
c67499c0	188	sk = inet_lookup_listener_slow(net, head, daddr, hnum, dif);
99a92ff5 HX	189	}
	190	if (sk) {
	191	sherry_cache:
	192	sock_hold(sk);
	193	}
	194	read_unlock(&hashinfo->lhash_lock);
	195	return sk;
	196	}
8f491069	197	EXPORT_SYMBOL_GPL(__inet_lookup_listener);
a7f5e7f1	198
c67499c0 PE	199	struct sock * __inet_lookup_established(struct net *net,
c67499c0 PE	200	struct inet_hashinfo *hashinfo,
77a5ba55 PE	201	const __be32 saddr, const __be16 sport,
	202	const __be32 daddr, const u16 hnum,
	203	const int dif)
	204	{
	205	INET_ADDR_COOKIE(acookie, saddr, daddr)
	206	const __portpair ports = INET_COMBINED_PORTS(sport, hnum);
	207	struct sock *sk;
	208	const struct hlist_node *node;
	209	/* Optimize here for direct hit, only listening connections can
	210	* have wildcards anyways.
	211	*/
	212	unsigned int hash = inet_ehashfn(daddr, hnum, saddr, sport);
	213	struct inet_ehash_bucket *head = inet_ehash_bucket(hashinfo, hash);
	214	rwlock_t *lock = inet_ehash_lockp(hashinfo, hash);
	215
	216	prefetch(head->chain.first);
	217	read_lock(lock);
	218	sk_for_each(sk, node, &head->chain) {
c67499c0 PE	219	if (INET_MATCH(sk, net, hash, acookie,
c67499c0 PE	220	saddr, daddr, ports, dif))
77a5ba55 PE	221	goto hit; /* You sunk my battleship! */
	222	}
	223
	224	/* Must check for a TIME_WAIT'er before going to listener hash. */
	225	sk_for_each(sk, node, &head->twchain) {
c67499c0 PE	226	if (INET_TW_MATCH(sk, net, hash, acookie,
c67499c0 PE	227	saddr, daddr, ports, dif))
77a5ba55 PE	228	goto hit;
	229	}
	230	sk = NULL;
	231	out:
	232	read_unlock(lock);
	233	return sk;
	234	hit:
	235	sock_hold(sk);
	236	goto out;
	237	}
	238	EXPORT_SYMBOL_GPL(__inet_lookup_established);
	239
a7f5e7f1 ACM	240	/* called with local bh disabled */
	241	static int __inet_check_established(struct inet_timewait_death_row *death_row,
	242	struct sock *sk, __u16 lport,
	243	struct inet_timewait_sock **twp)
	244	{
	245	struct inet_hashinfo *hinfo = death_row->hashinfo;
	246	struct inet_sock *inet = inet_sk(sk);
fb99c848 AV	247	__be32 daddr = inet->rcv_saddr;
fb99c848 AV	248	__be32 saddr = inet->daddr;
a7f5e7f1 ACM	249	int dif = sk->sk_bound_dev_if;
a7f5e7f1 ACM	250	INET_ADDR_COOKIE(acookie, saddr, daddr)
4f765d84	251	const __portpair ports = INET_COMBINED_PORTS(inet->dport, lport);
a7f5e7f1 ACM	252	unsigned int hash = inet_ehashfn(daddr, lport, saddr, inet->dport);
a7f5e7f1 ACM	253	struct inet_ehash_bucket *head = inet_ehash_bucket(hinfo, hash);
230140cf	254	rwlock_t *lock = inet_ehash_lockp(hinfo, hash);
a7f5e7f1 ACM	255	struct sock *sk2;
	256	const struct hlist_node *node;
	257	struct inet_timewait_sock *tw;
c67499c0	258	struct net *net = sk->sk_net;
a7f5e7f1 ACM	259
a7f5e7f1 ACM	260	prefetch(head->chain.first);
230140cf	261	write_lock(lock);
a7f5e7f1 ACM	262
a7f5e7f1 ACM	263	/* Check TIME-WAIT sockets first. */
dbca9b27	264	sk_for_each(sk2, node, &head->twchain) {
a7f5e7f1 ACM	265	tw = inet_twsk(sk2);
a7f5e7f1 ACM	266
c67499c0 PE	267	if (INET_TW_MATCH(sk2, net, hash, acookie,
c67499c0 PE	268	saddr, daddr, ports, dif)) {
a7f5e7f1 ACM	269	if (twsk_unique(sk, sk2, twp))
	270	goto unique;
	271	else
	272	goto not_unique;
	273	}
	274	}
	275	tw = NULL;
	276
	277	/* And established part... */
	278	sk_for_each(sk2, node, &head->chain) {
c67499c0 PE	279	if (INET_MATCH(sk2, net, hash, acookie,
c67499c0 PE	280	saddr, daddr, ports, dif))
a7f5e7f1 ACM	281	goto not_unique;
	282	}
	283
	284	unique:
	285	/* Must record num and sport now. Otherwise we will see
	286	* in hash table socket with a funny identity. */
	287	inet->num = lport;
	288	inet->sport = htons(lport);
	289	sk->sk_hash = hash;
	290	BUG_TRAP(sk_unhashed(sk));
	291	__sk_add_node(sk, &head->chain);
65f76517	292	sock_prot_inuse_add(sk->sk_prot, 1);
230140cf	293	write_unlock(lock);
a7f5e7f1 ACM	294
	295	if (twp) {
	296	*twp = tw;
	297	NET_INC_STATS_BH(LINUX_MIB_TIMEWAITRECYCLED);
	298	} else if (tw) {
	299	/* Silly. Should hash-dance instead... */
	300	inet_twsk_deschedule(tw, death_row);
	301	NET_INC_STATS_BH(LINUX_MIB_TIMEWAITRECYCLED);
	302
	303	inet_twsk_put(tw);
	304	}
	305
	306	return 0;
	307
	308	not_unique:
230140cf	309	write_unlock(lock);
a7f5e7f1 ACM	310	return -EADDRNOTAVAIL;
	311	}
	312
	313	static inline u32 inet_sk_port_offset(const struct sock *sk)
	314	{
	315	const struct inet_sock *inet = inet_sk(sk);
e905a9ed	316	return secure_ipv4_port_ephemeral(inet->rcv_saddr, inet->daddr,
a7f5e7f1 ACM	317	inet->dport);
	318	}
	319
152da81d PE	320	void __inet_hash_nolisten(struct inet_hashinfo hashinfo, struct sock sk)
	321	{
	322	struct hlist_head *list;
	323	rwlock_t *lock;
	324	struct inet_ehash_bucket *head;
	325
	326	BUG_TRAP(sk_unhashed(sk));
	327
	328	sk->sk_hash = inet_sk_ehashfn(sk);
	329	head = inet_ehash_bucket(hashinfo, sk->sk_hash);
	330	list = &head->chain;
	331	lock = inet_ehash_lockp(hashinfo, sk->sk_hash);
	332
	333	write_lock(lock);
	334	__sk_add_node(sk, list);
65f76517	335	sock_prot_inuse_add(sk->sk_prot, 1);
152da81d PE	336	write_unlock(lock);
	337	}
	338	EXPORT_SYMBOL_GPL(__inet_hash_nolisten);
	339
	340	void __inet_hash(struct inet_hashinfo hashinfo, struct sock sk)
	341	{
	342	struct hlist_head *list;
	343	rwlock_t *lock;
	344
	345	if (sk->sk_state != TCP_LISTEN) {
	346	__inet_hash_nolisten(hashinfo, sk);
	347	return;
	348	}
	349
	350	BUG_TRAP(sk_unhashed(sk));
	351	list = &hashinfo->listening_hash[inet_sk_listen_hashfn(sk)];
	352	lock = &hashinfo->lhash_lock;
	353
	354	inet_listen_wlock(hashinfo);
	355	__sk_add_node(sk, list);
65f76517	356	sock_prot_inuse_add(sk->sk_prot, 1);
152da81d PE	357	write_unlock(lock);
	358	wake_up(&hashinfo->lhash_wait);
	359	}
	360	EXPORT_SYMBOL_GPL(__inet_hash);
	361
5ee31fc1 PE	362	int __inet_hash_connect(struct inet_timewait_death_row *death_row,
	363	struct sock *sk,
	364	int (check_established)(struct inet_timewait_death_row ,
	365	struct sock , __u16, struct inet_timewait_sock *),
	366	void (hash)(struct inet_hashinfo , struct sock *))
a7f5e7f1 ACM	367	{
	368	struct inet_hashinfo *hinfo = death_row->hashinfo;
	369	const unsigned short snum = inet_sk(sk)->num;
e905a9ed YH	370	struct inet_bind_hashbucket *head;
e905a9ed YH	371	struct inet_bind_bucket *tb;
a7f5e7f1	372	int ret;
941b1d22	373	struct net *net = sk->sk_net;
a7f5e7f1	374
e905a9ed	375	if (!snum) {
227b60f5	376	int i, remaining, low, high, port;
a7f5e7f1 ACM	377	static u32 hint;
	378	u32 offset = hint + inet_sk_port_offset(sk);
	379	struct hlist_node *node;
e905a9ed	380	struct inet_timewait_sock *tw = NULL;
a7f5e7f1	381
227b60f5	382	inet_get_local_port_range(&low, &high);
a25de534	383	remaining = (high - low) + 1;
227b60f5	384
e905a9ed	385	local_bh_disable();
227b60f5 SH	386	for (i = 1; i <= remaining; i++) {
227b60f5 SH	387	port = low + (i + offset) % remaining;
e905a9ed YH	388	head = &hinfo->bhash[inet_bhashfn(port, hinfo->bhash_size)];
e905a9ed YH	389	spin_lock(&head->lock);
a7f5e7f1	390
e905a9ed YH	391	/* Does not bother with rcv_saddr checks,
	392	* because the established check is already
	393	* unique enough.
	394	*/
a7f5e7f1	395	inet_bind_bucket_for_each(tb, node, &head->chain) {
941b1d22	396	if (tb->ib_net == net && tb->port == port) {
e905a9ed YH	397	BUG_TRAP(!hlist_empty(&tb->owners));
	398	if (tb->fastreuse >= 0)
	399	goto next_port;
5ee31fc1 PE	400	if (!check_established(death_row, sk,
5ee31fc1 PE	401	port, &tw))
e905a9ed YH	402	goto ok;
	403	goto next_port;
	404	}
	405	}
	406
941b1d22 PE	407	tb = inet_bind_bucket_create(hinfo->bind_bucket_cachep,
941b1d22 PE	408	net, head, port);
e905a9ed YH	409	if (!tb) {
	410	spin_unlock(&head->lock);
	411	break;
	412	}
	413	tb->fastreuse = -1;
	414	goto ok;
	415
	416	next_port:
	417	spin_unlock(&head->lock);
	418	}
	419	local_bh_enable();
	420
	421	return -EADDRNOTAVAIL;
a7f5e7f1 ACM	422
	423	ok:
	424	hint += i;
	425
e905a9ed YH	426	/* Head lock still held and bh's disabled */
e905a9ed YH	427	inet_bind_hash(sk, tb, port);
a7f5e7f1	428	if (sk_unhashed(sk)) {
e905a9ed	429	inet_sk(sk)->sport = htons(port);
5ee31fc1	430	hash(hinfo, sk);
e905a9ed YH	431	}
e905a9ed YH	432	spin_unlock(&head->lock);
a7f5e7f1	433
e905a9ed YH	434	if (tw) {
	435	inet_twsk_deschedule(tw, death_row);
	436	inet_twsk_put(tw);
	437	}
a7f5e7f1 ACM	438
	439	ret = 0;
	440	goto out;
e905a9ed	441	}
a7f5e7f1	442
e905a9ed YH	443	head = &hinfo->bhash[inet_bhashfn(snum, hinfo->bhash_size)];
e905a9ed YH	444	tb = inet_csk(sk)->icsk_bind_hash;
a7f5e7f1 ACM	445	spin_lock_bh(&head->lock);
a7f5e7f1 ACM	446	if (sk_head(&tb->owners) == sk && !sk->sk_bind_node.next) {
5ee31fc1	447	hash(hinfo, sk);
a7f5e7f1 ACM	448	spin_unlock_bh(&head->lock);
	449	return 0;
	450	} else {
	451	spin_unlock(&head->lock);
	452	/* No definite answer... Walk to established hash table */
5ee31fc1	453	ret = check_established(death_row, sk, snum, NULL);
a7f5e7f1 ACM	454	out:
	455	local_bh_enable();
	456	return ret;
	457	}
	458	}
5ee31fc1 PE	459	EXPORT_SYMBOL_GPL(__inet_hash_connect);
	460
	461	/*
	462	* Bind a port for a connect operation and hash it.
	463	*/
	464	int inet_hash_connect(struct inet_timewait_death_row *death_row,
	465	struct sock *sk)
	466	{
	467	return __inet_hash_connect(death_row, sk,
	468	__inet_check_established, __inet_hash_nolisten);
	469	}
a7f5e7f1 ACM	470
a7f5e7f1 ACM	471	EXPORT_SYMBOL_GPL(inet_hash_connect);