[net-next-2.6.git] / include / net / inet_hashtables.h

/*
 * 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.
 *
 * 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.
 */

#ifndef _INET_HASHTABLES_H
#define _INET_HASHTABLES_H


#include <linux/interrupt.h>
#include <linux/ipv6.h>
#include <linux/list.h>
#include <linux/slab.h>
#include <linux/socket.h>
#include <linux/spinlock.h>
#include <linux/types.h>
#include <linux/wait.h>

#include <net/inet_connection_sock.h>
#include <net/inet_sock.h>
#include <net/route.h>
#include <net/sock.h>
#include <net/tcp_states.h>

#include <asm/atomic.h>
#include <asm/byteorder.h>

/* This is for all connections with a full identity, no wildcards.
 * New scheme, half the table is for TIME_WAIT, the other half is
 * for the rest.  I'll experiment with dynamic table growth later.
 */
struct inet_ehash_bucket {
	rwlock_t	  lock;
	struct hlist_head chain;
};

/* There are a few simple rules, which allow for local port reuse by
 * an application.  In essence:
 *
 *	1) Sockets bound to different interfaces may share a local port.
 *	   Failing that, goto test 2.
 *	2) If all sockets have sk->sk_reuse set, and none of them are in
 *	   TCP_LISTEN state, the port may be shared.
 *	   Failing that, goto test 3.
 *	3) If all sockets are bound to a specific inet_sk(sk)->rcv_saddr local
 *	   address, and none of them are the same, the port may be
 *	   shared.
 *	   Failing this, the port cannot be shared.
 *
 * The interesting point, is test #2.  This is what an FTP server does
 * all day.  To optimize this case we use a specific flag bit defined
 * below.  As we add sockets to a bind bucket list, we perform a
 * check of: (newsk->sk_reuse && (newsk->sk_state != TCP_LISTEN))
 * As long as all sockets added to a bind bucket pass this test,
 * the flag bit will be set.
 * The resulting situation is that tcp_v[46]_verify_bind() can just check
 * for this flag bit, if it is set and the socket trying to bind has
 * sk->sk_reuse set, we don't even have to walk the owners list at all,
 * we return that it is ok to bind this socket to the requested local port.
 *
 * Sounds like a lot of work, but it is worth it.  In a more naive
 * implementation (ie. current FreeBSD etc.) the entire list of ports
 * must be walked for each data port opened by an ftp server.  Needless
 * to say, this does not scale at all.  With a couple thousand FTP
 * users logged onto your box, isn't it nice to know that new data
 * ports are created in O(1) time?  I thought so. ;-)	-DaveM
 */
struct inet_bind_bucket {
	unsigned short		port;
	signed short		fastreuse;
	struct hlist_node	node;
	struct hlist_head	owners;
};

#define inet_bind_bucket_for_each(tb, node, head) \
	hlist_for_each_entry(tb, node, head, node)

struct inet_bind_hashbucket {
	spinlock_t		lock;
	struct hlist_head	chain;
};

/* This is for listening sockets, thus all sockets which possess wildcards. */
#define INET_LHTABLE_SIZE	32	/* Yes, really, this is all you need. */

struct inet_hashinfo {
	/* This is for sockets with full identity only.  Sockets here will
	 * always be without wildcards and will have the following invariant:
	 *
	 *          TCP_ESTABLISHED <= sk->sk_state < TCP_CLOSE
	 *
	 * First half of the table is for sockets not in TIME_WAIT, second half
	 * is for TIME_WAIT sockets only.
	 */
	struct inet_ehash_bucket	*ehash;

	/* Ok, let's try this, I give up, we do need a local binding
	 * TCP hash as well as the others for fast bind/connect.
	 */
	struct inet_bind_hashbucket	*bhash;

	int				bhash_size;
	unsigned int			ehash_size;

	/* All sockets in TCP_LISTEN state will be in here.  This is the only
	 * table where wildcard'd TCP sockets can exist.  Hash function here
	 * is just local port number.
	 */
	struct hlist_head		listening_hash[INET_LHTABLE_SIZE];

	/* All the above members are written once at bootup and
	 * never written again _or_ are predominantly read-access.
	 *
	 * Now align to a new cache line as all the following members
	 * are often dirty.
	 */
	rwlock_t			lhash_lock ____cacheline_aligned;
	atomic_t			lhash_users;
	wait_queue_head_t		lhash_wait;
	kmem_cache_t			*bind_bucket_cachep;
};

static inline struct inet_ehash_bucket *inet_ehash_bucket(
	struct inet_hashinfo *hashinfo,
	unsigned int hash)
{
	return &hashinfo->ehash[hash & (hashinfo->ehash_size - 1)];
}

extern struct inet_bind_bucket *
		    inet_bind_bucket_create(kmem_cache_t *cachep,
					    struct inet_bind_hashbucket *head,
					    const unsigned short snum);
extern void inet_bind_bucket_destroy(kmem_cache_t *cachep,
				     struct inet_bind_bucket *tb);

static inline int inet_bhashfn(const __u16 lport, const int bhash_size)
{
	return lport & (bhash_size - 1);
}

extern void inet_bind_hash(struct sock *sk, struct inet_bind_bucket *tb,
			   const unsigned short snum);

/* These can have wildcards, don't try too hard. */
static inline int inet_lhashfn(const unsigned short num)
{
	return num & (INET_LHTABLE_SIZE - 1);
}

static inline int inet_sk_listen_hashfn(const struct sock *sk)
{
	return inet_lhashfn(inet_sk(sk)->num);
}

/* Caller must disable local BH processing. */
static inline void __inet_inherit_port(struct inet_hashinfo *table,
				       struct sock *sk, struct sock *child)
{
	const int bhash = inet_bhashfn(inet_sk(child)->num, table->bhash_size);
	struct inet_bind_hashbucket *head = &table->bhash[bhash];
	struct inet_bind_bucket *tb;

	spin_lock(&head->lock);
	tb = inet_csk(sk)->icsk_bind_hash;
	sk_add_bind_node(child, &tb->owners);
	inet_csk(child)->icsk_bind_hash = tb;
	spin_unlock(&head->lock);
}

static inline void inet_inherit_port(struct inet_hashinfo *table,
				     struct sock *sk, struct sock *child)
{
	local_bh_disable();
	__inet_inherit_port(table, sk, child);
	local_bh_enable();
}

extern void inet_put_port(struct inet_hashinfo *table, struct sock *sk);

extern void inet_listen_wlock(struct inet_hashinfo *hashinfo);

/*
 * - We may sleep inside this lock.
 * - If sleeping is not required (or called from BH),
 *   use plain read_(un)lock(&inet_hashinfo.lhash_lock).
 */
static inline void inet_listen_lock(struct inet_hashinfo *hashinfo)
{
	/* read_lock synchronizes to candidates to writers */
	read_lock(&hashinfo->lhash_lock);
	atomic_inc(&hashinfo->lhash_users);
	read_unlock(&hashinfo->lhash_lock);
}

static inline void inet_listen_unlock(struct inet_hashinfo *hashinfo)
{
	if (atomic_dec_and_test(&hashinfo->lhash_users))
		wake_up(&hashinfo->lhash_wait);
}

static inline void __inet_hash(struct inet_hashinfo *hashinfo,
			       struct sock *sk, const int listen_possible)
{
	struct hlist_head *list;
	rwlock_t *lock;

	BUG_TRAP(sk_unhashed(sk));
	if (listen_possible && sk->sk_state == TCP_LISTEN) {
		list = &hashinfo->listening_hash[inet_sk_listen_hashfn(sk)];
		lock = &hashinfo->lhash_lock;
		inet_listen_wlock(hashinfo);
	} else {
		struct inet_ehash_bucket *head;
		sk->sk_hash = inet_sk_ehashfn(sk);
		head = inet_ehash_bucket(hashinfo, sk->sk_hash);
		list = &head->chain;
		lock = &head->lock;
		write_lock(lock);
	}
	__sk_add_node(sk, list);
	sock_prot_inc_use(sk->sk_prot);
	write_unlock(lock);
	if (listen_possible && sk->sk_state == TCP_LISTEN)
		wake_up(&hashinfo->lhash_wait);
}

static inline void inet_hash(struct inet_hashinfo *hashinfo, struct sock *sk)
{
	if (sk->sk_state != TCP_CLOSE) {
		local_bh_disable();
		__inet_hash(hashinfo, sk, 1);
		local_bh_enable();
	}
}

static inline void inet_unhash(struct inet_hashinfo *hashinfo, struct sock *sk)
{
	rwlock_t *lock;

	if (sk_unhashed(sk))
		goto out;

	if (sk->sk_state == TCP_LISTEN) {
		local_bh_disable();
		inet_listen_wlock(hashinfo);
		lock = &hashinfo->lhash_lock;
	} else {
		lock = &inet_ehash_bucket(hashinfo, sk->sk_hash)->lock;
		write_lock_bh(lock);
	}

	if (__sk_del_node_init(sk))
		sock_prot_dec_use(sk->sk_prot);
	write_unlock_bh(lock);
out:
	if (sk->sk_state == TCP_LISTEN)
		wake_up(&hashinfo->lhash_wait);
}

static inline int inet_iif(const struct sk_buff *skb)
{
	return ((struct rtable *)skb->dst)->rt_iif;
}

extern struct sock *__inet_lookup_listener(struct inet_hashinfo *hashinfo,
					   const u32 daddr,
					   const unsigned short hnum,
					   const int dif);

static inline struct sock *inet_lookup_listener(struct inet_hashinfo *hashinfo,
						u32 daddr, u16 dport, int dif)
{
	return __inet_lookup_listener(hashinfo, daddr, ntohs(dport), dif);
}

/* Socket demux engine toys. */
#ifdef __BIG_ENDIAN
#define INET_COMBINED_PORTS(__sport, __dport) \
	(((__u32)(__sport) << 16) | (__u32)(__dport))
#else /* __LITTLE_ENDIAN */
#define INET_COMBINED_PORTS(__sport, __dport) \
	(((__u32)(__dport) << 16) | (__u32)(__sport))
#endif

#if (BITS_PER_LONG == 64)
#ifdef __BIG_ENDIAN
#define INET_ADDR_COOKIE(__name, __saddr, __daddr) \
	const __u64 __name = (((__u64)(__saddr)) << 32) | ((__u64)(__daddr));
#else /* __LITTLE_ENDIAN */
#define INET_ADDR_COOKIE(__name, __saddr, __daddr) \
	const __u64 __name = (((__u64)(__daddr)) << 32) | ((__u64)(__saddr));
#endif /* __BIG_ENDIAN */
#define INET_MATCH(__sk, __hash, __cookie, __saddr, __daddr, __ports, __dif)\
	(((__sk)->sk_hash == (__hash))				&&	\
	 ((*((__u64 *)&(inet_sk(__sk)->daddr))) == (__cookie))	&&	\
	 ((*((__u32 *)&(inet_sk(__sk)->dport))) == (__ports))	&&	\
	 (!((__sk)->sk_bound_dev_if) || ((__sk)->sk_bound_dev_if == (__dif))))
#define INET_TW_MATCH(__sk, __hash, __cookie, __saddr, __daddr, __ports, __dif)\
	(((__sk)->sk_hash == (__hash))				&&	\
	 ((*((__u64 *)&(inet_twsk(__sk)->tw_daddr))) == (__cookie)) &&	\
	 ((*((__u32 *)&(inet_twsk(__sk)->tw_dport))) == (__ports)) &&	\
	 (!((__sk)->sk_bound_dev_if) || ((__sk)->sk_bound_dev_if == (__dif))))
#else /* 32-bit arch */
#define INET_ADDR_COOKIE(__name, __saddr, __daddr)
#define INET_MATCH(__sk, __hash, __cookie, __saddr, __daddr, __ports, __dif)	\
	(((__sk)->sk_hash == (__hash))				&&	\
	 (inet_sk(__sk)->daddr		== (__saddr))		&&	\
	 (inet_sk(__sk)->rcv_saddr	== (__daddr))		&&	\
	 ((*((__u32 *)&(inet_sk(__sk)->dport))) == (__ports))	&&	\
	 (!((__sk)->sk_bound_dev_if) || ((__sk)->sk_bound_dev_if == (__dif))))
#define INET_TW_MATCH(__sk, __hash,__cookie, __saddr, __daddr, __ports, __dif)	\
	(((__sk)->sk_hash == (__hash))				&&	\
	 (inet_twsk(__sk)->tw_daddr	== (__saddr))		&&	\
	 (inet_twsk(__sk)->tw_rcv_saddr	== (__daddr))		&&	\
	 ((*((__u32 *)&(inet_twsk(__sk)->tw_dport))) == (__ports)) &&	\
	 (!((__sk)->sk_bound_dev_if) || ((__sk)->sk_bound_dev_if == (__dif))))
#endif /* 64-bit arch */

/*
 * Sockets in TCP_CLOSE state are _always_ taken out of the hash, so we need
 * not check it for lookups anymore, thanks Alexey. -DaveM
 *
 * Local BH must be disabled here.
 */
static inline struct sock *
	__inet_lookup_established(struct inet_hashinfo *hashinfo,
				  const u32 saddr, const u16 sport,
				  const u32 daddr, const u16 hnum,
				  const int dif)
{
	INET_ADDR_COOKIE(acookie, saddr, daddr)
	const __u32 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);

	prefetch(head->chain.first);
	read_lock(&head->lock);
	sk_for_each(sk, node, &head->chain) {
		if (INET_MATCH(sk, 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 + hashinfo->ehash_size)->chain) {
		if (INET_TW_MATCH(sk, hash, acookie, saddr, daddr, ports, dif))
			goto hit;
	}
	sk = NULL;
out:
	read_unlock(&head->lock);
	return sk;
hit:
	sock_hold(sk);
	goto out;
}

static inline struct sock *
	inet_lookup_established(struct inet_hashinfo *hashinfo,
				const u32 saddr, const u16 sport,
				const u32 daddr, const u16 dport,
				const int dif)
{
	return __inet_lookup_established(hashinfo, saddr, sport, daddr,
					 ntohs(dport), dif);
}

static inline struct sock *__inet_lookup(struct inet_hashinfo *hashinfo,
					 const u32 saddr, const u16 sport,
					 const u32 daddr, const u16 dport,
					 const int dif)
{
	u16 hnum = ntohs(dport);
	struct sock *sk = __inet_lookup_established(hashinfo, saddr, sport, daddr,
						    hnum, dif);
	return sk ? : __inet_lookup_listener(hashinfo, daddr, hnum, dif);
}

static inline struct sock *inet_lookup(struct inet_hashinfo *hashinfo,
				       const u32 saddr, const u16 sport,
				       const u32 daddr, const u16 dport,
				       const int dif)
{
	struct sock *sk;

	local_bh_disable();
	sk = __inet_lookup(hashinfo, saddr, sport, daddr, dport, dif);
	local_bh_enable();

	return sk;
}

extern int inet_hash_connect(struct inet_timewait_death_row *death_row,
			     struct sock *sk);
#endif /* _INET_HASHTABLES_H */
Commit	Line	Data
304a1618 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	* Authors: Lotsa people, from code originally in tcp
	7	*
	8	* This program is free software; you can redistribute it and/or
	9	* modify it under the terms of the GNU General Public License
	10	* as published by the Free Software Foundation; either version
	11	* 2 of the License, or (at your option) any later version.
	12	*/
	13
	14	#ifndef _INET_HASHTABLES_H
	15	#define _INET_HASHTABLES_H
	16
8feaf0c0	17
2d8c4ce5	18	#include <linux/interrupt.h>
33b62231	19	#include <linux/ipv6.h>
77d8bf9c ACM	20	#include <linux/list.h>
77d8bf9c ACM	21	#include <linux/slab.h>
33b62231	22	#include <linux/socket.h>
77d8bf9c	23	#include <linux/spinlock.h>
304a1618	24	#include <linux/types.h>
f3f05f70	25	#include <linux/wait.h>
304a1618	26
463c84b9	27	#include <net/inet_connection_sock.h>
14c85021	28	#include <net/inet_sock.h>
0a5578cf	29	#include <net/route.h>
2d8c4ce5	30	#include <net/sock.h>
c752f073	31	#include <net/tcp_states.h>
2d8c4ce5	32
f3f05f70	33	#include <asm/atomic.h>
e48c414e	34	#include <asm/byteorder.h>
f3f05f70	35
77d8bf9c ACM	36	/* This is for all connections with a full identity, no wildcards.
	37	* New scheme, half the table is for TIME_WAIT, the other half is
	38	* for the rest. I'll experiment with dynamic table growth later.
	39	*/
	40	struct inet_ehash_bucket {
	41	rwlock_t lock;
	42	struct hlist_head chain;
6d255361	43	};
77d8bf9c ACM	44
	45	/* There are a few simple rules, which allow for local port reuse by
	46	* an application. In essence:
	47	*
	48	* 1) Sockets bound to different interfaces may share a local port.
	49	* Failing that, goto test 2.
	50	* 2) If all sockets have sk->sk_reuse set, and none of them are in
	51	* TCP_LISTEN state, the port may be shared.
	52	* Failing that, goto test 3.
	53	* 3) If all sockets are bound to a specific inet_sk(sk)->rcv_saddr local
	54	* address, and none of them are the same, the port may be
	55	* shared.
	56	* Failing this, the port cannot be shared.
	57	*
	58	* The interesting point, is test #2. This is what an FTP server does
	59	* all day. To optimize this case we use a specific flag bit defined
	60	* below. As we add sockets to a bind bucket list, we perform a
	61	* check of: (newsk->sk_reuse && (newsk->sk_state != TCP_LISTEN))
	62	* As long as all sockets added to a bind bucket pass this test,
	63	* the flag bit will be set.
	64	* The resulting situation is that tcp_v[46]_verify_bind() can just check
	65	* for this flag bit, if it is set and the socket trying to bind has
	66	* sk->sk_reuse set, we don't even have to walk the owners list at all,
	67	* we return that it is ok to bind this socket to the requested local port.
	68	*
	69	* Sounds like a lot of work, but it is worth it. In a more naive
	70	* implementation (ie. current FreeBSD etc.) the entire list of ports
	71	* must be walked for each data port opened by an ftp server. Needless
	72	* to say, this does not scale at all. With a couple thousand FTP
	73	* users logged onto your box, isn't it nice to know that new data
	74	* ports are created in O(1) time? I thought so. ;-) -DaveM
	75	*/
	76	struct inet_bind_bucket {
	77	unsigned short port;
	78	signed short fastreuse;
	79	struct hlist_node node;
	80	struct hlist_head owners;
	81	};
	82
	83	#define inet_bind_bucket_for_each(tb, node, head) \
	84	hlist_for_each_entry(tb, node, head, node)
	85
	86	struct inet_bind_hashbucket {
	87	spinlock_t lock;
	88	struct hlist_head chain;
	89	};
	90
	91	/* This is for listening sockets, thus all sockets which possess wildcards. */
	92	#define INET_LHTABLE_SIZE 32 /* Yes, really, this is all you need. */
	93
	94	struct inet_hashinfo {
	95	/* This is for sockets with full identity only. Sockets here will
	96	* always be without wildcards and will have the following invariant:
	97	*
	98	* TCP_ESTABLISHED <= sk->sk_state < TCP_CLOSE
	99	*
	100	* First half of the table is for sockets not in TIME_WAIT, second half
	101	* is for TIME_WAIT sockets only.
	102	*/
	103	struct inet_ehash_bucket *ehash;
	104
	105	/* Ok, let's try this, I give up, we do need a local binding
	106	* TCP hash as well as the others for fast bind/connect.
	107	*/
108	struct inet_bind_hashbucket *bhash;
109
110	int bhash_size;
81c3d547	111	unsigned int ehash_size;
77d8bf9c ACM	112
	113	/* All sockets in TCP_LISTEN state will be in here. This is the only
	114	* table where wildcard'd TCP sockets can exist. Hash function here
	115	* is just local port number.
	116	*/
	117	struct hlist_head listening_hash[INET_LHTABLE_SIZE];
	118
	119	/* All the above members are written once at bootup and
	120	* never written again _or_ are predominantly read-access.
	121	*
	122	* Now align to a new cache line as all the following members
	123	* are often dirty.
	124	*/
	125	rwlock_t lhash_lock ____cacheline_aligned;
	126	atomic_t lhash_users;
	127	wait_queue_head_t lhash_wait;
2d8c4ce5	128	kmem_cache_t *bind_bucket_cachep;
77d8bf9c ACM	129	};
77d8bf9c ACM	130
81c3d547 ED	131	static inline struct inet_ehash_bucket *inet_ehash_bucket(
	132	struct inet_hashinfo *hashinfo,
	133	unsigned int hash)
	134	{
	135	return &hashinfo->ehash[hash & (hashinfo->ehash_size - 1)];
304a1618 ACM	136	}
304a1618 ACM	137
77d8bf9c ACM	138	extern struct inet_bind_bucket *
	139	inet_bind_bucket_create(kmem_cache_t *cachep,
	140	struct inet_bind_hashbucket *head,
	141	const unsigned short snum);
	142	extern void inet_bind_bucket_destroy(kmem_cache_t *cachep,
	143	struct inet_bind_bucket *tb);
	144
	145	static inline int inet_bhashfn(const __u16 lport, const int bhash_size)
	146	{
	147	return lport & (bhash_size - 1);
	148	}
	149
2d8c4ce5 ACM	150	extern void inet_bind_hash(struct sock sk, struct inet_bind_bucket tb,
	151	const unsigned short snum);
	152
77d8bf9c ACM	153	/* These can have wildcards, don't try too hard. */
	154	static inline int inet_lhashfn(const unsigned short num)
	155	{
	156	return num & (INET_LHTABLE_SIZE - 1);
	157	}
	158
	159	static inline int inet_sk_listen_hashfn(const struct sock *sk)
	160	{
	161	return inet_lhashfn(inet_sk(sk)->num);
	162	}
	163
2d8c4ce5 ACM	164	/* Caller must disable local BH processing. */
	165	static inline void __inet_inherit_port(struct inet_hashinfo *table,
	166	struct sock sk, struct sock child)
	167	{
	168	const int bhash = inet_bhashfn(inet_sk(child)->num, table->bhash_size);
	169	struct inet_bind_hashbucket *head = &table->bhash[bhash];
	170	struct inet_bind_bucket *tb;
	171
	172	spin_lock(&head->lock);
463c84b9	173	tb = inet_csk(sk)->icsk_bind_hash;
2d8c4ce5	174	sk_add_bind_node(child, &tb->owners);
463c84b9	175	inet_csk(child)->icsk_bind_hash = tb;
2d8c4ce5 ACM	176	spin_unlock(&head->lock);
	177	}
	178
	179	static inline void inet_inherit_port(struct inet_hashinfo *table,
	180	struct sock sk, struct sock child)
	181	{
	182	local_bh_disable();
	183	__inet_inherit_port(table, sk, child);
	184	local_bh_enable();
	185	}
	186
	187	extern void inet_put_port(struct inet_hashinfo table, struct sock sk);
	188
f3f05f70 ACM	189	extern void inet_listen_wlock(struct inet_hashinfo *hashinfo);
	190
	191	/*
	192	* - We may sleep inside this lock.
	193	* - If sleeping is not required (or called from BH),
	194	* use plain read_(un)lock(&inet_hashinfo.lhash_lock).
	195	*/
	196	static inline void inet_listen_lock(struct inet_hashinfo *hashinfo)
	197	{
	198	/* read_lock synchronizes to candidates to writers */
	199	read_lock(&hashinfo->lhash_lock);
	200	atomic_inc(&hashinfo->lhash_users);
	201	read_unlock(&hashinfo->lhash_lock);
	202	}
	203
	204	static inline void inet_listen_unlock(struct inet_hashinfo *hashinfo)
	205	{
	206	if (atomic_dec_and_test(&hashinfo->lhash_users))
	207	wake_up(&hashinfo->lhash_wait);
	208	}
	209
	210	static inline void __inet_hash(struct inet_hashinfo *hashinfo,
	211	struct sock *sk, const int listen_possible)
	212	{
	213	struct hlist_head *list;
	214	rwlock_t *lock;
	215
	216	BUG_TRAP(sk_unhashed(sk));
	217	if (listen_possible && sk->sk_state == TCP_LISTEN) {
	218	list = &hashinfo->listening_hash[inet_sk_listen_hashfn(sk)];
	219	lock = &hashinfo->lhash_lock;
	220	inet_listen_wlock(hashinfo);
	221	} else {
81c3d547 ED	222	struct inet_ehash_bucket *head;
	223	sk->sk_hash = inet_sk_ehashfn(sk);
	224	head = inet_ehash_bucket(hashinfo, sk->sk_hash);
	225	list = &head->chain;
	226	lock = &head->lock;
f3f05f70 ACM	227	write_lock(lock);
	228	}
	229	__sk_add_node(sk, list);
	230	sock_prot_inc_use(sk->sk_prot);
	231	write_unlock(lock);
	232	if (listen_possible && sk->sk_state == TCP_LISTEN)
	233	wake_up(&hashinfo->lhash_wait);
	234	}
81849d10 ACM	235
	236	static inline void inet_hash(struct inet_hashinfo hashinfo, struct sock sk)
	237	{
	238	if (sk->sk_state != TCP_CLOSE) {
	239	local_bh_disable();
	240	__inet_hash(hashinfo, sk, 1);
	241	local_bh_enable();
	242	}
	243	}
	244
	245	static inline void inet_unhash(struct inet_hashinfo hashinfo, struct sock sk)
	246	{
	247	rwlock_t *lock;
	248
	249	if (sk_unhashed(sk))
	250	goto out;
	251
	252	if (sk->sk_state == TCP_LISTEN) {
	253	local_bh_disable();
	254	inet_listen_wlock(hashinfo);
	255	lock = &hashinfo->lhash_lock;
	256	} else {
81c3d547 ED	257	lock = &inet_ehash_bucket(hashinfo, sk->sk_hash)->lock;
81c3d547 ED	258	write_lock_bh(lock);
81849d10 ACM	259	}
	260
	261	if (__sk_del_node_init(sk))
	262	sock_prot_dec_use(sk->sk_prot);
	263	write_unlock_bh(lock);
	264	out:
	265	if (sk->sk_state == TCP_LISTEN)
	266	wake_up(&hashinfo->lhash_wait);
	267	}
33b62231	268
0a5578cf ACM	269	static inline int inet_iif(const struct sk_buff *skb)
	270	{
	271	return ((struct rtable *)skb->dst)->rt_iif;
	272	}
	273
8f491069 HX	274	extern struct sock __inet_lookup_listener(struct inet_hashinfo hashinfo,
	275	const u32 daddr,
	276	const unsigned short hnum,
	277	const int dif);
	278
	279	static inline struct sock inet_lookup_listener(struct inet_hashinfo hashinfo,
	280	u32 daddr, u16 dport, int dif)
	281	{
	282	return __inet_lookup_listener(hashinfo, daddr, ntohs(dport), dif);
	283	}
8feaf0c0 ACM	284
	285	/* Socket demux engine toys. */
	286	#ifdef __BIG_ENDIAN
	287	#define INET_COMBINED_PORTS(__sport, __dport) \
	288	(((__u32)(__sport) << 16) \| (__u32)(__dport))
	289	#else /* __LITTLE_ENDIAN */
	290	#define INET_COMBINED_PORTS(__sport, __dport) \
	291	(((__u32)(__dport) << 16) \| (__u32)(__sport))
	292	#endif
	293
	294	#if (BITS_PER_LONG == 64)
	295	#ifdef __BIG_ENDIAN
	296	#define INET_ADDR_COOKIE(__name, __saddr, __daddr) \
	297	const __u64 __name = (((__u64)(__saddr)) << 32) \| ((__u64)(__daddr));
	298	#else /* __LITTLE_ENDIAN */
	299	#define INET_ADDR_COOKIE(__name, __saddr, __daddr) \
	300	const __u64 __name = (((__u64)(__daddr)) << 32) \| ((__u64)(__saddr));
	301	#endif /* __BIG_ENDIAN */
81c3d547 ED	302	#define INET_MATCH(__sk, __hash, __cookie, __saddr, __daddr, __ports, __dif)\
	303	(((__sk)->sk_hash == (__hash)) && \
	304	((((__u64 )&(inet_sk(__sk)->daddr))) == (__cookie)) && \
8feaf0c0 ACM	305	((((__u32 )&(inet_sk(__sk)->dport))) == (__ports)) && \
8feaf0c0 ACM	306	(!((__sk)->sk_bound_dev_if) \|\| ((__sk)->sk_bound_dev_if == (__dif))))
81c3d547 ED	307	#define INET_TW_MATCH(__sk, __hash, __cookie, __saddr, __daddr, __ports, __dif)\
	308	(((__sk)->sk_hash == (__hash)) && \
	309	((((__u64 )&(inet_twsk(__sk)->tw_daddr))) == (__cookie)) && \
8feaf0c0 ACM	310	((((__u32 )&(inet_twsk(__sk)->tw_dport))) == (__ports)) && \
	311	(!((__sk)->sk_bound_dev_if) \|\| ((__sk)->sk_bound_dev_if == (__dif))))
	312	#else /* 32-bit arch */
	313	#define INET_ADDR_COOKIE(__name, __saddr, __daddr)
81c3d547 ED	314	#define INET_MATCH(__sk, __hash, __cookie, __saddr, __daddr, __ports, __dif) \
	315	(((__sk)->sk_hash == (__hash)) && \
	316	(inet_sk(__sk)->daddr == (__saddr)) && \
8feaf0c0 ACM	317	(inet_sk(__sk)->rcv_saddr == (__daddr)) && \
	318	((((__u32 )&(inet_sk(__sk)->dport))) == (__ports)) && \
	319	(!((__sk)->sk_bound_dev_if) \|\| ((__sk)->sk_bound_dev_if == (__dif))))
81c3d547 ED	320	#define INET_TW_MATCH(__sk, __hash,__cookie, __saddr, __daddr, __ports, __dif) \
	321	(((__sk)->sk_hash == (__hash)) && \
	322	(inet_twsk(__sk)->tw_daddr == (__saddr)) && \
8feaf0c0 ACM	323	(inet_twsk(__sk)->tw_rcv_saddr == (__daddr)) && \
	324	((((__u32 )&(inet_twsk(__sk)->tw_dport))) == (__ports)) && \
	325	(!((__sk)->sk_bound_dev_if) \|\| ((__sk)->sk_bound_dev_if == (__dif))))
	326	#endif /* 64-bit arch */
e48c414e ACM	327
	328	/*
	329	* Sockets in TCP_CLOSE state are _always_ taken out of the hash, so we need
	330	* not check it for lookups anymore, thanks Alexey. -DaveM
	331	*
	332	* Local BH must be disabled here.
	333	*/
	334	static inline struct sock *
	335	__inet_lookup_established(struct inet_hashinfo *hashinfo,
	336	const u32 saddr, const u16 sport,
	337	const u32 daddr, const u16 hnum,
	338	const int dif)
	339	{
	340	INET_ADDR_COOKIE(acookie, saddr, daddr)
	341	const __u32 ports = INET_COMBINED_PORTS(sport, hnum);
	342	struct sock *sk;
	343	const struct hlist_node *node;
	344	/* Optimize here for direct hit, only listening connections can
	345	* have wildcards anyways.
	346	*/
81c3d547 ED	347	unsigned int hash = inet_ehashfn(daddr, hnum, saddr, sport);
81c3d547 ED	348	struct inet_ehash_bucket *head = inet_ehash_bucket(hashinfo, hash);
e48c414e	349
81c3d547	350	prefetch(head->chain.first);
e48c414e ACM	351	read_lock(&head->lock);
e48c414e ACM	352	sk_for_each(sk, node, &head->chain) {
81c3d547	353	if (INET_MATCH(sk, hash, acookie, saddr, daddr, ports, dif))
e48c414e ACM	354	goto hit; /* You sunk my battleship! */
	355	}
	356
	357	/* Must check for a TIME_WAIT'er before going to listener hash. */
	358	sk_for_each(sk, node, &(head + hashinfo->ehash_size)->chain) {
81c3d547	359	if (INET_TW_MATCH(sk, hash, acookie, saddr, daddr, ports, dif))
e48c414e ACM	360	goto hit;
	361	}
	362	sk = NULL;
	363	out:
	364	read_unlock(&head->lock);
	365	return sk;
	366	hit:
	367	sock_hold(sk);
	368	goto out;
	369	}
	370
8f491069 HX	371	static inline struct sock *
	372	inet_lookup_established(struct inet_hashinfo *hashinfo,
	373	const u32 saddr, const u16 sport,
	374	const u32 daddr, const u16 dport,
	375	const int dif)
	376	{
	377	return __inet_lookup_established(hashinfo, saddr, sport, daddr,
	378	ntohs(dport), dif);
	379	}
	380
e48c414e ACM	381	static inline struct sock __inet_lookup(struct inet_hashinfo hashinfo,
e48c414e ACM	382	const u32 saddr, const u16 sport,
8f491069	383	const u32 daddr, const u16 dport,
e48c414e ACM	384	const int dif)
e48c414e ACM	385	{
8f491069	386	u16 hnum = ntohs(dport);
e48c414e ACM	387	struct sock *sk = __inet_lookup_established(hashinfo, saddr, sport, daddr,
e48c414e ACM	388	hnum, dif);
8f491069	389	return sk ? : __inet_lookup_listener(hashinfo, daddr, hnum, dif);
e48c414e ACM	390	}
	391
	392	static inline struct sock inet_lookup(struct inet_hashinfo hashinfo,
	393	const u32 saddr, const u16 sport,
	394	const u32 daddr, const u16 dport,
	395	const int dif)
	396	{
	397	struct sock *sk;
	398
	399	local_bh_disable();
8f491069	400	sk = __inet_lookup(hashinfo, saddr, sport, daddr, dport, dif);
e48c414e ACM	401	local_bh_enable();
	402
	403	return sk;
	404	}
a7f5e7f1 ACM	405
	406	extern int inet_hash_connect(struct inet_timewait_death_row *death_row,
	407	struct sock *sk);
304a1618	408	#endif /* _INET_HASHTABLES_H */