[mirror_ubuntu-artful-kernel.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.
 * One chain is dedicated to TIME_WAIT sockets.
 * I'll experiment with dynamic table growth later.
 */
struct inet_ehash_bucket {
	rwlock_t	  lock;
	struct hlist_head chain;
	struct hlist_head twchain;
};

/* 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
	 *
	 * TIME_WAIT sockets use a separate chain (twchain).
	 */
	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;
	struct kmem_cache			*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(struct kmem_cache *cachep,
					    struct inet_bind_hashbucket *head,
					    const unsigned short snum);
extern void inet_bind_bucket_destroy(struct kmem_cache *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 __be32 daddr,
					   const unsigned short hnum,
					   const int dif);

static inline struct sock *inet_lookup_listener(struct inet_hashinfo *hashinfo,
						__be32 daddr, __be16 dport, int dif)
{
	return __inet_lookup_listener(hashinfo, daddr, ntohs(dport), dif);
}

/* Socket demux engine toys. */
/* What happens here is ugly; there's a pair of adjacent fields in
   struct inet_sock; __be16 dport followed by __u16 num.  We want to
   search by pair, so we combine the keys into a single 32bit value
   and compare with 32bit value read from &...->dport.  Let's at least
   make sure that it's not mixed with anything else...
   On 64bit targets we combine comparisons with pair of adjacent __be32
   fields in the same way.
*/
typedef __u32 __bitwise __portpair;
#ifdef __BIG_ENDIAN
#define INET_COMBINED_PORTS(__sport, __dport) \
	((__force __portpair)(((__force __u32)(__be16)(__sport) << 16) | (__u32)(__dport)))
#else /* __LITTLE_ENDIAN */
#define INET_COMBINED_PORTS(__sport, __dport) \
	((__force __portpair)(((__u32)(__dport) << 16) | (__force __u32)(__be16)(__sport)))
#endif

#if (BITS_PER_LONG == 64)
typedef __u64 __bitwise __addrpair;
#ifdef __BIG_ENDIAN
#define INET_ADDR_COOKIE(__name, __saddr, __daddr) \
	const __addrpair __name = (__force __addrpair) ( \
				   (((__force __u64)(__be32)(__saddr)) << 32) | \
				   ((__force __u64)(__be32)(__daddr)));
#else /* __LITTLE_ENDIAN */
#define INET_ADDR_COOKIE(__name, __saddr, __daddr) \
	const __addrpair __name = (__force __addrpair) ( \
				   (((__force __u64)(__be32)(__daddr)) << 32) | \
				   ((__force __u64)(__be32)(__saddr)));
#endif /* __BIG_ENDIAN */
#define INET_MATCH(__sk, __hash, __cookie, __saddr, __daddr, __ports, __dif)\
	(((__sk)->sk_hash == (__hash))				&&	\
	 ((*((__addrpair *)&(inet_sk(__sk)->daddr))) == (__cookie))	&&	\
	 ((*((__portpair *)&(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))				&&	\
	 ((*((__addrpair *)&(inet_twsk(__sk)->tw_daddr))) == (__cookie)) &&	\
	 ((*((__portpair *)&(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))		&&	\
	 ((*((__portpair *)&(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))		&&	\
	 ((*((__portpair *)&(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 __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);

	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->twchain) {
		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 __be32 saddr, const __be16 sport,
				const __be32 daddr, const __be16 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 __be32 saddr, const __be16 sport,
					 const __be32 daddr, const __be16 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 __be32 saddr, const __be16 sport,
				       const __be32 daddr, const __be16 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	36	/* This is for all connections with a full identity, no wildcards.
dbca9b27 ED	37	* One chain is dedicated to TIME_WAIT sockets.
dbca9b27 ED	38	* I'll experiment with dynamic table growth later.
77d8bf9c ACM	39	*/
	40	struct inet_ehash_bucket {
	41	rwlock_t lock;
	42	struct hlist_head chain;
dbca9b27	43	struct hlist_head twchain;
6d255361	44	};
77d8bf9c ACM	45
	46	/* There are a few simple rules, which allow for local port reuse by
	47	* an application. In essence:
	48	*
	49	* 1) Sockets bound to different interfaces may share a local port.
	50	* Failing that, goto test 2.
	51	* 2) If all sockets have sk->sk_reuse set, and none of them are in
	52	* TCP_LISTEN state, the port may be shared.
	53	* Failing that, goto test 3.
	54	* 3) If all sockets are bound to a specific inet_sk(sk)->rcv_saddr local
	55	* address, and none of them are the same, the port may be
	56	* shared.
	57	* Failing this, the port cannot be shared.
	58	*
	59	* The interesting point, is test #2. This is what an FTP server does
	60	* all day. To optimize this case we use a specific flag bit defined
	61	* below. As we add sockets to a bind bucket list, we perform a
	62	* check of: (newsk->sk_reuse && (newsk->sk_state != TCP_LISTEN))
	63	* As long as all sockets added to a bind bucket pass this test,
	64	* the flag bit will be set.
	65	* The resulting situation is that tcp_v[46]_verify_bind() can just check
	66	* for this flag bit, if it is set and the socket trying to bind has
	67	* sk->sk_reuse set, we don't even have to walk the owners list at all,
	68	* we return that it is ok to bind this socket to the requested local port.
	69	*
	70	* Sounds like a lot of work, but it is worth it. In a more naive
	71	* implementation (ie. current FreeBSD etc.) the entire list of ports
	72	* must be walked for each data port opened by an ftp server. Needless
	73	* to say, this does not scale at all. With a couple thousand FTP
	74	* users logged onto your box, isn't it nice to know that new data
	75	* ports are created in O(1) time? I thought so. ;-) -DaveM
	76	*/
	77	struct inet_bind_bucket {
	78	unsigned short port;
	79	signed short fastreuse;
	80	struct hlist_node node;
	81	struct hlist_head owners;
	82	};
	83
	84	#define inet_bind_bucket_for_each(tb, node, head) \
	85	hlist_for_each_entry(tb, node, head, node)
	86
	87	struct inet_bind_hashbucket {
	88	spinlock_t lock;
	89	struct hlist_head chain;
	90	};
	91
	92	/* This is for listening sockets, thus all sockets which possess wildcards. */
	93	#define INET_LHTABLE_SIZE 32 /* Yes, really, this is all you need. */
	94
	95	struct inet_hashinfo {
	96	/* This is for sockets with full identity only. Sockets here will
	97	* always be without wildcards and will have the following invariant:
	98	*
	99	* TCP_ESTABLISHED <= sk->sk_state < TCP_CLOSE
	100	*
dbca9b27	101	* TIME_WAIT sockets use a separate chain (twchain).
77d8bf9c ACM	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;
e18b890b	128	struct kmem_cache *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	138	extern struct inet_bind_bucket *
e18b890b	139	inet_bind_bucket_create(struct kmem_cache *cachep,
77d8bf9c ACM	140	struct inet_bind_hashbucket *head,
77d8bf9c ACM	141	const unsigned short snum);
e18b890b	142	extern void inet_bind_bucket_destroy(struct kmem_cache *cachep,
77d8bf9c ACM	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	274	extern struct sock __inet_lookup_listener(struct inet_hashinfo hashinfo,
fb99c848	275	const __be32 daddr,
8f491069 HX	276	const unsigned short hnum,
	277	const int dif);
	278
	279	static inline struct sock inet_lookup_listener(struct inet_hashinfo hashinfo,
fb99c848	280	__be32 daddr, __be16 dport, int dif)
8f491069 HX	281	{
	282	return __inet_lookup_listener(hashinfo, daddr, ntohs(dport), dif);
	283	}
8feaf0c0 ACM	284
8feaf0c0 ACM	285	/* Socket demux engine toys. */
4f765d84 AV	286	/* What happens here is ugly; there's a pair of adjacent fields in
	287	struct inet_sock; __be16 dport followed by __u16 num. We want to
	288	search by pair, so we combine the keys into a single 32bit value
	289	and compare with 32bit value read from &...->dport. Let's at least
	290	make sure that it's not mixed with anything else...
	291	On 64bit targets we combine comparisons with pair of adjacent __be32
	292	fields in the same way.
	293	*/
	294	typedef __u32 __bitwise __portpair;
8feaf0c0 ACM	295	#ifdef __BIG_ENDIAN
8feaf0c0 ACM	296	#define INET_COMBINED_PORTS(__sport, __dport) \
4f765d84	297	((__force __portpair)(((__force __u32)(__be16)(__sport) << 16) \| (__u32)(__dport)))
8feaf0c0 ACM	298	#else /* __LITTLE_ENDIAN */
8feaf0c0 ACM	299	#define INET_COMBINED_PORTS(__sport, __dport) \
4f765d84	300	((__force __portpair)(((__u32)(__dport) << 16) \| (__force __u32)(__be16)(__sport)))
8feaf0c0 ACM	301	#endif
	302
	303	#if (BITS_PER_LONG == 64)
4f765d84	304	typedef __u64 __bitwise __addrpair;
8feaf0c0 ACM	305	#ifdef __BIG_ENDIAN
8feaf0c0 ACM	306	#define INET_ADDR_COOKIE(__name, __saddr, __daddr) \
4f765d84 AV	307	const __addrpair __name = (__force __addrpair) ( \
	308	(((__force __u64)(__be32)(__saddr)) << 32) \| \
	309	((__force __u64)(__be32)(__daddr)));
8feaf0c0 ACM	310	#else /* __LITTLE_ENDIAN */
8feaf0c0 ACM	311	#define INET_ADDR_COOKIE(__name, __saddr, __daddr) \
4f765d84 AV	312	const __addrpair __name = (__force __addrpair) ( \
	313	(((__force __u64)(__be32)(__daddr)) << 32) \| \
	314	((__force __u64)(__be32)(__saddr)));
8feaf0c0	315	#endif /* __BIG_ENDIAN */
81c3d547 ED	316	#define INET_MATCH(__sk, __hash, __cookie, __saddr, __daddr, __ports, __dif)\
81c3d547 ED	317	(((__sk)->sk_hash == (__hash)) && \
4f765d84 AV	318	((((__addrpair )&(inet_sk(__sk)->daddr))) == (__cookie)) && \
4f765d84 AV	319	((((__portpair )&(inet_sk(__sk)->dport))) == (__ports)) && \
8feaf0c0	320	(!((__sk)->sk_bound_dev_if) \|\| ((__sk)->sk_bound_dev_if == (__dif))))
81c3d547 ED	321	#define INET_TW_MATCH(__sk, __hash, __cookie, __saddr, __daddr, __ports, __dif)\
81c3d547 ED	322	(((__sk)->sk_hash == (__hash)) && \
4f765d84 AV	323	((((__addrpair )&(inet_twsk(__sk)->tw_daddr))) == (__cookie)) && \
4f765d84 AV	324	((((__portpair )&(inet_twsk(__sk)->tw_dport))) == (__ports)) && \
8feaf0c0 ACM	325	(!((__sk)->sk_bound_dev_if) \|\| ((__sk)->sk_bound_dev_if == (__dif))))
	326	#else /* 32-bit arch */
	327	#define INET_ADDR_COOKIE(__name, __saddr, __daddr)
81c3d547 ED	328	#define INET_MATCH(__sk, __hash, __cookie, __saddr, __daddr, __ports, __dif) \
	329	(((__sk)->sk_hash == (__hash)) && \
	330	(inet_sk(__sk)->daddr == (__saddr)) && \
8feaf0c0	331	(inet_sk(__sk)->rcv_saddr == (__daddr)) && \
4f765d84	332	((((__portpair )&(inet_sk(__sk)->dport))) == (__ports)) && \
8feaf0c0	333	(!((__sk)->sk_bound_dev_if) \|\| ((__sk)->sk_bound_dev_if == (__dif))))
81c3d547 ED	334	#define INET_TW_MATCH(__sk, __hash,__cookie, __saddr, __daddr, __ports, __dif) \
	335	(((__sk)->sk_hash == (__hash)) && \
	336	(inet_twsk(__sk)->tw_daddr == (__saddr)) && \
8feaf0c0	337	(inet_twsk(__sk)->tw_rcv_saddr == (__daddr)) && \
4f765d84	338	((((__portpair )&(inet_twsk(__sk)->tw_dport))) == (__ports)) && \
8feaf0c0 ACM	339	(!((__sk)->sk_bound_dev_if) \|\| ((__sk)->sk_bound_dev_if == (__dif))))
8feaf0c0 ACM	340	#endif /* 64-bit arch */
e48c414e ACM	341
	342	/*
	343	* Sockets in TCP_CLOSE state are _always_ taken out of the hash, so we need
	344	* not check it for lookups anymore, thanks Alexey. -DaveM
	345	*
	346	* Local BH must be disabled here.
	347	*/
	348	static inline struct sock *
	349	__inet_lookup_established(struct inet_hashinfo *hashinfo,
fb99c848 AV	350	const __be32 saddr, const __be16 sport,
fb99c848 AV	351	const __be32 daddr, const u16 hnum,
e48c414e ACM	352	const int dif)
	353	{
	354	INET_ADDR_COOKIE(acookie, saddr, daddr)
4f765d84	355	const __portpair ports = INET_COMBINED_PORTS(sport, hnum);
e48c414e ACM	356	struct sock *sk;
	357	const struct hlist_node *node;
	358	/* Optimize here for direct hit, only listening connections can
	359	* have wildcards anyways.
	360	*/
81c3d547 ED	361	unsigned int hash = inet_ehashfn(daddr, hnum, saddr, sport);
81c3d547 ED	362	struct inet_ehash_bucket *head = inet_ehash_bucket(hashinfo, hash);
e48c414e	363
81c3d547	364	prefetch(head->chain.first);
e48c414e ACM	365	read_lock(&head->lock);
e48c414e ACM	366	sk_for_each(sk, node, &head->chain) {
81c3d547	367	if (INET_MATCH(sk, hash, acookie, saddr, daddr, ports, dif))
e48c414e ACM	368	goto hit; /* You sunk my battleship! */
	369	}
	370
	371	/* Must check for a TIME_WAIT'er before going to listener hash. */
dbca9b27	372	sk_for_each(sk, node, &head->twchain) {
81c3d547	373	if (INET_TW_MATCH(sk, hash, acookie, saddr, daddr, ports, dif))
e48c414e ACM	374	goto hit;
	375	}
	376	sk = NULL;
	377	out:
	378	read_unlock(&head->lock);
	379	return sk;
	380	hit:
	381	sock_hold(sk);
	382	goto out;
	383	}
	384
8f491069 HX	385	static inline struct sock *
8f491069 HX	386	inet_lookup_established(struct inet_hashinfo *hashinfo,
fb99c848 AV	387	const __be32 saddr, const __be16 sport,
fb99c848 AV	388	const __be32 daddr, const __be16 dport,
8f491069 HX	389	const int dif)
	390	{
	391	return __inet_lookup_established(hashinfo, saddr, sport, daddr,
	392	ntohs(dport), dif);
	393	}
	394
e48c414e	395	static inline struct sock __inet_lookup(struct inet_hashinfo hashinfo,
fb99c848 AV	396	const __be32 saddr, const __be16 sport,
fb99c848 AV	397	const __be32 daddr, const __be16 dport,
e48c414e ACM	398	const int dif)
e48c414e ACM	399	{
8f491069	400	u16 hnum = ntohs(dport);
e48c414e ACM	401	struct sock *sk = __inet_lookup_established(hashinfo, saddr, sport, daddr,
e48c414e ACM	402	hnum, dif);
8f491069	403	return sk ? : __inet_lookup_listener(hashinfo, daddr, hnum, dif);
e48c414e ACM	404	}
	405
	406	static inline struct sock inet_lookup(struct inet_hashinfo hashinfo,
fb99c848 AV	407	const __be32 saddr, const __be16 sport,
fb99c848 AV	408	const __be32 daddr, const __be16 dport,
e48c414e ACM	409	const int dif)
	410	{
	411	struct sock *sk;
	412
	413	local_bh_disable();
8f491069	414	sk = __inet_lookup(hashinfo, saddr, sport, daddr, dport, dif);
e48c414e ACM	415	local_bh_enable();
	416
	417	return sk;
	418	}
a7f5e7f1 ACM	419
	420	extern int inet_hash_connect(struct inet_timewait_death_row *death_row,
	421	struct sock *sk);
304a1618	422	#endif /* _INET_HASHTABLES_H */