[linux-block.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 <linux/vmalloc.h>

#include <net/inet_connection_sock.h>
#include <net/inet_sock.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 {
	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 {
	struct net		*ib_net;
	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;
	rwlock_t			*ehash_locks;
	unsigned int			ehash_size;
	unsigned int			ehash_locks_mask;

	/* 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;

	unsigned int			bhash_size;
	/* Note : 4 bytes padding on 64 bit arches */

	/* 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)];
}

static inline rwlock_t *inet_ehash_lockp(
	struct inet_hashinfo *hashinfo,
	unsigned int hash)
{
	return &hashinfo->ehash_locks[hash & hashinfo->ehash_locks_mask];
}

static inline int inet_ehash_locks_alloc(struct inet_hashinfo *hashinfo)
{
	unsigned int i, size = 256;
#if defined(CONFIG_PROVE_LOCKING)
	unsigned int nr_pcpus = 2;
#else
	unsigned int nr_pcpus = num_possible_cpus();
#endif
	if (nr_pcpus >= 4)
		size = 512;
	if (nr_pcpus >= 8)
		size = 1024;
	if (nr_pcpus >= 16)
		size = 2048;
	if (nr_pcpus >= 32)
		size = 4096;
	if (sizeof(rwlock_t) != 0) {
#ifdef CONFIG_NUMA
		if (size * sizeof(rwlock_t) > PAGE_SIZE)
			hashinfo->ehash_locks = vmalloc(size * sizeof(rwlock_t));
		else
#endif
		hashinfo->ehash_locks =	kmalloc(size * sizeof(rwlock_t),
						GFP_KERNEL);
		if (!hashinfo->ehash_locks)
			return ENOMEM;
		for (i = 0; i < size; i++)
			rwlock_init(&hashinfo->ehash_locks[i]);
	}
	hashinfo->ehash_locks_mask = size - 1;
	return 0;
}

static inline void inet_ehash_locks_free(struct inet_hashinfo *hashinfo)
{
	if (hashinfo->ehash_locks) {
#ifdef CONFIG_NUMA
		unsigned int size = (hashinfo->ehash_locks_mask + 1) *
							sizeof(rwlock_t);
		if (size > PAGE_SIZE)
			vfree(hashinfo->ehash_locks);
		else
#endif
		kfree(hashinfo->ehash_locks);
		hashinfo->ehash_locks = NULL;
	}
}

extern struct inet_bind_bucket *
		    inet_bind_bucket_create(struct kmem_cache *cachep,
					    struct net *net,
					    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);
}

extern void __inet_hash(struct inet_hashinfo *hashinfo, struct sock *sk);
extern void __inet_hash_nolisten(struct inet_hashinfo *hinfo, struct sock *sk);

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);
		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_lockp(hashinfo, sk->sk_hash);
		write_lock_bh(lock);
	}

	if (__sk_del_node_init(sk))
		sock_prot_inuse_add(sk->sk_prot, -1);
	write_unlock_bh(lock);
out:
	if (sk->sk_state == TCP_LISTEN)
		wake_up(&hashinfo->lhash_wait);
}

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.
 */
extern struct sock * __inet_lookup_established(struct inet_hashinfo *hashinfo,
		const __be32 saddr, const __be16 sport,
		const __be32 daddr, const u16 hnum, const int dif);

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,
		int (*check_established)(struct inet_timewait_death_row *,
			struct sock *, __u16, struct inet_timewait_sock **),
		void (*hash)(struct inet_hashinfo *, struct sock *));
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>
9e4505c4	26	#include <linux/vmalloc.h>
304a1618	27
463c84b9	28	#include <net/inet_connection_sock.h>
14c85021	29	#include <net/inet_sock.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	*/
77d8bf9c ACM	40	struct inet_ehash_bucket {
77d8bf9c	41	struct hlist_head chain;
dbca9b27	42	struct hlist_head twchain;
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 {
941b1d22	77	struct net *ib_net;
77d8bf9c ACM	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	*/
77d8bf9c ACM	103	struct inet_ehash_bucket *ehash;
230140cf ED	104	rwlock_t *ehash_locks;
	105	unsigned int ehash_size;
	106	unsigned int ehash_locks_mask;
77d8bf9c ACM	107
	108	/* Ok, let's try this, I give up, we do need a local binding
	109	* TCP hash as well as the others for fast bind/connect.
	110	*/
	111	struct inet_bind_hashbucket *bhash;
	112
cfcabdcc	113	unsigned int bhash_size;
230140cf	114	/* Note : 4 bytes padding on 64 bit arches */
77d8bf9c ACM	115
	116	/* All sockets in TCP_LISTEN state will be in here. This is the only
	117	* table where wildcard'd TCP sockets can exist. Hash function here
	118	* is just local port number.
	119	*/
	120	struct hlist_head listening_hash[INET_LHTABLE_SIZE];
	121
	122	/* All the above members are written once at bootup and
	123	* never written again _or_ are predominantly read-access.
	124	*
	125	* Now align to a new cache line as all the following members
	126	* are often dirty.
	127	*/
	128	rwlock_t lhash_lock ____cacheline_aligned;
	129	atomic_t lhash_users;
	130	wait_queue_head_t lhash_wait;
e18b890b	131	struct kmem_cache *bind_bucket_cachep;
77d8bf9c ACM	132	};
77d8bf9c ACM	133
81c3d547 ED	134	static inline struct inet_ehash_bucket *inet_ehash_bucket(
	135	struct inet_hashinfo *hashinfo,
	136	unsigned int hash)
	137	{
	138	return &hashinfo->ehash[hash & (hashinfo->ehash_size - 1)];
304a1618 ACM	139	}
304a1618 ACM	140
230140cf ED	141	static inline rwlock_t *inet_ehash_lockp(
	142	struct inet_hashinfo *hashinfo,
	143	unsigned int hash)
	144	{
	145	return &hashinfo->ehash_locks[hash & hashinfo->ehash_locks_mask];
	146	}
	147
	148	static inline int inet_ehash_locks_alloc(struct inet_hashinfo *hashinfo)
	149	{
	150	unsigned int i, size = 256;
	151	#if defined(CONFIG_PROVE_LOCKING)
	152	unsigned int nr_pcpus = 2;
	153	#else
	154	unsigned int nr_pcpus = num_possible_cpus();
	155	#endif
	156	if (nr_pcpus >= 4)
	157	size = 512;
	158	if (nr_pcpus >= 8)
	159	size = 1024;
	160	if (nr_pcpus >= 16)
	161	size = 2048;
	162	if (nr_pcpus >= 32)
	163	size = 4096;
	164	if (sizeof(rwlock_t) != 0) {
	165	#ifdef CONFIG_NUMA
	166	if (size * sizeof(rwlock_t) > PAGE_SIZE)
	167	hashinfo->ehash_locks = vmalloc(size * sizeof(rwlock_t));
	168	else
	169	#endif
	170	hashinfo->ehash_locks = kmalloc(size * sizeof(rwlock_t),
	171	GFP_KERNEL);
	172	if (!hashinfo->ehash_locks)
	173	return ENOMEM;
	174	for (i = 0; i < size; i++)
	175	rwlock_init(&hashinfo->ehash_locks[i]);
	176	}
	177	hashinfo->ehash_locks_mask = size - 1;
	178	return 0;
	179	}
	180
	181	static inline void inet_ehash_locks_free(struct inet_hashinfo *hashinfo)
	182	{
	183	if (hashinfo->ehash_locks) {
	184	#ifdef CONFIG_NUMA
	185	unsigned int size = (hashinfo->ehash_locks_mask + 1) *
	186	sizeof(rwlock_t);
	187	if (size > PAGE_SIZE)
	188	vfree(hashinfo->ehash_locks);
	189	else
230140cf	190	#endif
218ad12f	191	kfree(hashinfo->ehash_locks);
230140cf ED	192	hashinfo->ehash_locks = NULL;
	193	}
	194	}
	195
77d8bf9c	196	extern struct inet_bind_bucket *
e18b890b	197	inet_bind_bucket_create(struct kmem_cache *cachep,
941b1d22	198	struct net *net,
77d8bf9c ACM	199	struct inet_bind_hashbucket *head,
77d8bf9c ACM	200	const unsigned short snum);
e18b890b	201	extern void inet_bind_bucket_destroy(struct kmem_cache *cachep,
77d8bf9c ACM	202	struct inet_bind_bucket *tb);
	203
	204	static inline int inet_bhashfn(const __u16 lport, const int bhash_size)
	205	{
	206	return lport & (bhash_size - 1);
	207	}
	208
2d8c4ce5 ACM	209	extern void inet_bind_hash(struct sock sk, struct inet_bind_bucket tb,
	210	const unsigned short snum);
	211
77d8bf9c ACM	212	/* These can have wildcards, don't try too hard. */
	213	static inline int inet_lhashfn(const unsigned short num)
	214	{
	215	return num & (INET_LHTABLE_SIZE - 1);
	216	}
	217
	218	static inline int inet_sk_listen_hashfn(const struct sock *sk)
	219	{
	220	return inet_lhashfn(inet_sk(sk)->num);
	221	}
	222
2d8c4ce5 ACM	223	/* Caller must disable local BH processing. */
	224	static inline void __inet_inherit_port(struct inet_hashinfo *table,
	225	struct sock sk, struct sock child)
	226	{
	227	const int bhash = inet_bhashfn(inet_sk(child)->num, table->bhash_size);
	228	struct inet_bind_hashbucket *head = &table->bhash[bhash];
	229	struct inet_bind_bucket *tb;
	230
	231	spin_lock(&head->lock);
463c84b9	232	tb = inet_csk(sk)->icsk_bind_hash;
2d8c4ce5	233	sk_add_bind_node(child, &tb->owners);
463c84b9	234	inet_csk(child)->icsk_bind_hash = tb;
2d8c4ce5 ACM	235	spin_unlock(&head->lock);
	236	}
	237
	238	static inline void inet_inherit_port(struct inet_hashinfo *table,
	239	struct sock sk, struct sock child)
	240	{
	241	local_bh_disable();
	242	__inet_inherit_port(table, sk, child);
	243	local_bh_enable();
	244	}
	245
	246	extern void inet_put_port(struct inet_hashinfo table, struct sock sk);
	247
f3f05f70 ACM	248	extern void inet_listen_wlock(struct inet_hashinfo *hashinfo);
	249
	250	/*
	251	* - We may sleep inside this lock.
	252	* - If sleeping is not required (or called from BH),
	253	* use plain read_(un)lock(&inet_hashinfo.lhash_lock).
	254	*/
	255	static inline void inet_listen_lock(struct inet_hashinfo *hashinfo)
	256	{
	257	/* read_lock synchronizes to candidates to writers */
	258	read_lock(&hashinfo->lhash_lock);
	259	atomic_inc(&hashinfo->lhash_users);
	260	read_unlock(&hashinfo->lhash_lock);
	261	}
	262
	263	static inline void inet_listen_unlock(struct inet_hashinfo *hashinfo)
	264	{
	265	if (atomic_dec_and_test(&hashinfo->lhash_users))
	266	wake_up(&hashinfo->lhash_wait);
	267	}
	268
152da81d PE	269	extern void __inet_hash(struct inet_hashinfo hashinfo, struct sock sk);
152da81d PE	270	extern void __inet_hash_nolisten(struct inet_hashinfo hinfo, struct sock sk);
81849d10 ACM	271
	272	static inline void inet_hash(struct inet_hashinfo hashinfo, struct sock sk)
	273	{
	274	if (sk->sk_state != TCP_CLOSE) {
	275	local_bh_disable();
152da81d	276	__inet_hash(hashinfo, sk);
81849d10 ACM	277	local_bh_enable();
	278	}
	279	}
	280
	281	static inline void inet_unhash(struct inet_hashinfo hashinfo, struct sock sk)
	282	{
	283	rwlock_t *lock;
	284
	285	if (sk_unhashed(sk))
	286	goto out;
	287
	288	if (sk->sk_state == TCP_LISTEN) {
	289	local_bh_disable();
	290	inet_listen_wlock(hashinfo);
	291	lock = &hashinfo->lhash_lock;
	292	} else {
230140cf	293	lock = inet_ehash_lockp(hashinfo, sk->sk_hash);
81c3d547	294	write_lock_bh(lock);
81849d10 ACM	295	}
	296
	297	if (__sk_del_node_init(sk))
65f76517	298	sock_prot_inuse_add(sk->sk_prot, -1);
81849d10 ACM	299	write_unlock_bh(lock);
	300	out:
	301	if (sk->sk_state == TCP_LISTEN)
	302	wake_up(&hashinfo->lhash_wait);
	303	}
33b62231	304
8f491069	305	extern struct sock __inet_lookup_listener(struct inet_hashinfo hashinfo,
fb99c848	306	const __be32 daddr,
8f491069 HX	307	const unsigned short hnum,
	308	const int dif);
	309
	310	static inline struct sock inet_lookup_listener(struct inet_hashinfo hashinfo,
fb99c848	311	__be32 daddr, __be16 dport, int dif)
8f491069 HX	312	{
	313	return __inet_lookup_listener(hashinfo, daddr, ntohs(dport), dif);
	314	}
8feaf0c0 ACM	315
8feaf0c0 ACM	316	/* Socket demux engine toys. */
4f765d84 AV	317	/* What happens here is ugly; there's a pair of adjacent fields in
	318	struct inet_sock; __be16 dport followed by __u16 num. We want to
	319	search by pair, so we combine the keys into a single 32bit value
	320	and compare with 32bit value read from &...->dport. Let's at least
	321	make sure that it's not mixed with anything else...
	322	On 64bit targets we combine comparisons with pair of adjacent __be32
	323	fields in the same way.
	324	*/
	325	typedef __u32 __bitwise __portpair;
8feaf0c0 ACM	326	#ifdef __BIG_ENDIAN
8feaf0c0 ACM	327	#define INET_COMBINED_PORTS(__sport, __dport) \
4f765d84	328	((__force __portpair)(((__force __u32)(__be16)(__sport) << 16) \| (__u32)(__dport)))
8feaf0c0 ACM	329	#else /* __LITTLE_ENDIAN */
8feaf0c0 ACM	330	#define INET_COMBINED_PORTS(__sport, __dport) \
4f765d84	331	((__force __portpair)(((__u32)(__dport) << 16) \| (__force __u32)(__be16)(__sport)))
8feaf0c0 ACM	332	#endif
	333
	334	#if (BITS_PER_LONG == 64)
4f765d84	335	typedef __u64 __bitwise __addrpair;
8feaf0c0 ACM	336	#ifdef __BIG_ENDIAN
8feaf0c0 ACM	337	#define INET_ADDR_COOKIE(__name, __saddr, __daddr) \
4f765d84 AV	338	const __addrpair __name = (__force __addrpair) ( \
	339	(((__force __u64)(__be32)(__saddr)) << 32) \| \
	340	((__force __u64)(__be32)(__daddr)));
8feaf0c0 ACM	341	#else /* __LITTLE_ENDIAN */
8feaf0c0 ACM	342	#define INET_ADDR_COOKIE(__name, __saddr, __daddr) \
4f765d84 AV	343	const __addrpair __name = (__force __addrpair) ( \
	344	(((__force __u64)(__be32)(__daddr)) << 32) \| \
	345	((__force __u64)(__be32)(__saddr)));
8feaf0c0	346	#endif /* __BIG_ENDIAN */
81c3d547 ED	347	#define INET_MATCH(__sk, __hash, __cookie, __saddr, __daddr, __ports, __dif)\
81c3d547 ED	348	(((__sk)->sk_hash == (__hash)) && \
4f765d84 AV	349	((((__addrpair )&(inet_sk(__sk)->daddr))) == (__cookie)) && \
4f765d84 AV	350	((((__portpair )&(inet_sk(__sk)->dport))) == (__ports)) && \
8feaf0c0	351	(!((__sk)->sk_bound_dev_if) \|\| ((__sk)->sk_bound_dev_if == (__dif))))
81c3d547 ED	352	#define INET_TW_MATCH(__sk, __hash, __cookie, __saddr, __daddr, __ports, __dif)\
81c3d547 ED	353	(((__sk)->sk_hash == (__hash)) && \
4f765d84 AV	354	((((__addrpair )&(inet_twsk(__sk)->tw_daddr))) == (__cookie)) && \
4f765d84 AV	355	((((__portpair )&(inet_twsk(__sk)->tw_dport))) == (__ports)) && \
8feaf0c0 ACM	356	(!((__sk)->sk_bound_dev_if) \|\| ((__sk)->sk_bound_dev_if == (__dif))))
	357	#else /* 32-bit arch */
	358	#define INET_ADDR_COOKIE(__name, __saddr, __daddr)
81c3d547 ED	359	#define INET_MATCH(__sk, __hash, __cookie, __saddr, __daddr, __ports, __dif) \
	360	(((__sk)->sk_hash == (__hash)) && \
	361	(inet_sk(__sk)->daddr == (__saddr)) && \
8feaf0c0	362	(inet_sk(__sk)->rcv_saddr == (__daddr)) && \
4f765d84	363	((((__portpair )&(inet_sk(__sk)->dport))) == (__ports)) && \
8feaf0c0	364	(!((__sk)->sk_bound_dev_if) \|\| ((__sk)->sk_bound_dev_if == (__dif))))
81c3d547 ED	365	#define INET_TW_MATCH(__sk, __hash,__cookie, __saddr, __daddr, __ports, __dif) \
	366	(((__sk)->sk_hash == (__hash)) && \
	367	(inet_twsk(__sk)->tw_daddr == (__saddr)) && \
8feaf0c0	368	(inet_twsk(__sk)->tw_rcv_saddr == (__daddr)) && \
4f765d84	369	((((__portpair )&(inet_twsk(__sk)->tw_dport))) == (__ports)) && \
8feaf0c0 ACM	370	(!((__sk)->sk_bound_dev_if) \|\| ((__sk)->sk_bound_dev_if == (__dif))))
8feaf0c0 ACM	371	#endif /* 64-bit arch */
e48c414e ACM	372
	373	/*
	374	* Sockets in TCP_CLOSE state are _always_ taken out of the hash, so we need
	375	* not check it for lookups anymore, thanks Alexey. -DaveM
	376	*
	377	* Local BH must be disabled here.
	378	*/
77a5ba55 PE	379	extern struct sock * __inet_lookup_established(struct inet_hashinfo *hashinfo,
	380	const __be32 saddr, const __be16 sport,
	381	const __be32 daddr, const u16 hnum, const int dif);
e48c414e	382
8f491069 HX	383	static inline struct sock *
8f491069 HX	384	inet_lookup_established(struct inet_hashinfo *hashinfo,
fb99c848 AV	385	const __be32 saddr, const __be16 sport,
fb99c848 AV	386	const __be32 daddr, const __be16 dport,
8f491069 HX	387	const int dif)
	388	{
	389	return __inet_lookup_established(hashinfo, saddr, sport, daddr,
	390	ntohs(dport), dif);
	391	}
	392
e48c414e	393	static inline struct sock __inet_lookup(struct inet_hashinfo hashinfo,
fb99c848 AV	394	const __be32 saddr, const __be16 sport,
fb99c848 AV	395	const __be32 daddr, const __be16 dport,
e48c414e ACM	396	const int dif)
e48c414e ACM	397	{
8f491069	398	u16 hnum = ntohs(dport);
e48c414e ACM	399	struct sock *sk = __inet_lookup_established(hashinfo, saddr, sport, daddr,
e48c414e ACM	400	hnum, dif);
8f491069	401	return sk ? : __inet_lookup_listener(hashinfo, daddr, hnum, dif);
e48c414e ACM	402	}
	403
	404	static inline struct sock inet_lookup(struct inet_hashinfo hashinfo,
fb99c848 AV	405	const __be32 saddr, const __be16 sport,
fb99c848 AV	406	const __be32 daddr, const __be16 dport,
e48c414e ACM	407	const int dif)
	408	{
	409	struct sock *sk;
	410
	411	local_bh_disable();
8f491069	412	sk = __inet_lookup(hashinfo, saddr, sport, daddr, dport, dif);
e48c414e ACM	413	local_bh_enable();
	414
	415	return sk;
	416	}
a7f5e7f1	417
5ee31fc1 PE	418	extern int __inet_hash_connect(struct inet_timewait_death_row *death_row,
	419	struct sock *sk,
	420	int (check_established)(struct inet_timewait_death_row ,
	421	struct sock , __u16, struct inet_timewait_sock *),
	422	void (hash)(struct inet_hashinfo , struct sock *));
a7f5e7f1 ACM	423	extern int inet_hash_connect(struct inet_timewait_death_row *death_row,
a7f5e7f1 ACM	424	struct sock *sk);
304a1618	425	#endif /* _INET_HASHTABLES_H */