[deliverable/linux.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/ip.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/route.h>
#include <net/tcp_states.h>
#include <net/netns/hash.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(struct net *net,
		const __u16 lport, const int bhash_size)
{
	return (lport + net_hash_mix(net)) & (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(struct net *net, const unsigned short num)
{
	return (num + net_hash_mix(net)) & (INET_LHTABLE_SIZE - 1);
}

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

/* Caller must disable local BH processing. */
extern void __inet_inherit_port(struct sock *sk, struct sock *child);

extern void inet_put_port(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_nolisten(struct sock *sk);
extern void inet_hash(struct sock *sk);
extern void inet_unhash(struct sock *sk);

extern struct sock *__inet_lookup_listener(struct net *net,
					   struct inet_hashinfo *hashinfo,
					   const __be32 daddr,
					   const unsigned short hnum,
					   const int dif);

static inline struct sock *inet_lookup_listener(struct net *net,
		struct inet_hashinfo *hashinfo,
		__be32 daddr, __be16 dport, int dif)
{
	return __inet_lookup_listener(net, 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, __net, __hash, __cookie, __saddr, __daddr, __ports, __dif)\
	(((__sk)->sk_hash == (__hash)) && sock_net((__sk)) == (__net)	&&	\
	 ((*((__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, __net, __hash, __cookie, __saddr, __daddr, __ports, __dif)\
	(((__sk)->sk_hash == (__hash)) && sock_net((__sk)) == (__net)	&&	\
	 ((*((__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, __net, __hash, __cookie, __saddr, __daddr, __ports, __dif)	\
	(((__sk)->sk_hash == (__hash)) && sock_net((__sk)) == (__net)	&&	\
	 (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, __net, __hash,__cookie, __saddr, __daddr, __ports, __dif)	\
	(((__sk)->sk_hash == (__hash)) && sock_net((__sk)) == (__net)	&&	\
	 (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 net *net,
		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 net *net, struct inet_hashinfo *hashinfo,
				const __be32 saddr, const __be16 sport,
				const __be32 daddr, const __be16 dport,
				const int dif)
{
	return __inet_lookup_established(net, hashinfo, saddr, sport, daddr,
					 ntohs(dport), dif);
}

static inline struct sock *__inet_lookup(struct net *net,
					 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(net, hashinfo,
				saddr, sport, daddr, hnum, dif);

	return sk ? : __inet_lookup_listener(net, hashinfo, daddr, hnum, dif);
}

static inline struct sock *inet_lookup(struct net *net,
				       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(net, hashinfo, saddr, sport, daddr, dport, dif);
	local_bh_enable();

	return sk;
}

static inline struct sock *__inet_lookup_skb(struct inet_hashinfo *hashinfo,
					     struct sk_buff *skb,
					     const __be16 sport,
					     const __be16 dport)
{
	struct sock *sk;
	const struct iphdr *iph = ip_hdr(skb);

	if (unlikely(sk = skb_steal_sock(skb)))
		return sk;
	else
		return __inet_lookup(dev_net(skb->dst->dev), hashinfo,
				     iph->saddr, sport,
				     iph->daddr, dport, inet_iif(skb));
}

extern int __inet_hash_connect(struct inet_timewait_death_row *death_row,
		struct sock *sk, u32 port_offset,
		int (*check_established)(struct inet_timewait_death_row *,
			struct sock *, __u16, struct inet_timewait_sock **),
			       void (*hash)(struct sock *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>
9a1f27c4	19	#include <linux/ip.h>
33b62231	20	#include <linux/ipv6.h>
77d8bf9c ACM	21	#include <linux/list.h>
77d8bf9c ACM	22	#include <linux/slab.h>
33b62231	23	#include <linux/socket.h>
77d8bf9c	24	#include <linux/spinlock.h>
304a1618	25	#include <linux/types.h>
f3f05f70	26	#include <linux/wait.h>
9e4505c4	27	#include <linux/vmalloc.h>
304a1618	28
463c84b9	29	#include <net/inet_connection_sock.h>
14c85021	30	#include <net/inet_sock.h>
2d8c4ce5	31	#include <net/sock.h>
9a1f27c4	32	#include <net/route.h>
c752f073	33	#include <net/tcp_states.h>
0b441916	34	#include <net/netns/hash.h>
2d8c4ce5	35
f3f05f70	36	#include <asm/atomic.h>
e48c414e	37	#include <asm/byteorder.h>
f3f05f70	38
77d8bf9c	39	/* This is for all connections with a full identity, no wildcards.
dbca9b27 ED	40	* One chain is dedicated to TIME_WAIT sockets.
dbca9b27 ED	41	* I'll experiment with dynamic table growth later.
77d8bf9c ACM	42	*/
77d8bf9c ACM	43	struct inet_ehash_bucket {
77d8bf9c	44	struct hlist_head chain;
dbca9b27	45	struct hlist_head twchain;
6d255361	46	};
77d8bf9c ACM	47
	48	/* There are a few simple rules, which allow for local port reuse by
	49	* an application. In essence:
	50	*
	51	* 1) Sockets bound to different interfaces may share a local port.
	52	* Failing that, goto test 2.
	53	* 2) If all sockets have sk->sk_reuse set, and none of them are in
	54	* TCP_LISTEN state, the port may be shared.
	55	* Failing that, goto test 3.
	56	* 3) If all sockets are bound to a specific inet_sk(sk)->rcv_saddr local
	57	* address, and none of them are the same, the port may be
	58	* shared.
	59	* Failing this, the port cannot be shared.
	60	*
	61	* The interesting point, is test #2. This is what an FTP server does
	62	* all day. To optimize this case we use a specific flag bit defined
	63	* below. As we add sockets to a bind bucket list, we perform a
	64	* check of: (newsk->sk_reuse && (newsk->sk_state != TCP_LISTEN))
	65	* As long as all sockets added to a bind bucket pass this test,
	66	* the flag bit will be set.
	67	* The resulting situation is that tcp_v[46]_verify_bind() can just check
	68	* for this flag bit, if it is set and the socket trying to bind has
	69	* sk->sk_reuse set, we don't even have to walk the owners list at all,
	70	* we return that it is ok to bind this socket to the requested local port.
	71	*
	72	* Sounds like a lot of work, but it is worth it. In a more naive
	73	* implementation (ie. current FreeBSD etc.) the entire list of ports
	74	* must be walked for each data port opened by an ftp server. Needless
	75	* to say, this does not scale at all. With a couple thousand FTP
	76	* users logged onto your box, isn't it nice to know that new data
	77	* ports are created in O(1) time? I thought so. ;-) -DaveM
	78	*/
	79	struct inet_bind_bucket {
941b1d22	80	struct net *ib_net;
77d8bf9c ACM	81	unsigned short port;
	82	signed short fastreuse;
	83	struct hlist_node node;
	84	struct hlist_head owners;
	85	};
	86
	87	#define inet_bind_bucket_for_each(tb, node, head) \
	88	hlist_for_each_entry(tb, node, head, node)
	89
	90	struct inet_bind_hashbucket {
	91	spinlock_t lock;
	92	struct hlist_head chain;
	93	};
	94
	95	/* This is for listening sockets, thus all sockets which possess wildcards. */
	96	#define INET_LHTABLE_SIZE 32 /* Yes, really, this is all you need. */
	97
	98	struct inet_hashinfo {
	99	/* This is for sockets with full identity only. Sockets here will
	100	* always be without wildcards and will have the following invariant:
	101	*
	102	* TCP_ESTABLISHED <= sk->sk_state < TCP_CLOSE
	103	*
dbca9b27	104	* TIME_WAIT sockets use a separate chain (twchain).
77d8bf9c ACM	105	*/
77d8bf9c ACM	106	struct inet_ehash_bucket *ehash;
230140cf ED	107	rwlock_t *ehash_locks;
	108	unsigned int ehash_size;
	109	unsigned int ehash_locks_mask;
77d8bf9c ACM	110
	111	/* Ok, let's try this, I give up, we do need a local binding
	112	* TCP hash as well as the others for fast bind/connect.
	113	*/
	114	struct inet_bind_hashbucket *bhash;
	115
cfcabdcc	116	unsigned int bhash_size;
230140cf	117	/* Note : 4 bytes padding on 64 bit arches */
77d8bf9c ACM	118
	119	/* All sockets in TCP_LISTEN state will be in here. This is the only
	120	* table where wildcard'd TCP sockets can exist. Hash function here
	121	* is just local port number.
	122	*/
	123	struct hlist_head listening_hash[INET_LHTABLE_SIZE];
	124
	125	/* All the above members are written once at bootup and
	126	* never written again _or_ are predominantly read-access.
	127	*
	128	* Now align to a new cache line as all the following members
	129	* are often dirty.
	130	*/
	131	rwlock_t lhash_lock ____cacheline_aligned;
	132	atomic_t lhash_users;
	133	wait_queue_head_t lhash_wait;
e18b890b	134	struct kmem_cache *bind_bucket_cachep;
77d8bf9c ACM	135	};
77d8bf9c ACM	136
81c3d547 ED	137	static inline struct inet_ehash_bucket *inet_ehash_bucket(
	138	struct inet_hashinfo *hashinfo,
	139	unsigned int hash)
	140	{
	141	return &hashinfo->ehash[hash & (hashinfo->ehash_size - 1)];
304a1618 ACM	142	}
304a1618 ACM	143
230140cf ED	144	static inline rwlock_t *inet_ehash_lockp(
	145	struct inet_hashinfo *hashinfo,
	146	unsigned int hash)
	147	{
	148	return &hashinfo->ehash_locks[hash & hashinfo->ehash_locks_mask];
	149	}
	150
	151	static inline int inet_ehash_locks_alloc(struct inet_hashinfo *hashinfo)
	152	{
	153	unsigned int i, size = 256;
	154	#if defined(CONFIG_PROVE_LOCKING)
	155	unsigned int nr_pcpus = 2;
	156	#else
	157	unsigned int nr_pcpus = num_possible_cpus();
	158	#endif
	159	if (nr_pcpus >= 4)
	160	size = 512;
	161	if (nr_pcpus >= 8)
	162	size = 1024;
	163	if (nr_pcpus >= 16)
	164	size = 2048;
	165	if (nr_pcpus >= 32)
	166	size = 4096;
	167	if (sizeof(rwlock_t) != 0) {
	168	#ifdef CONFIG_NUMA
	169	if (size * sizeof(rwlock_t) > PAGE_SIZE)
	170	hashinfo->ehash_locks = vmalloc(size * sizeof(rwlock_t));
	171	else
	172	#endif
	173	hashinfo->ehash_locks = kmalloc(size * sizeof(rwlock_t),
	174	GFP_KERNEL);
	175	if (!hashinfo->ehash_locks)
	176	return ENOMEM;
	177	for (i = 0; i < size; i++)
	178	rwlock_init(&hashinfo->ehash_locks[i]);
	179	}
	180	hashinfo->ehash_locks_mask = size - 1;
	181	return 0;
	182	}
	183
	184	static inline void inet_ehash_locks_free(struct inet_hashinfo *hashinfo)
	185	{
	186	if (hashinfo->ehash_locks) {
	187	#ifdef CONFIG_NUMA
	188	unsigned int size = (hashinfo->ehash_locks_mask + 1) *
	189	sizeof(rwlock_t);
	190	if (size > PAGE_SIZE)
	191	vfree(hashinfo->ehash_locks);
	192	else
230140cf	193	#endif
218ad12f	194	kfree(hashinfo->ehash_locks);
230140cf ED	195	hashinfo->ehash_locks = NULL;
	196	}
	197	}
	198
77d8bf9c	199	extern struct inet_bind_bucket *
e18b890b	200	inet_bind_bucket_create(struct kmem_cache *cachep,
941b1d22	201	struct net *net,
77d8bf9c ACM	202	struct inet_bind_hashbucket *head,
77d8bf9c ACM	203	const unsigned short snum);
e18b890b	204	extern void inet_bind_bucket_destroy(struct kmem_cache *cachep,
77d8bf9c ACM	205	struct inet_bind_bucket *tb);
77d8bf9c ACM	206
7f635ab7 PE	207	static inline int inet_bhashfn(struct net *net,
7f635ab7 PE	208	const __u16 lport, const int bhash_size)
77d8bf9c	209	{
0b441916	210	return (lport + net_hash_mix(net)) & (bhash_size - 1);
77d8bf9c ACM	211	}
77d8bf9c ACM	212
2d8c4ce5 ACM	213	extern void inet_bind_hash(struct sock sk, struct inet_bind_bucket tb,
	214	const unsigned short snum);
	215
77d8bf9c	216	/* These can have wildcards, don't try too hard. */
2086a650	217	static inline int inet_lhashfn(struct net *net, const unsigned short num)
77d8bf9c	218	{
0b441916	219	return (num + net_hash_mix(net)) & (INET_LHTABLE_SIZE - 1);
77d8bf9c ACM	220	}
	221
	222	static inline int inet_sk_listen_hashfn(const struct sock *sk)
	223	{
2086a650	224	return inet_lhashfn(sock_net(sk), inet_sk(sk)->num);
77d8bf9c ACM	225	}
77d8bf9c ACM	226
2d8c4ce5	227	/* Caller must disable local BH processing. */
53083773	228	extern void __inet_inherit_port(struct sock sk, struct sock child);
2d8c4ce5	229
ab1e0a13	230	extern void inet_put_port(struct sock *sk);
2d8c4ce5	231
f3f05f70 ACM	232	extern void inet_listen_wlock(struct inet_hashinfo *hashinfo);
	233
	234	/*
	235	* - We may sleep inside this lock.
	236	* - If sleeping is not required (or called from BH),
	237	* use plain read_(un)lock(&inet_hashinfo.lhash_lock).
	238	*/
	239	static inline void inet_listen_lock(struct inet_hashinfo *hashinfo)
	240	{
	241	/* read_lock synchronizes to candidates to writers */
	242	read_lock(&hashinfo->lhash_lock);
	243	atomic_inc(&hashinfo->lhash_users);
	244	read_unlock(&hashinfo->lhash_lock);
	245	}
	246
	247	static inline void inet_listen_unlock(struct inet_hashinfo *hashinfo)
	248	{
	249	if (atomic_dec_and_test(&hashinfo->lhash_users))
	250	wake_up(&hashinfo->lhash_wait);
	251	}
	252
ab1e0a13 ACM	253	extern void __inet_hash_nolisten(struct sock *sk);
	254	extern void inet_hash(struct sock *sk);
	255	extern void inet_unhash(struct sock *sk);
33b62231	256
c67499c0 PE	257	extern struct sock __inet_lookup_listener(struct net net,
c67499c0 PE	258	struct inet_hashinfo *hashinfo,
fb99c848	259	const __be32 daddr,
8f491069 HX	260	const unsigned short hnum,
	261	const int dif);
	262
c67499c0 PE	263	static inline struct sock inet_lookup_listener(struct net net,
	264	struct inet_hashinfo *hashinfo,
	265	__be32 daddr, __be16 dport, int dif)
8f491069	266	{
c67499c0	267	return __inet_lookup_listener(net, hashinfo, daddr, ntohs(dport), dif);
8f491069	268	}
8feaf0c0 ACM	269
8feaf0c0 ACM	270	/* Socket demux engine toys. */
4f765d84 AV	271	/* What happens here is ugly; there's a pair of adjacent fields in
	272	struct inet_sock; __be16 dport followed by __u16 num. We want to
	273	search by pair, so we combine the keys into a single 32bit value
	274	and compare with 32bit value read from &...->dport. Let's at least
	275	make sure that it's not mixed with anything else...
	276	On 64bit targets we combine comparisons with pair of adjacent __be32
	277	fields in the same way.
	278	*/
	279	typedef __u32 __bitwise __portpair;
8feaf0c0 ACM	280	#ifdef __BIG_ENDIAN
8feaf0c0 ACM	281	#define INET_COMBINED_PORTS(__sport, __dport) \
4f765d84	282	((__force __portpair)(((__force __u32)(__be16)(__sport) << 16) \| (__u32)(__dport)))
8feaf0c0 ACM	283	#else /* __LITTLE_ENDIAN */
8feaf0c0 ACM	284	#define INET_COMBINED_PORTS(__sport, __dport) \
4f765d84	285	((__force __portpair)(((__u32)(__dport) << 16) \| (__force __u32)(__be16)(__sport)))
8feaf0c0 ACM	286	#endif
	287
	288	#if (BITS_PER_LONG == 64)
4f765d84	289	typedef __u64 __bitwise __addrpair;
8feaf0c0 ACM	290	#ifdef __BIG_ENDIAN
8feaf0c0 ACM	291	#define INET_ADDR_COOKIE(__name, __saddr, __daddr) \
4f765d84 AV	292	const __addrpair __name = (__force __addrpair) ( \
	293	(((__force __u64)(__be32)(__saddr)) << 32) \| \
	294	((__force __u64)(__be32)(__daddr)));
8feaf0c0 ACM	295	#else /* __LITTLE_ENDIAN */
8feaf0c0 ACM	296	#define INET_ADDR_COOKIE(__name, __saddr, __daddr) \
4f765d84 AV	297	const __addrpair __name = (__force __addrpair) ( \
	298	(((__force __u64)(__be32)(__daddr)) << 32) \| \
	299	((__force __u64)(__be32)(__saddr)));
8feaf0c0	300	#endif /* __BIG_ENDIAN */
c67499c0	301	#define INET_MATCH(__sk, __net, __hash, __cookie, __saddr, __daddr, __ports, __dif)\
3b1e0a65	302	(((__sk)->sk_hash == (__hash)) && sock_net((__sk)) == (__net) && \
4f765d84 AV	303	((((__addrpair )&(inet_sk(__sk)->daddr))) == (__cookie)) && \
4f765d84 AV	304	((((__portpair )&(inet_sk(__sk)->dport))) == (__ports)) && \
8feaf0c0	305	(!((__sk)->sk_bound_dev_if) \|\| ((__sk)->sk_bound_dev_if == (__dif))))
c67499c0	306	#define INET_TW_MATCH(__sk, __net, __hash, __cookie, __saddr, __daddr, __ports, __dif)\
3b1e0a65	307	(((__sk)->sk_hash == (__hash)) && sock_net((__sk)) == (__net) && \
4f765d84 AV	308	((((__addrpair )&(inet_twsk(__sk)->tw_daddr))) == (__cookie)) && \
4f765d84 AV	309	((((__portpair )&(inet_twsk(__sk)->tw_dport))) == (__ports)) && \
8feaf0c0 ACM	310	(!((__sk)->sk_bound_dev_if) \|\| ((__sk)->sk_bound_dev_if == (__dif))))
	311	#else /* 32-bit arch */
	312	#define INET_ADDR_COOKIE(__name, __saddr, __daddr)
c67499c0	313	#define INET_MATCH(__sk, __net, __hash, __cookie, __saddr, __daddr, __ports, __dif) \
3b1e0a65	314	(((__sk)->sk_hash == (__hash)) && sock_net((__sk)) == (__net) && \
81c3d547	315	(inet_sk(__sk)->daddr == (__saddr)) && \
8feaf0c0	316	(inet_sk(__sk)->rcv_saddr == (__daddr)) && \
4f765d84	317	((((__portpair )&(inet_sk(__sk)->dport))) == (__ports)) && \
8feaf0c0	318	(!((__sk)->sk_bound_dev_if) \|\| ((__sk)->sk_bound_dev_if == (__dif))))
c67499c0	319	#define INET_TW_MATCH(__sk, __net, __hash,__cookie, __saddr, __daddr, __ports, __dif) \
3b1e0a65	320	(((__sk)->sk_hash == (__hash)) && sock_net((__sk)) == (__net) && \
81c3d547	321	(inet_twsk(__sk)->tw_daddr == (__saddr)) && \
8feaf0c0	322	(inet_twsk(__sk)->tw_rcv_saddr == (__daddr)) && \
4f765d84	323	((((__portpair )&(inet_twsk(__sk)->tw_dport))) == (__ports)) && \
8feaf0c0 ACM	324	(!((__sk)->sk_bound_dev_if) \|\| ((__sk)->sk_bound_dev_if == (__dif))))
8feaf0c0 ACM	325	#endif /* 64-bit arch */
e48c414e ACM	326
	327	/*
	328	* Sockets in TCP_CLOSE state are _always_ taken out of the hash, so we need
	329	* not check it for lookups anymore, thanks Alexey. -DaveM
	330	*
	331	* Local BH must be disabled here.
	332	*/
c67499c0 PE	333	extern struct sock * __inet_lookup_established(struct net *net,
c67499c0 PE	334	struct inet_hashinfo *hashinfo,
77a5ba55 PE	335	const __be32 saddr, const __be16 sport,
77a5ba55 PE	336	const __be32 daddr, const u16 hnum, const int dif);
e48c414e	337
8f491069	338	static inline struct sock *
c67499c0	339	inet_lookup_established(struct net net, struct inet_hashinfo hashinfo,
fb99c848 AV	340	const __be32 saddr, const __be16 sport,
fb99c848 AV	341	const __be32 daddr, const __be16 dport,
8f491069 HX	342	const int dif)
8f491069 HX	343	{
c67499c0	344	return __inet_lookup_established(net, hashinfo, saddr, sport, daddr,
8f491069 HX	345	ntohs(dport), dif);
	346	}
	347
c67499c0 PE	348	static inline struct sock __inet_lookup(struct net net,
c67499c0 PE	349	struct inet_hashinfo *hashinfo,
fb99c848 AV	350	const __be32 saddr, const __be16 sport,
fb99c848 AV	351	const __be32 daddr, const __be16 dport,
e48c414e ACM	352	const int dif)
e48c414e ACM	353	{
8f491069	354	u16 hnum = ntohs(dport);
c67499c0 PE	355	struct sock *sk = __inet_lookup_established(net, hashinfo,
	356	saddr, sport, daddr, hnum, dif);
	357
	358	return sk ? : __inet_lookup_listener(net, hashinfo, daddr, hnum, dif);
e48c414e ACM	359	}
e48c414e ACM	360
c67499c0 PE	361	static inline struct sock inet_lookup(struct net net,
c67499c0 PE	362	struct inet_hashinfo *hashinfo,
fb99c848 AV	363	const __be32 saddr, const __be16 sport,
fb99c848 AV	364	const __be32 daddr, const __be16 dport,
e48c414e ACM	365	const int dif)
	366	{
	367	struct sock *sk;
	368
	369	local_bh_disable();
c67499c0	370	sk = __inet_lookup(net, hashinfo, saddr, sport, daddr, dport, dif);
e48c414e ACM	371	local_bh_enable();
	372
	373	return sk;
	374	}
a7f5e7f1	375
9a1f27c4 ACM	376	static inline struct sock __inet_lookup_skb(struct inet_hashinfo hashinfo,
	377	struct sk_buff *skb,
	378	const __be16 sport,
	379	const __be16 dport)
	380	{
23542618	381	struct sock *sk;
9a1f27c4 ACM	382	const struct iphdr *iph = ip_hdr(skb);
9a1f27c4 ACM	383
23542618 KK	384	if (unlikely(sk = skb_steal_sock(skb)))
	385	return sk;
	386	else
	387	return __inet_lookup(dev_net(skb->dst->dev), hashinfo,
	388	iph->saddr, sport,
	389	iph->daddr, dport, inet_iif(skb));
9a1f27c4 ACM	390	}
9a1f27c4 ACM	391
5ee31fc1	392	extern int __inet_hash_connect(struct inet_timewait_death_row *death_row,
5d8c0aa9	393	struct sock *sk, u32 port_offset,
5ee31fc1 PE	394	int (check_established)(struct inet_timewait_death_row ,
5ee31fc1 PE	395	struct sock , __u16, struct inet_timewait_sock *),
ab1e0a13	396	void (hash)(struct sock sk));
a7f5e7f1 ACM	397	extern int inet_hash_connect(struct inet_timewait_death_row *death_row,
a7f5e7f1 ACM	398	struct sock *sk);
304a1618	399	#endif /* _INET_HASHTABLES_H */