[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/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 <linux/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_nulls_head chain;
	struct hlist_nulls_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 {
#ifdef CONFIG_NET_NS
	struct net		*ib_net;
#endif
	unsigned short		port;
	signed short		fastreuse;
	int			num_owners;
	struct hlist_node	node;
	struct hlist_head	owners;
};

static inline struct net *ib_net(struct inet_bind_bucket *ib)
{
	return read_pnet(&ib->ib_net);
}

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

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

/*
 * Sockets can be hashed in established or listening table
 * We must use different 'nulls' end-of-chain value for listening
 * hash table, or we might find a socket that was closed and
 * reallocated/inserted into established hash table
 */
#define LISTENING_NULLS_BASE (1U << 29)
struct inet_listen_hashbucket {
	spinlock_t		lock;
	struct hlist_nulls_head	head;
};

/* 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;
	spinlock_t			*ehash_locks;
	unsigned int			ehash_mask;
	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;
	/* 4 bytes hole on 64 bit */

	struct kmem_cache		*bind_bucket_cachep;

	/* 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
	 * might be often dirty.
	 */
	/* 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 inet_listen_hashbucket	listening_hash[INET_LHTABLE_SIZE]
					____cacheline_aligned_in_smp;

	atomic_t			bsockets;
};

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

static inline spinlock_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(spinlock_t) != 0) {
#ifdef CONFIG_NUMA
		if (size * sizeof(spinlock_t) > PAGE_SIZE)
			hashinfo->ehash_locks = vmalloc(size * sizeof(spinlock_t));
		else
#endif
		hashinfo->ehash_locks =	kmalloc(size * sizeof(spinlock_t),
						GFP_KERNEL);
		if (!hashinfo->ehash_locks)
			return ENOMEM;
		for (i = 0; i < size; i++)
			spin_lock_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(spinlock_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)->inet_num);
}

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

extern void inet_put_port(struct sock *sk);

void inet_hashinfo_init(struct inet_hashinfo *h);

extern int __inet_hash_nolisten(struct sock *sk, struct inet_timewait_sock *tw);
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.
*/
#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)
#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, __cookie, __saddr, __daddr, __ports, __dif)	\
	((inet_sk(__sk)->inet_portpair == (__ports))		&&	\
	 (inet_sk(__sk)->inet_addrpair == (__cookie))		&&	\
	 (!(__sk)->sk_bound_dev_if	||				\
	   ((__sk)->sk_bound_dev_if == (__dif))) 		&& 	\
	 net_eq(sock_net(__sk), (__net)))
#define INET_TW_MATCH(__sk, __net, __cookie, __saddr, __daddr, __ports, __dif)\
	((inet_twsk(__sk)->tw_portpair == (__ports))	&&		\
	 (inet_twsk(__sk)->tw_addrpair == (__cookie))	&&		\
	 (!(__sk)->sk_bound_dev_if	||				\
	   ((__sk)->sk_bound_dev_if == (__dif)))	&&		\
	 net_eq(sock_net(__sk), (__net)))
#else /* 32-bit arch */
#define INET_ADDR_COOKIE(__name, __saddr, __daddr)
#define INET_MATCH(__sk, __net, __cookie, __saddr, __daddr, __ports, __dif) \
	((inet_sk(__sk)->inet_portpair == (__ports))	&&		\
	 (inet_sk(__sk)->inet_daddr	== (__saddr))	&&		\
	 (inet_sk(__sk)->inet_rcv_saddr	== (__daddr))	&&		\
	 (!(__sk)->sk_bound_dev_if	||				\
	   ((__sk)->sk_bound_dev_if == (__dif))) 	&&		\
	 net_eq(sock_net(__sk), (__net)))
#define INET_TW_MATCH(__sk, __net, __cookie, __saddr, __daddr, __ports, __dif) \
	((inet_twsk(__sk)->tw_portpair == (__ports))	&&		\
	 (inet_twsk(__sk)->tw_daddr	== (__saddr))	&&		\
	 (inet_twsk(__sk)->tw_rcv_saddr	== (__daddr))	&&		\
	 (!(__sk)->sk_bound_dev_if	||				\
	   ((__sk)->sk_bound_dev_if == (__dif))) 	&&		\
	 net_eq(sock_net(__sk), (__net)))
#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 = skb_steal_sock(skb);
	const struct iphdr *iph = ip_hdr(skb);

	if (sk)
		return sk;
	else
		return __inet_lookup(dev_net(skb_dst(skb)->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 **),
		int (*hash)(struct sock *sk, struct inet_timewait_sock *twp));

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
60063497	36	#include <linux/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 {
3ab5aee7 ED	44	struct hlist_nulls_head chain;
3ab5aee7 ED	45	struct hlist_nulls_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 {
7a9546ee	80	#ifdef CONFIG_NET_NS
941b1d22	81	struct net *ib_net;
7a9546ee	82	#endif
77d8bf9c ACM	83	unsigned short port;
77d8bf9c ACM	84	signed short fastreuse;
a9d8f911	85	int num_owners;
77d8bf9c ACM	86	struct hlist_node node;
	87	struct hlist_head owners;
	88	};
	89
7a9546ee ED	90	static inline struct net ib_net(struct inet_bind_bucket ib)
	91	{
	92	return read_pnet(&ib->ib_net);
	93	}
	94
5256f2ef LAG	95	#define inet_bind_bucket_for_each(tb, pos, head) \
5256f2ef LAG	96	hlist_for_each_entry(tb, pos, head, node)
77d8bf9c ACM	97
	98	struct inet_bind_hashbucket {
	99	spinlock_t lock;
	100	struct hlist_head chain;
	101	};
	102
c25eb3bf ED	103	/*
	104	* Sockets can be hashed in established or listening table
	105	* We must use different 'nulls' end-of-chain value for listening
	106	* hash table, or we might find a socket that was closed and
	107	* reallocated/inserted into established hash table
	108	*/
	109	#define LISTENING_NULLS_BASE (1U << 29)
5caea4ea ED	110	struct inet_listen_hashbucket {
5caea4ea ED	111	spinlock_t lock;
c25eb3bf	112	struct hlist_nulls_head head;
5caea4ea ED	113	};
5caea4ea ED	114
77d8bf9c ACM	115	/* This is for listening sockets, thus all sockets which possess wildcards. */
	116	#define INET_LHTABLE_SIZE 32 /* Yes, really, this is all you need. */
	117
	118	struct inet_hashinfo {
	119	/* This is for sockets with full identity only. Sockets here will
	120	* always be without wildcards and will have the following invariant:
	121	*
	122	* TCP_ESTABLISHED <= sk->sk_state < TCP_CLOSE
	123	*
dbca9b27	124	* TIME_WAIT sockets use a separate chain (twchain).
77d8bf9c ACM	125	*/
77d8bf9c ACM	126	struct inet_ehash_bucket *ehash;
9db66bdc	127	spinlock_t *ehash_locks;
f373b53b	128	unsigned int ehash_mask;
230140cf	129	unsigned int ehash_locks_mask;
77d8bf9c ACM	130
	131	/* Ok, let's try this, I give up, we do need a local binding
	132	* TCP hash as well as the others for fast bind/connect.
	133	*/
	134	struct inet_bind_hashbucket *bhash;
	135
cfcabdcc	136	unsigned int bhash_size;
24dd1fa1	137	/* 4 bytes hole on 64 bit */
77d8bf9c	138
5caea4ea	139	struct kmem_cache *bind_bucket_cachep;
77d8bf9c ACM	140
	141	/* All the above members are written once at bootup and
	142	* never written again _or_ are predominantly read-access.
	143	*
	144	* Now align to a new cache line as all the following members
5caea4ea ED	145	* might be often dirty.
	146	*/
	147	/* All sockets in TCP_LISTEN state will be in here. This is the only
	148	* table where wildcard'd TCP sockets can exist. Hash function here
	149	* is just local port number.
77d8bf9c	150	*/
5caea4ea ED	151	struct inet_listen_hashbucket listening_hash[INET_LHTABLE_SIZE]
	152	____cacheline_aligned_in_smp;
	153
24dd1fa1	154	atomic_t bsockets;
77d8bf9c ACM	155	};
77d8bf9c ACM	156
81c3d547 ED	157	static inline struct inet_ehash_bucket *inet_ehash_bucket(
	158	struct inet_hashinfo *hashinfo,
	159	unsigned int hash)
	160	{
f373b53b	161	return &hashinfo->ehash[hash & hashinfo->ehash_mask];
304a1618 ACM	162	}
304a1618 ACM	163
9db66bdc	164	static inline spinlock_t *inet_ehash_lockp(
230140cf ED	165	struct inet_hashinfo *hashinfo,
	166	unsigned int hash)
	167	{
	168	return &hashinfo->ehash_locks[hash & hashinfo->ehash_locks_mask];
	169	}
	170
	171	static inline int inet_ehash_locks_alloc(struct inet_hashinfo *hashinfo)
	172	{
	173	unsigned int i, size = 256;
	174	#if defined(CONFIG_PROVE_LOCKING)
	175	unsigned int nr_pcpus = 2;
	176	#else
	177	unsigned int nr_pcpus = num_possible_cpus();
	178	#endif
	179	if (nr_pcpus >= 4)
	180	size = 512;
	181	if (nr_pcpus >= 8)
	182	size = 1024;
	183	if (nr_pcpus >= 16)
	184	size = 2048;
	185	if (nr_pcpus >= 32)
	186	size = 4096;
94cd3e6c	187	if (sizeof(spinlock_t) != 0) {
230140cf	188	#ifdef CONFIG_NUMA
9db66bdc ED	189	if (size * sizeof(spinlock_t) > PAGE_SIZE)
9db66bdc ED	190	hashinfo->ehash_locks = vmalloc(size * sizeof(spinlock_t));
230140cf ED	191	else
230140cf ED	192	#endif
9db66bdc	193	hashinfo->ehash_locks = kmalloc(size * sizeof(spinlock_t),
230140cf ED	194	GFP_KERNEL);
	195	if (!hashinfo->ehash_locks)
	196	return ENOMEM;
	197	for (i = 0; i < size; i++)
9db66bdc	198	spin_lock_init(&hashinfo->ehash_locks[i]);
230140cf ED	199	}
	200	hashinfo->ehash_locks_mask = size - 1;
	201	return 0;
	202	}
	203
	204	static inline void inet_ehash_locks_free(struct inet_hashinfo *hashinfo)
	205	{
	206	if (hashinfo->ehash_locks) {
	207	#ifdef CONFIG_NUMA
	208	unsigned int size = (hashinfo->ehash_locks_mask + 1) *
9db66bdc	209	sizeof(spinlock_t);
230140cf ED	210	if (size > PAGE_SIZE)
	211	vfree(hashinfo->ehash_locks);
	212	else
230140cf	213	#endif
218ad12f	214	kfree(hashinfo->ehash_locks);
230140cf ED	215	hashinfo->ehash_locks = NULL;
	216	}
	217	}
	218
77d8bf9c	219	extern struct inet_bind_bucket *
e18b890b	220	inet_bind_bucket_create(struct kmem_cache *cachep,
941b1d22	221	struct net *net,
77d8bf9c ACM	222	struct inet_bind_hashbucket *head,
77d8bf9c ACM	223	const unsigned short snum);
e18b890b	224	extern void inet_bind_bucket_destroy(struct kmem_cache *cachep,
77d8bf9c ACM	225	struct inet_bind_bucket *tb);
77d8bf9c ACM	226
7f635ab7 PE	227	static inline int inet_bhashfn(struct net *net,
7f635ab7 PE	228	const __u16 lport, const int bhash_size)
77d8bf9c	229	{
0b441916	230	return (lport + net_hash_mix(net)) & (bhash_size - 1);
77d8bf9c ACM	231	}
77d8bf9c ACM	232
2d8c4ce5 ACM	233	extern void inet_bind_hash(struct sock sk, struct inet_bind_bucket tb,
	234	const unsigned short snum);
	235
77d8bf9c	236	/* These can have wildcards, don't try too hard. */
2086a650	237	static inline int inet_lhashfn(struct net *net, const unsigned short num)
77d8bf9c	238	{
0b441916	239	return (num + net_hash_mix(net)) & (INET_LHTABLE_SIZE - 1);
77d8bf9c ACM	240	}
	241
	242	static inline int inet_sk_listen_hashfn(const struct sock *sk)
	243	{
c720c7e8	244	return inet_lhashfn(sock_net(sk), inet_sk(sk)->inet_num);
77d8bf9c ACM	245	}
77d8bf9c ACM	246
2d8c4ce5	247	/* Caller must disable local BH processing. */
093d2823	248	extern int __inet_inherit_port(struct sock sk, struct sock child);
2d8c4ce5	249
ab1e0a13	250	extern void inet_put_port(struct sock *sk);
2d8c4ce5	251
5caea4ea	252	void inet_hashinfo_init(struct inet_hashinfo *h);
f3f05f70	253
9327f705	254	extern int __inet_hash_nolisten(struct sock sk, struct inet_timewait_sock tw);
ab1e0a13 ACM	255	extern void inet_hash(struct sock *sk);
ab1e0a13 ACM	256	extern void inet_unhash(struct sock *sk);
33b62231	257
c67499c0 PE	258	extern struct sock __inet_lookup_listener(struct net net,
c67499c0 PE	259	struct inet_hashinfo *hashinfo,
fb99c848	260	const __be32 daddr,
8f491069 HX	261	const unsigned short hnum,
	262	const int dif);
	263
c67499c0 PE	264	static inline struct sock inet_lookup_listener(struct net net,
	265	struct inet_hashinfo *hashinfo,
	266	__be32 daddr, __be16 dport, int dif)
8f491069	267	{
c67499c0	268	return __inet_lookup_listener(net, hashinfo, daddr, ntohs(dport), dif);
8f491069	269	}
8feaf0c0 ACM	270
8feaf0c0 ACM	271	/* Socket demux engine toys. */
4f765d84 AV	272	/* What happens here is ugly; there's a pair of adjacent fields in
	273	struct inet_sock; __be16 dport followed by __u16 num. We want to
	274	search by pair, so we combine the keys into a single 32bit value
	275	and compare with 32bit value read from &...->dport. Let's at least
	276	make sure that it's not mixed with anything else...
	277	On 64bit targets we combine comparisons with pair of adjacent __be32
	278	fields in the same way.
	279	*/
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)
	289	#ifdef __BIG_ENDIAN
	290	#define INET_ADDR_COOKIE(__name, __saddr, __daddr) \
4f765d84 AV	291	const __addrpair __name = (__force __addrpair) ( \
	292	(((__force __u64)(__be32)(__saddr)) << 32) \| \
	293	((__force __u64)(__be32)(__daddr)));
8feaf0c0 ACM	294	#else /* __LITTLE_ENDIAN */
8feaf0c0 ACM	295	#define INET_ADDR_COOKIE(__name, __saddr, __daddr) \
4f765d84 AV	296	const __addrpair __name = (__force __addrpair) ( \
	297	(((__force __u64)(__be32)(__daddr)) << 32) \| \
	298	((__force __u64)(__be32)(__saddr)));
8feaf0c0	299	#endif /* __BIG_ENDIAN */
ce43b03e ED	300	#define INET_MATCH(__sk, __net, __cookie, __saddr, __daddr, __ports, __dif) \
	301	((inet_sk(__sk)->inet_portpair == (__ports)) && \
	302	(inet_sk(__sk)->inet_addrpair == (__cookie)) && \
	303	(!(__sk)->sk_bound_dev_if \|\| \
	304	((__sk)->sk_bound_dev_if == (__dif))) && \
	305	net_eq(sock_net(__sk), (__net)))
	306	#define INET_TW_MATCH(__sk, __net, __cookie, __saddr, __daddr, __ports, __dif)\
	307	((inet_twsk(__sk)->tw_portpair == (__ports)) && \
	308	(inet_twsk(__sk)->tw_addrpair == (__cookie)) && \
	309	(!(__sk)->sk_bound_dev_if \|\| \
	310	((__sk)->sk_bound_dev_if == (__dif))) && \
	311	net_eq(sock_net(__sk), (__net)))
8feaf0c0 ACM	312	#else /* 32-bit arch */
8feaf0c0 ACM	313	#define INET_ADDR_COOKIE(__name, __saddr, __daddr)
ce43b03e ED	314	#define INET_MATCH(__sk, __net, __cookie, __saddr, __daddr, __ports, __dif) \
	315	((inet_sk(__sk)->inet_portpair == (__ports)) && \
	316	(inet_sk(__sk)->inet_daddr == (__saddr)) && \
	317	(inet_sk(__sk)->inet_rcv_saddr == (__daddr)) && \
	318	(!(__sk)->sk_bound_dev_if \|\| \
	319	((__sk)->sk_bound_dev_if == (__dif))) && \
	320	net_eq(sock_net(__sk), (__net)))
	321	#define INET_TW_MATCH(__sk, __net, __cookie, __saddr, __daddr, __ports, __dif) \
	322	((inet_twsk(__sk)->tw_portpair == (__ports)) && \
	323	(inet_twsk(__sk)->tw_daddr == (__saddr)) && \
	324	(inet_twsk(__sk)->tw_rcv_saddr == (__daddr)) && \
	325	(!(__sk)->sk_bound_dev_if \|\| \
	326	((__sk)->sk_bound_dev_if == (__dif))) && \
	327	net_eq(sock_net(__sk), (__net)))
8feaf0c0	328	#endif /* 64-bit arch */
e48c414e ACM	329
	330	/*
	331	* Sockets in TCP_CLOSE state are _always_ taken out of the hash, so we need
	332	* not check it for lookups anymore, thanks Alexey. -DaveM
	333	*
	334	* Local BH must be disabled here.
	335	*/
c67499c0 PE	336	extern struct sock * __inet_lookup_established(struct net *net,
c67499c0 PE	337	struct inet_hashinfo *hashinfo,
77a5ba55 PE	338	const __be32 saddr, const __be16 sport,
77a5ba55 PE	339	const __be32 daddr, const u16 hnum, const int dif);
e48c414e	340
8f491069	341	static inline struct sock *
c67499c0	342	inet_lookup_established(struct net net, struct inet_hashinfo hashinfo,
fb99c848 AV	343	const __be32 saddr, const __be16 sport,
fb99c848 AV	344	const __be32 daddr, const __be16 dport,
8f491069 HX	345	const int dif)
8f491069 HX	346	{
c67499c0	347	return __inet_lookup_established(net, hashinfo, saddr, sport, daddr,
8f491069 HX	348	ntohs(dport), dif);
	349	}
	350
c67499c0 PE	351	static inline struct sock __inet_lookup(struct net net,
c67499c0 PE	352	struct inet_hashinfo *hashinfo,
fb99c848 AV	353	const __be32 saddr, const __be16 sport,
fb99c848 AV	354	const __be32 daddr, const __be16 dport,
e48c414e ACM	355	const int dif)
e48c414e ACM	356	{
8f491069	357	u16 hnum = ntohs(dport);
c67499c0 PE	358	struct sock *sk = __inet_lookup_established(net, hashinfo,
	359	saddr, sport, daddr, hnum, dif);
	360
	361	return sk ? : __inet_lookup_listener(net, hashinfo, daddr, hnum, dif);
e48c414e ACM	362	}
e48c414e ACM	363
c67499c0 PE	364	static inline struct sock inet_lookup(struct net net,
c67499c0 PE	365	struct inet_hashinfo *hashinfo,
fb99c848 AV	366	const __be32 saddr, const __be16 sport,
fb99c848 AV	367	const __be32 daddr, const __be16 dport,
e48c414e ACM	368	const int dif)
	369	{
	370	struct sock *sk;
	371
	372	local_bh_disable();
c67499c0	373	sk = __inet_lookup(net, hashinfo, saddr, sport, daddr, dport, dif);
e48c414e ACM	374	local_bh_enable();
	375
	376	return sk;
	377	}
a7f5e7f1	378
9a1f27c4 ACM	379	static inline struct sock __inet_lookup_skb(struct inet_hashinfo hashinfo,
	380	struct sk_buff *skb,
	381	const __be16 sport,
	382	const __be16 dport)
	383	{
41063e9d	384	struct sock *sk = skb_steal_sock(skb);
9a1f27c4 ACM	385	const struct iphdr *iph = ip_hdr(skb);
9a1f27c4 ACM	386
41063e9d	387	if (sk)
23542618 KK	388	return sk;
23542618 KK	389	else
adf30907	390	return __inet_lookup(dev_net(skb_dst(skb)->dev), hashinfo,
23542618 KK	391	iph->saddr, sport,
23542618 KK	392	iph->daddr, dport, inet_iif(skb));
9a1f27c4 ACM	393	}
9a1f27c4 ACM	394
5ee31fc1	395	extern int __inet_hash_connect(struct inet_timewait_death_row *death_row,
9327f705 ED	396	struct sock *sk,
9327f705 ED	397	u32 port_offset,
5ee31fc1 PE	398	int (check_established)(struct inet_timewait_death_row ,
5ee31fc1 PE	399	struct sock , __u16, struct inet_timewait_sock *),
9327f705 ED	400	int (hash)(struct sock sk, struct inet_timewait_sock *twp));
9327f705 ED	401
a7f5e7f1 ACM	402	extern int inet_hash_connect(struct inet_timewait_death_row *death_row,
a7f5e7f1 ACM	403	struct sock *sk);
304a1618	404	#endif /* _INET_HASHTABLES_H */