inet_hashtables.h 12.3 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
/*
 * 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

17

18
#include <linux/interrupt.h>
19
#include <linux/ip.h>
20
#include <linux/ipv6.h>
21 22
#include <linux/list.h>
#include <linux/slab.h>
23
#include <linux/socket.h>
24
#include <linux/spinlock.h>
25
#include <linux/types.h>
26
#include <linux/wait.h>
27

28
#include <net/inet_connection_sock.h>
29
#include <net/inet_sock.h>
30
#include <net/sock.h>
31
#include <net/route.h>
32
#include <net/tcp_states.h>
33
#include <net/netns/hash.h>
34

35
#include <linux/atomic.h>
36
#include <asm/byteorder.h>
37

38
/* This is for all connections with a full identity, no wildcards.
Eric Dumazet's avatar
Eric Dumazet committed
39 40
 * The 'e' prefix stands for Establish, but we really put all sockets
 * but LISTEN ones.
41 42
 */
struct inet_ehash_bucket {
43
	struct hlist_nulls_head chain;
44
};
45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77

/* 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 {
78
	possible_net_t		ib_net;
79
	unsigned short		port;
80 81 82
	signed char		fastreuse;
	signed char		fastreuseport;
	kuid_t			fastuid;
83
	int			num_owners;
84 85 86 87
	struct hlist_node	node;
	struct hlist_head	owners;
};

88 89 90 91 92
static inline struct net *ib_net(struct inet_bind_bucket *ib)
{
	return read_pnet(&ib->ib_net);
}

93 94
#define inet_bind_bucket_for_each(tb, head) \
	hlist_for_each_entry(tb, head, node)
95 96 97 98 99 100

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

101 102 103
/*
 * Sockets can be hashed in established or listening table
 */
104 105
struct inet_listen_hashbucket {
	spinlock_t		lock;
106
	struct hlist_head	head;
107 108
};

109 110 111 112 113 114 115 116 117 118 119
/* 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
	 *
	 */
	struct inet_ehash_bucket	*ehash;
120
	spinlock_t			*ehash_locks;
121
	unsigned int			ehash_mask;
122
	unsigned int			ehash_locks_mask;
123 124 125 126 127 128

	/* 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;

129
	unsigned int			bhash_size;
130
	/* 4 bytes hole on 64 bit */
131

132
	struct kmem_cache		*bind_bucket_cachep;
133 134 135 136 137

	/* 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
138 139 140 141 142
	 * 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.
143
	 */
144 145
	struct inet_listen_hashbucket	listening_hash[INET_LHTABLE_SIZE]
					____cacheline_aligned_in_smp;
146 147
};

148 149 150 151
static inline struct inet_ehash_bucket *inet_ehash_bucket(
	struct inet_hashinfo *hashinfo,
	unsigned int hash)
{
152
	return &hashinfo->ehash[hash & hashinfo->ehash_mask];
153 154
}

155
static inline spinlock_t *inet_ehash_lockp(
156 157 158 159 160 161
	struct inet_hashinfo *hashinfo,
	unsigned int hash)
{
	return &hashinfo->ehash_locks[hash & hashinfo->ehash_locks_mask];
}

162
int inet_ehash_locks_alloc(struct inet_hashinfo *hashinfo);
163 164 165

static inline void inet_ehash_locks_free(struct inet_hashinfo *hashinfo)
{
166 167
	kvfree(hashinfo->ehash_locks);
	hashinfo->ehash_locks = NULL;
168 169
}

170 171 172 173 174 175 176
struct inet_bind_bucket *
inet_bind_bucket_create(struct kmem_cache *cachep, struct net *net,
			struct inet_bind_hashbucket *head,
			const unsigned short snum);
void inet_bind_bucket_destroy(struct kmem_cache *cachep,
			      struct inet_bind_bucket *tb);

177 178
static inline u32 inet_bhashfn(const struct net *net, const __u16 lport,
			       const u32 bhash_size)
179
{
180
	return (lport + net_hash_mix(net)) & (bhash_size - 1);
181 182
}

183 184
void inet_bind_hash(struct sock *sk, struct inet_bind_bucket *tb,
		    const unsigned short snum);
185

186
/* These can have wildcards, don't try too hard. */
187
static inline u32 inet_lhashfn(const struct net *net, const unsigned short num)
188
{
189
	return (num + net_hash_mix(net)) & (INET_LHTABLE_SIZE - 1);
190 191 192 193
}

static inline int inet_sk_listen_hashfn(const struct sock *sk)
{
194
	return inet_lhashfn(sock_net(sk), inet_sk(sk)->inet_num);
195 196
}

197
/* Caller must disable local BH processing. */
198
int __inet_inherit_port(const struct sock *sk, struct sock *child);
199

200
void inet_put_port(struct sock *sk);
201

202
void inet_hashinfo_init(struct inet_hashinfo *h);
203

204 205
bool inet_ehash_insert(struct sock *sk, struct sock *osk);
bool inet_ehash_nolisten(struct sock *sk, struct sock *osk);
206 207 208 209
int __inet_hash(struct sock *sk, struct sock *osk,
		int (*saddr_same)(const struct sock *sk1,
				  const struct sock *sk2,
				  bool match_wildcard));
210
int inet_hash(struct sock *sk);
211
void inet_unhash(struct sock *sk);
212

213 214
struct sock *__inet_lookup_listener(struct net *net,
				    struct inet_hashinfo *hashinfo,
215
				    struct sk_buff *skb, int doff,
216 217 218 219
				    const __be32 saddr, const __be16 sport,
				    const __be32 daddr,
				    const unsigned short hnum,
				    const int dif);
220

221 222
static inline struct sock *inet_lookup_listener(struct net *net,
		struct inet_hashinfo *hashinfo,
223
		struct sk_buff *skb, int doff,
224
		__be32 saddr, __be16 sport,
225
		__be32 daddr, __be16 dport, int dif)
226
{
227
	return __inet_lookup_listener(net, hashinfo, skb, doff, saddr, sport,
228
				      daddr, ntohs(dport), dif);
229
}
230 231

/* Socket demux engine toys. */
232 233 234 235 236 237 238 239
/* 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.
*/
240 241
#ifdef __BIG_ENDIAN
#define INET_COMBINED_PORTS(__sport, __dport) \
242
	((__force __portpair)(((__force __u32)(__be16)(__sport) << 16) | (__u32)(__dport)))
243 244
#else /* __LITTLE_ENDIAN */
#define INET_COMBINED_PORTS(__sport, __dport) \
245
	((__force __portpair)(((__u32)(__dport) << 16) | (__force __u32)(__be16)(__sport)))
246 247 248 249 250
#endif

#if (BITS_PER_LONG == 64)
#ifdef __BIG_ENDIAN
#define INET_ADDR_COOKIE(__name, __saddr, __daddr) \
251 252
	const __addrpair __name = (__force __addrpair) ( \
				   (((__force __u64)(__be32)(__saddr)) << 32) | \
253
				   ((__force __u64)(__be32)(__daddr)))
254 255
#else /* __LITTLE_ENDIAN */
#define INET_ADDR_COOKIE(__name, __saddr, __daddr) \
256 257
	const __addrpair __name = (__force __addrpair) ( \
				   (((__force __u64)(__be32)(__daddr)) << 32) | \
258
				   ((__force __u64)(__be32)(__saddr)))
259
#endif /* __BIG_ENDIAN */
260
#define INET_MATCH(__sk, __net, __cookie, __saddr, __daddr, __ports, __dif)	\
261 262
	(((__sk)->sk_portpair == (__ports))			&&	\
	 ((__sk)->sk_addrpair == (__cookie))			&&	\
263 264 265
	 (!(__sk)->sk_bound_dev_if	||				\
	   ((__sk)->sk_bound_dev_if == (__dif))) 		&& 	\
	 net_eq(sock_net(__sk), (__net)))
266
#else /* 32-bit arch */
267 268 269
#define INET_ADDR_COOKIE(__name, __saddr, __daddr) \
	const int __name __deprecated __attribute__((unused))

270
#define INET_MATCH(__sk, __net, __cookie, __saddr, __daddr, __ports, __dif) \
271 272 273
	(((__sk)->sk_portpair == (__ports))		&&		\
	 ((__sk)->sk_daddr	== (__saddr))		&&		\
	 ((__sk)->sk_rcv_saddr	== (__daddr))		&&		\
274 275 276
	 (!(__sk)->sk_bound_dev_if	||				\
	   ((__sk)->sk_bound_dev_if == (__dif))) 	&&		\
	 net_eq(sock_net(__sk), (__net)))
277
#endif /* 64-bit arch */
278

279
/* Sockets in TCP_CLOSE state are _always_ taken out of the hash, so we need
280 281
 * not check it for lookups anymore, thanks Alexey. -DaveM
 */
282 283 284 285 286
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);
287

288
static inline struct sock *
289
	inet_lookup_established(struct net *net, struct inet_hashinfo *hashinfo,
290 291
				const __be32 saddr, const __be16 sport,
				const __be32 daddr, const __be16 dport,
292 293
				const int dif)
{
294
	return __inet_lookup_established(net, hashinfo, saddr, sport, daddr,
295 296 297
					 ntohs(dport), dif);
}

298 299
static inline struct sock *__inet_lookup(struct net *net,
					 struct inet_hashinfo *hashinfo,
300
					 struct sk_buff *skb, int doff,
301 302
					 const __be32 saddr, const __be16 sport,
					 const __be32 daddr, const __be16 dport,
303 304
					 const int dif,
					 bool *refcounted)
305
{
306
	u16 hnum = ntohs(dport);
307
	struct sock *sk;
308

309 310 311 312 313 314 315 316
	sk = __inet_lookup_established(net, hashinfo, saddr, sport,
				       daddr, hnum, dif);
	*refcounted = true;
	if (sk)
		return sk;
	*refcounted = false;
	return __inet_lookup_listener(net, hashinfo, skb, doff, saddr,
				      sport, daddr, hnum, dif);
317 318
}

319 320
static inline struct sock *inet_lookup(struct net *net,
				       struct inet_hashinfo *hashinfo,
321
				       struct sk_buff *skb, int doff,
322 323
				       const __be32 saddr, const __be16 sport,
				       const __be32 daddr, const __be16 dport,
324 325 326
				       const int dif)
{
	struct sock *sk;
327
	bool refcounted;
328

329
	sk = __inet_lookup(net, hashinfo, skb, doff, saddr, sport, daddr,
330
			   dport, dif, &refcounted);
331

332 333
	if (sk && !refcounted && !atomic_inc_not_zero(&sk->sk_refcnt))
		sk = NULL;
334 335
	return sk;
}
336

337 338
static inline struct sock *__inet_lookup_skb(struct inet_hashinfo *hashinfo,
					     struct sk_buff *skb,
339
					     int doff,
340
					     const __be16 sport,
341 342
					     const __be16 dport,
					     bool *refcounted)
343
{
344
	struct sock *sk = skb_steal_sock(skb);
345 346
	const struct iphdr *iph = ip_hdr(skb);

347
	*refcounted = true;
348
	if (sk)
349
		return sk;
350 351 352 353 354

	return __inet_lookup(dev_net(skb_dst(skb)->dev), hashinfo, skb,
			     doff, iph->saddr, sport,
			     iph->daddr, dport, inet_iif(skb),
			     refcounted);
355 356
}

357
u32 sk_ehashfn(const struct sock *sk);
358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376
u32 inet6_ehashfn(const struct net *net,
		  const struct in6_addr *laddr, const u16 lport,
		  const struct in6_addr *faddr, const __be16 fport);

static inline void sk_daddr_set(struct sock *sk, __be32 addr)
{
	sk->sk_daddr = addr; /* alias of inet_daddr */
#if IS_ENABLED(CONFIG_IPV6)
	ipv6_addr_set_v4mapped(addr, &sk->sk_v6_daddr);
#endif
}

static inline void sk_rcv_saddr_set(struct sock *sk, __be32 addr)
{
	sk->sk_rcv_saddr = addr; /* alias of inet_rcv_saddr */
#if IS_ENABLED(CONFIG_IPV6)
	ipv6_addr_set_v4mapped(addr, &sk->sk_v6_rcv_saddr);
#endif
}
377

378 379 380 381
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,
382
						 struct inet_timewait_sock **));
383

384 385
int inet_hash_connect(struct inet_timewait_death_row *death_row,
		      struct sock *sk);
386
#endif /* _INET_HASHTABLES_H */