inet_timewait_sock.c 12.5 KB
Newer Older
1 2 3 4 5 6 7 8 9 10
/*
 * 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.
 *
 *		Generic TIME_WAIT sockets functions
 *
 *		From code orinally in TCP
 */

11
#include <linux/kernel.h>
12
#include <linux/kmemcheck.h>
13 14
#include <net/inet_hashtables.h>
#include <net/inet_timewait_sock.h>
15
#include <net/ip.h>
16 17

/* Must be called with locally disabled BHs. */
18 19
static void __inet_twsk_kill(struct inet_timewait_sock *tw,
			     struct inet_hashinfo *hashinfo)
20 21 22 23
{
	struct inet_bind_hashbucket *bhead;
	struct inet_bind_bucket *tb;
	/* Unlink from established hashes. */
24
	spinlock_t *lock = inet_ehash_lockp(hashinfo, tw->tw_hash);
25

26
	spin_lock(lock);
27
	if (hlist_nulls_unhashed(&tw->tw_node)) {
28
		spin_unlock(lock);
29 30
		return;
	}
31 32
	hlist_nulls_del_rcu(&tw->tw_node);
	sk_nulls_node_init(&tw->tw_node);
33
	spin_unlock(lock);
34 35

	/* Disassociate with bind bucket. */
36 37
	bhead = &hashinfo->bhash[inet_bhashfn(twsk_net(tw), tw->tw_num,
			hashinfo->bhash_size)];
38 39 40 41 42 43 44 45 46 47 48 49 50 51 52
	spin_lock(&bhead->lock);
	tb = tw->tw_tb;
	__hlist_del(&tw->tw_bind_node);
	tw->tw_tb = NULL;
	inet_bind_bucket_destroy(hashinfo->bind_bucket_cachep, tb);
	spin_unlock(&bhead->lock);
#ifdef SOCK_REFCNT_DEBUG
	if (atomic_read(&tw->tw_refcnt) != 1) {
		printk(KERN_DEBUG "%s timewait_sock %p refcnt=%d\n",
		       tw->tw_prot->name, tw, atomic_read(&tw->tw_refcnt));
	}
#endif
	inet_twsk_put(tw);
}

53
static noinline void inet_twsk_free(struct inet_timewait_sock *tw)
54
{
55 56
	struct module *owner = tw->tw_prot->owner;
	twsk_destructor((struct sock *)tw);
57
#ifdef SOCK_REFCNT_DEBUG
58
	pr_debug("%s timewait_sock %p released\n", tw->tw_prot->name, tw);
59
#endif
60 61 62 63 64 65 66 67 68
	release_net(twsk_net(tw));
	kmem_cache_free(tw->tw_prot->twsk_prot->twsk_slab, tw);
	module_put(owner);
}

void inet_twsk_put(struct inet_timewait_sock *tw)
{
	if (atomic_dec_and_test(&tw->tw_refcnt))
		inet_twsk_free(tw);
69 70 71
}
EXPORT_SYMBOL_GPL(inet_twsk_put);

72 73 74 75 76 77 78 79 80
/*
 * Enter the time wait state. This is called with locally disabled BH.
 * Essentially we whip up a timewait bucket, copy the relevant info into it
 * from the SK, and mess with hash chains and list linkage.
 */
void __inet_twsk_hashdance(struct inet_timewait_sock *tw, struct sock *sk,
			   struct inet_hashinfo *hashinfo)
{
	const struct inet_sock *inet = inet_sk(sk);
81
	const struct inet_connection_sock *icsk = inet_csk(sk);
82
	struct inet_ehash_bucket *ehead = inet_ehash_bucket(hashinfo, sk->sk_hash);
83
	spinlock_t *lock = inet_ehash_lockp(hashinfo, sk->sk_hash);
84 85 86 87 88
	struct inet_bind_hashbucket *bhead;
	/* Step 1: Put TW into bind hash. Original socket stays there too.
	   Note, that any socket with inet->num != 0 MUST be bound in
	   binding cache, even if it is closed.
	 */
89 90
	bhead = &hashinfo->bhash[inet_bhashfn(twsk_net(tw), inet->num,
			hashinfo->bhash_size)];
91
	spin_lock(&bhead->lock);
92
	tw->tw_tb = icsk->icsk_bind_hash;
93
	WARN_ON(!icsk->icsk_bind_hash);
94 95 96
	inet_twsk_add_bind_node(tw, &tw->tw_tb->owners);
	spin_unlock(&bhead->lock);

97
	spin_lock(lock);
98

99 100 101 102 103
	/*
	 * Step 2: Hash TW into TIMEWAIT chain.
	 * Should be done before removing sk from established chain
	 * because readers are lockless and search established first.
	 */
104
	atomic_inc(&tw->tw_refcnt);
105 106 107 108 109
	inet_twsk_add_node_rcu(tw, &ehead->twchain);

	/* Step 3: Remove SK from established hash. */
	if (__sk_nulls_del_node_init_rcu(sk))
		sock_prot_inuse_add(sock_net(sk), sk->sk_prot, -1);
110

111
	spin_unlock(lock);
112
}
113

114 115
EXPORT_SYMBOL_GPL(__inet_twsk_hashdance);

116 117
struct inet_timewait_sock *inet_twsk_alloc(const struct sock *sk, const int state)
{
118 119
	struct inet_timewait_sock *tw =
		kmem_cache_alloc(sk->sk_prot_creator->twsk_prot->twsk_slab,
120
				 GFP_ATOMIC);
121 122 123
	if (tw != NULL) {
		const struct inet_sock *inet = inet_sk(sk);

124 125
		kmemcheck_annotate_bitfield(tw, flags);

126 127 128 129 130 131 132 133 134 135 136
		/* Give us an identity. */
		tw->tw_daddr	    = inet->daddr;
		tw->tw_rcv_saddr    = inet->rcv_saddr;
		tw->tw_bound_dev_if = sk->sk_bound_dev_if;
		tw->tw_num	    = inet->num;
		tw->tw_state	    = TCP_TIME_WAIT;
		tw->tw_substate	    = state;
		tw->tw_sport	    = inet->sport;
		tw->tw_dport	    = inet->dport;
		tw->tw_family	    = sk->sk_family;
		tw->tw_reuse	    = sk->sk_reuse;
137
		tw->tw_hash	    = sk->sk_hash;
138
		tw->tw_ipv6only	    = 0;
139
		tw->tw_transparent  = inet->transparent;
140
		tw->tw_prot	    = sk->sk_prot_creator;
141
		twsk_net_set(tw, hold_net(sock_net(sk)));
142 143
		atomic_set(&tw->tw_refcnt, 1);
		inet_twsk_dead_node_init(tw);
144
		__module_get(tw->tw_prot->owner);
145 146 147 148
	}

	return tw;
}
149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173

EXPORT_SYMBOL_GPL(inet_twsk_alloc);

/* Returns non-zero if quota exceeded.  */
static int inet_twdr_do_twkill_work(struct inet_timewait_death_row *twdr,
				    const int slot)
{
	struct inet_timewait_sock *tw;
	struct hlist_node *node;
	unsigned int killed;
	int ret;

	/* NOTE: compare this to previous version where lock
	 * was released after detaching chain. It was racy,
	 * because tw buckets are scheduled in not serialized context
	 * in 2.3 (with netfilter), and with softnet it is common, because
	 * soft irqs are not sequenced.
	 */
	killed = 0;
	ret = 0;
rescan:
	inet_twsk_for_each_inmate(tw, node, &twdr->cells[slot]) {
		__inet_twsk_del_dead_node(tw);
		spin_unlock(&twdr->death_lock);
		__inet_twsk_kill(tw, twdr->hashinfo);
174 175 176
#ifdef CONFIG_NET_NS
		NET_INC_STATS_BH(twsk_net(tw), LINUX_MIB_TIMEWAITED);
#endif
177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194
		inet_twsk_put(tw);
		killed++;
		spin_lock(&twdr->death_lock);
		if (killed > INET_TWDR_TWKILL_QUOTA) {
			ret = 1;
			break;
		}

		/* While we dropped twdr->death_lock, another cpu may have
		 * killed off the next TW bucket in the list, therefore
		 * do a fresh re-read of the hlist head node with the
		 * lock reacquired.  We still use the hlist traversal
		 * macro in order to get the prefetches.
		 */
		goto rescan;
	}

	twdr->tw_count -= killed;
195 196 197
#ifndef CONFIG_NET_NS
	NET_ADD_STATS_BH(&init_net, LINUX_MIB_TIMEWAITED, killed);
#endif
198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220
	return ret;
}

void inet_twdr_hangman(unsigned long data)
{
	struct inet_timewait_death_row *twdr;
	int unsigned need_timer;

	twdr = (struct inet_timewait_death_row *)data;
	spin_lock(&twdr->death_lock);

	if (twdr->tw_count == 0)
		goto out;

	need_timer = 0;
	if (inet_twdr_do_twkill_work(twdr, twdr->slot)) {
		twdr->thread_slots |= (1 << twdr->slot);
		schedule_work(&twdr->twkill_work);
		need_timer = 1;
	} else {
		/* We purged the entire slot, anything left?  */
		if (twdr->tw_count)
			need_timer = 1;
221
		twdr->slot = ((twdr->slot + 1) & (INET_TWDR_TWKILL_SLOTS - 1));
222 223 224 225 226 227 228 229 230
	}
	if (need_timer)
		mod_timer(&twdr->tw_timer, jiffies + twdr->period);
out:
	spin_unlock(&twdr->death_lock);
}

EXPORT_SYMBOL_GPL(inet_twdr_hangman);

231
void inet_twdr_twkill_work(struct work_struct *work)
232
{
233 234
	struct inet_timewait_death_row *twdr =
		container_of(work, struct inet_timewait_death_row, twkill_work);
235 236
	int i;

237 238
	BUILD_BUG_ON((INET_TWDR_TWKILL_SLOTS - 1) >
			(sizeof(twdr->thread_slots) * 8));
239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327

	while (twdr->thread_slots) {
		spin_lock_bh(&twdr->death_lock);
		for (i = 0; i < INET_TWDR_TWKILL_SLOTS; i++) {
			if (!(twdr->thread_slots & (1 << i)))
				continue;

			while (inet_twdr_do_twkill_work(twdr, i) != 0) {
				if (need_resched()) {
					spin_unlock_bh(&twdr->death_lock);
					schedule();
					spin_lock_bh(&twdr->death_lock);
				}
			}

			twdr->thread_slots &= ~(1 << i);
		}
		spin_unlock_bh(&twdr->death_lock);
	}
}

EXPORT_SYMBOL_GPL(inet_twdr_twkill_work);

/* These are always called from BH context.  See callers in
 * tcp_input.c to verify this.
 */

/* This is for handling early-kills of TIME_WAIT sockets. */
void inet_twsk_deschedule(struct inet_timewait_sock *tw,
			  struct inet_timewait_death_row *twdr)
{
	spin_lock(&twdr->death_lock);
	if (inet_twsk_del_dead_node(tw)) {
		inet_twsk_put(tw);
		if (--twdr->tw_count == 0)
			del_timer(&twdr->tw_timer);
	}
	spin_unlock(&twdr->death_lock);
	__inet_twsk_kill(tw, twdr->hashinfo);
}

EXPORT_SYMBOL(inet_twsk_deschedule);

void inet_twsk_schedule(struct inet_timewait_sock *tw,
		       struct inet_timewait_death_row *twdr,
		       const int timeo, const int timewait_len)
{
	struct hlist_head *list;
	int slot;

	/* timeout := RTO * 3.5
	 *
	 * 3.5 = 1+2+0.5 to wait for two retransmits.
	 *
	 * RATIONALE: if FIN arrived and we entered TIME-WAIT state,
	 * our ACK acking that FIN can be lost. If N subsequent retransmitted
	 * FINs (or previous seqments) are lost (probability of such event
	 * is p^(N+1), where p is probability to lose single packet and
	 * time to detect the loss is about RTO*(2^N - 1) with exponential
	 * backoff). Normal timewait length is calculated so, that we
	 * waited at least for one retransmitted FIN (maximal RTO is 120sec).
	 * [ BTW Linux. following BSD, violates this requirement waiting
	 *   only for 60sec, we should wait at least for 240 secs.
	 *   Well, 240 consumes too much of resources 8)
	 * ]
	 * This interval is not reduced to catch old duplicate and
	 * responces to our wandering segments living for two MSLs.
	 * However, if we use PAWS to detect
	 * old duplicates, we can reduce the interval to bounds required
	 * by RTO, rather than MSL. So, if peer understands PAWS, we
	 * kill tw bucket after 3.5*RTO (it is important that this number
	 * is greater than TS tick!) and detect old duplicates with help
	 * of PAWS.
	 */
	slot = (timeo + (1 << INET_TWDR_RECYCLE_TICK) - 1) >> INET_TWDR_RECYCLE_TICK;

	spin_lock(&twdr->death_lock);

	/* Unlink it, if it was scheduled */
	if (inet_twsk_del_dead_node(tw))
		twdr->tw_count--;
	else
		atomic_inc(&tw->tw_refcnt);

	if (slot >= INET_TWDR_RECYCLE_SLOTS) {
		/* Schedule to slow timer */
		if (timeo >= timewait_len) {
			slot = INET_TWDR_TWKILL_SLOTS - 1;
		} else {
328
			slot = DIV_ROUND_UP(timeo, twdr->period);
329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389
			if (slot >= INET_TWDR_TWKILL_SLOTS)
				slot = INET_TWDR_TWKILL_SLOTS - 1;
		}
		tw->tw_ttd = jiffies + timeo;
		slot = (twdr->slot + slot) & (INET_TWDR_TWKILL_SLOTS - 1);
		list = &twdr->cells[slot];
	} else {
		tw->tw_ttd = jiffies + (slot << INET_TWDR_RECYCLE_TICK);

		if (twdr->twcal_hand < 0) {
			twdr->twcal_hand = 0;
			twdr->twcal_jiffie = jiffies;
			twdr->twcal_timer.expires = twdr->twcal_jiffie +
					      (slot << INET_TWDR_RECYCLE_TICK);
			add_timer(&twdr->twcal_timer);
		} else {
			if (time_after(twdr->twcal_timer.expires,
				       jiffies + (slot << INET_TWDR_RECYCLE_TICK)))
				mod_timer(&twdr->twcal_timer,
					  jiffies + (slot << INET_TWDR_RECYCLE_TICK));
			slot = (twdr->twcal_hand + slot) & (INET_TWDR_RECYCLE_SLOTS - 1);
		}
		list = &twdr->twcal_row[slot];
	}

	hlist_add_head(&tw->tw_death_node, list);

	if (twdr->tw_count++ == 0)
		mod_timer(&twdr->tw_timer, jiffies + twdr->period);
	spin_unlock(&twdr->death_lock);
}

EXPORT_SYMBOL_GPL(inet_twsk_schedule);

void inet_twdr_twcal_tick(unsigned long data)
{
	struct inet_timewait_death_row *twdr;
	int n, slot;
	unsigned long j;
	unsigned long now = jiffies;
	int killed = 0;
	int adv = 0;

	twdr = (struct inet_timewait_death_row *)data;

	spin_lock(&twdr->death_lock);
	if (twdr->twcal_hand < 0)
		goto out;

	slot = twdr->twcal_hand;
	j = twdr->twcal_jiffie;

	for (n = 0; n < INET_TWDR_RECYCLE_SLOTS; n++) {
		if (time_before_eq(j, now)) {
			struct hlist_node *node, *safe;
			struct inet_timewait_sock *tw;

			inet_twsk_for_each_inmate_safe(tw, node, safe,
						       &twdr->twcal_row[slot]) {
				__inet_twsk_del_dead_node(tw);
				__inet_twsk_kill(tw, twdr->hashinfo);
390 391 392
#ifdef CONFIG_NET_NS
				NET_INC_STATS_BH(twsk_net(tw), LINUX_MIB_TIMEWAITKILLED);
#endif
393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415
				inet_twsk_put(tw);
				killed++;
			}
		} else {
			if (!adv) {
				adv = 1;
				twdr->twcal_jiffie = j;
				twdr->twcal_hand = slot;
			}

			if (!hlist_empty(&twdr->twcal_row[slot])) {
				mod_timer(&twdr->twcal_timer, j);
				goto out;
			}
		}
		j += 1 << INET_TWDR_RECYCLE_TICK;
		slot = (slot + 1) & (INET_TWDR_RECYCLE_SLOTS - 1);
	}
	twdr->twcal_hand = -1;

out:
	if ((twdr->tw_count -= killed) == 0)
		del_timer(&twdr->tw_timer);
416 417 418
#ifndef CONFIG_NET_NS
	NET_ADD_STATS_BH(&init_net, LINUX_MIB_TIMEWAITKILLED, killed);
#endif
419 420 421 422
	spin_unlock(&twdr->death_lock);
}

EXPORT_SYMBOL_GPL(inet_twdr_twcal_tick);
423 424 425 426 427 428

void inet_twsk_purge(struct net *net, struct inet_hashinfo *hashinfo,
		     struct inet_timewait_death_row *twdr, int family)
{
	struct inet_timewait_sock *tw;
	struct sock *sk;
429
	struct hlist_nulls_node *node;
430 431 432
	int h;

	local_bh_disable();
433
	for (h = 0; h <= hashinfo->ehash_mask; h++) {
434 435
		struct inet_ehash_bucket *head =
			inet_ehash_bucket(hashinfo, h);
436
		spinlock_t *lock = inet_ehash_lockp(hashinfo, h);
437
restart:
438
		spin_lock(lock);
439
		sk_nulls_for_each(sk, node, &head->twchain) {
440 441 442 443 444 445 446

			tw = inet_twsk(sk);
			if (!net_eq(twsk_net(tw), net) ||
			    tw->tw_family != family)
				continue;

			atomic_inc(&tw->tw_refcnt);
447
			spin_unlock(lock);
448 449 450 451 452
			inet_twsk_deschedule(tw, twdr);
			inet_twsk_put(tw);

			goto restart;
		}
453
		spin_unlock(lock);
454 455 456 457
	}
	local_bh_enable();
}
EXPORT_SYMBOL_GPL(inet_twsk_purge);