tcp_timer.c 18.2 KB
Newer Older
L
Linus Torvalds 已提交
1 2 3 4 5 6 7
/*
 * 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.
 *
 *		Implementation of the Transmission Control Protocol(TCP).
 *
8
 * Authors:	Ross Biro
L
Linus Torvalds 已提交
9 10 11 12 13 14 15 16 17 18 19 20 21
 *		Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
 *		Mark Evans, <evansmp@uhura.aston.ac.uk>
 *		Corey Minyard <wf-rch!minyard@relay.EU.net>
 *		Florian La Roche, <flla@stud.uni-sb.de>
 *		Charles Hedrick, <hedrick@klinzhai.rutgers.edu>
 *		Linus Torvalds, <torvalds@cs.helsinki.fi>
 *		Alan Cox, <gw4pts@gw4pts.ampr.org>
 *		Matthew Dillon, <dillon@apollo.west.oic.com>
 *		Arnt Gulbrandsen, <agulbra@nvg.unit.no>
 *		Jorge Cwik, <jorge@laser.satlink.net>
 */

#include <linux/module.h>
22
#include <linux/gfp.h>
L
Linus Torvalds 已提交
23 24
#include <net/tcp.h>

25 26 27 28 29 30 31 32
int sysctl_tcp_syn_retries __read_mostly = TCP_SYN_RETRIES;
int sysctl_tcp_synack_retries __read_mostly = TCP_SYNACK_RETRIES;
int sysctl_tcp_keepalive_time __read_mostly = TCP_KEEPALIVE_TIME;
int sysctl_tcp_keepalive_probes __read_mostly = TCP_KEEPALIVE_PROBES;
int sysctl_tcp_keepalive_intvl __read_mostly = TCP_KEEPALIVE_INTVL;
int sysctl_tcp_retries1 __read_mostly = TCP_RETR1;
int sysctl_tcp_retries2 __read_mostly = TCP_RETR2;
int sysctl_tcp_orphan_retries __read_mostly;
A
Andreas Petlund 已提交
33
int sysctl_tcp_thin_linear_timeouts __read_mostly;
L
Linus Torvalds 已提交
34 35 36 37 38 39 40

static void tcp_write_err(struct sock *sk)
{
	sk->sk_err = sk->sk_err_soft ? : ETIMEDOUT;
	sk->sk_error_report(sk);

	tcp_done(sk);
41
	NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPABORTONTIMEOUT);
L
Linus Torvalds 已提交
42 43 44 45 46 47 48
}

/* Do not allow orphaned sockets to eat all our resources.
 * This is direct violation of TCP specs, but it is required
 * to prevent DoS attacks. It is called when a retransmission timeout
 * or zero probe timeout occurs on orphaned socket.
 *
S
Stephen Hemminger 已提交
49
 * Criteria is still not confirmed experimentally and may change.
L
Linus Torvalds 已提交
50 51 52 53 54 55 56 57
 * We kill the socket, if:
 * 1. If number of orphaned sockets exceeds an administratively configured
 *    limit.
 * 2. If we have strong memory pressure.
 */
static int tcp_out_of_resources(struct sock *sk, int do_reset)
{
	struct tcp_sock *tp = tcp_sk(sk);
58
	int shift = 0;
L
Linus Torvalds 已提交
59

60
	/* If peer does not open window for long time, or did not transmit
L
Linus Torvalds 已提交
61 62
	 * anything for long time, penalize it. */
	if ((s32)(tcp_time_stamp - tp->lsndtime) > 2*TCP_RTO_MAX || !do_reset)
63
		shift++;
L
Linus Torvalds 已提交
64 65 66

	/* If some dubious ICMP arrived, penalize even more. */
	if (sk->sk_err_soft)
67
		shift++;
L
Linus Torvalds 已提交
68

A
Arun Sharma 已提交
69
	if (tcp_check_oom(sk, shift)) {
L
Linus Torvalds 已提交
70 71 72 73 74 75 76 77 78
		/* Catch exceptional cases, when connection requires reset.
		 *      1. Last segment was sent recently. */
		if ((s32)(tcp_time_stamp - tp->lsndtime) <= TCP_TIMEWAIT_LEN ||
		    /*  2. Window is closed. */
		    (!tp->snd_wnd && !tp->packets_out))
			do_reset = 1;
		if (do_reset)
			tcp_send_active_reset(sk, GFP_ATOMIC);
		tcp_done(sk);
79
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPABORTONMEMORY);
L
Linus Torvalds 已提交
80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101
		return 1;
	}
	return 0;
}

/* Calculate maximal number or retries on an orphaned socket. */
static int tcp_orphan_retries(struct sock *sk, int alive)
{
	int retries = sysctl_tcp_orphan_retries; /* May be zero. */

	/* We know from an ICMP that something is wrong. */
	if (sk->sk_err_soft && !alive)
		retries = 0;

	/* However, if socket sent something recently, select some safe
	 * number of retries. 8 corresponds to >100 seconds with minimal
	 * RTO of 200msec. */
	if (retries == 0 && alive)
		retries = 8;
	return retries;
}

102 103 104 105 106 107 108 109 110
static void tcp_mtu_probing(struct inet_connection_sock *icsk, struct sock *sk)
{
	/* Black hole detection */
	if (sysctl_tcp_mtu_probing) {
		if (!icsk->icsk_mtup.enabled) {
			icsk->icsk_mtup.enabled = 1;
			tcp_sync_mss(sk, icsk->icsk_pmtu_cookie);
		} else {
			struct tcp_sock *tp = tcp_sk(sk);
111 112
			int mss;

113
			mss = tcp_mtu_to_mss(sk, icsk->icsk_mtup.search_low) >> 1;
114 115 116 117 118 119 120 121
			mss = min(sysctl_tcp_base_mss, mss);
			mss = max(mss, 68 - tp->tcp_header_len);
			icsk->icsk_mtup.search_low = tcp_mss_to_mtu(sk, mss);
			tcp_sync_mss(sk, icsk->icsk_pmtu_cookie);
		}
	}
}

122
/* This function calculates a "timeout" which is equivalent to the timeout of a
D
Daniel Mack 已提交
123
 * TCP connection after "boundary" unsuccessful, exponentially backed-off
124 125
 * retransmissions with an initial RTO of TCP_RTO_MIN or TCP_TIMEOUT_INIT if
 * syn_set flag is set.
126 127
 */
static bool retransmits_timed_out(struct sock *sk,
128
				  unsigned int boundary,
129
				  unsigned int timeout,
130
				  bool syn_set)
131
{
132
	unsigned int linear_backoff_thresh, start_ts;
133
	unsigned int rto_base = syn_set ? TCP_TIMEOUT_INIT : TCP_RTO_MIN;
134 135 136 137 138 139 140 141 142

	if (!inet_csk(sk)->icsk_retransmits)
		return false;

	if (unlikely(!tcp_sk(sk)->retrans_stamp))
		start_ts = TCP_SKB_CB(tcp_write_queue_head(sk))->when;
	else
		start_ts = tcp_sk(sk)->retrans_stamp;

143
	if (likely(timeout == 0)) {
144
		linear_backoff_thresh = ilog2(TCP_RTO_MAX/rto_base);
145

146
		if (boundary <= linear_backoff_thresh)
147
			timeout = ((2 << boundary) - 1) * rto_base;
148
		else
149
			timeout = ((2 << linear_backoff_thresh) - 1) * rto_base +
150 151
				(boundary - linear_backoff_thresh) * TCP_RTO_MAX;
	}
152 153 154
	return (tcp_time_stamp - start_ts) >= timeout;
}

L
Linus Torvalds 已提交
155 156 157
/* A write timeout has occurred. Process the after effects. */
static int tcp_write_timeout(struct sock *sk)
{
J
John Heffner 已提交
158
	struct inet_connection_sock *icsk = inet_csk(sk);
L
Linus Torvalds 已提交
159
	int retry_until;
160
	bool do_reset, syn_set = false;
L
Linus Torvalds 已提交
161 162

	if ((1 << sk->sk_state) & (TCPF_SYN_SENT | TCPF_SYN_RECV)) {
163
		if (icsk->icsk_retransmits)
E
Eric Dumazet 已提交
164
			dst_negative_advice(sk);
165
		retry_until = icsk->icsk_syn_retries ? : sysctl_tcp_syn_retries;
166
		syn_set = true;
L
Linus Torvalds 已提交
167
	} else {
168
		if (retransmits_timed_out(sk, sysctl_tcp_retries1, 0, 0)) {
J
John Heffner 已提交
169
			/* Black hole detection */
170
			tcp_mtu_probing(icsk, sk);
L
Linus Torvalds 已提交
171

E
Eric Dumazet 已提交
172
			dst_negative_advice(sk);
L
Linus Torvalds 已提交
173 174 175 176
		}

		retry_until = sysctl_tcp_retries2;
		if (sock_flag(sk, SOCK_DEAD)) {
177
			const int alive = (icsk->icsk_rto < TCP_RTO_MAX);
178

L
Linus Torvalds 已提交
179
			retry_until = tcp_orphan_retries(sk, alive);
180
			do_reset = alive ||
181
				!retransmits_timed_out(sk, retry_until, 0, 0);
L
Linus Torvalds 已提交
182

183
			if (tcp_out_of_resources(sk, do_reset))
L
Linus Torvalds 已提交
184 185 186 187
				return 1;
		}
	}

188
	if (retransmits_timed_out(sk, retry_until,
189
				  syn_set ? 0 : icsk->icsk_user_timeout, syn_set)) {
L
Linus Torvalds 已提交
190 191 192 193 194 195 196
		/* Has it gone just too far? */
		tcp_write_err(sk);
		return 1;
	}
	return 0;
}

197
void tcp_delack_timer_handler(struct sock *sk)
L
Linus Torvalds 已提交
198 199
{
	struct tcp_sock *tp = tcp_sk(sk);
200
	struct inet_connection_sock *icsk = inet_csk(sk);
L
Linus Torvalds 已提交
201

202
	sk_mem_reclaim_partial(sk);
L
Linus Torvalds 已提交
203

204
	if (sk->sk_state == TCP_CLOSE || !(icsk->icsk_ack.pending & ICSK_ACK_TIMER))
L
Linus Torvalds 已提交
205 206
		goto out;

207 208
	if (time_after(icsk->icsk_ack.timeout, jiffies)) {
		sk_reset_timer(sk, &icsk->icsk_delack_timer, icsk->icsk_ack.timeout);
L
Linus Torvalds 已提交
209 210
		goto out;
	}
211
	icsk->icsk_ack.pending &= ~ICSK_ACK_TIMER;
L
Linus Torvalds 已提交
212

213
	if (!skb_queue_empty(&tp->ucopy.prequeue)) {
L
Linus Torvalds 已提交
214 215
		struct sk_buff *skb;

216
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPSCHEDULERFAILED);
L
Linus Torvalds 已提交
217 218

		while ((skb = __skb_dequeue(&tp->ucopy.prequeue)) != NULL)
P
Peter Zijlstra 已提交
219
			sk_backlog_rcv(sk, skb);
L
Linus Torvalds 已提交
220 221 222 223

		tp->ucopy.memory = 0;
	}

224 225
	if (inet_csk_ack_scheduled(sk)) {
		if (!icsk->icsk_ack.pingpong) {
L
Linus Torvalds 已提交
226
			/* Delayed ACK missed: inflate ATO. */
227
			icsk->icsk_ack.ato = min(icsk->icsk_ack.ato << 1, icsk->icsk_rto);
L
Linus Torvalds 已提交
228 229 230 231
		} else {
			/* Delayed ACK missed: leave pingpong mode and
			 * deflate ATO.
			 */
232 233
			icsk->icsk_ack.pingpong = 0;
			icsk->icsk_ack.ato      = TCP_ATO_MIN;
L
Linus Torvalds 已提交
234 235
		}
		tcp_send_ack(sk);
236
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_DELAYEDACKS);
L
Linus Torvalds 已提交
237 238 239
	}

out:
240
	if (sk_under_memory_pressure(sk))
241
		sk_mem_reclaim(sk);
242 243 244 245 246 247 248 249 250 251 252 253 254
}

static void tcp_delack_timer(unsigned long data)
{
	struct sock *sk = (struct sock *)data;

	bh_lock_sock(sk);
	if (!sock_owned_by_user(sk)) {
		tcp_delack_timer_handler(sk);
	} else {
		inet_csk(sk)->icsk_ack.blocked = 1;
		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_DELAYEDACKLOCKED);
		/* deleguate our work to tcp_release_cb() */
255 256
		if (!test_and_set_bit(TCP_DELACK_TIMER_DEFERRED, &tcp_sk(sk)->tsq_flags))
			sock_hold(sk);
257
	}
L
Linus Torvalds 已提交
258 259 260 261 262 263
	bh_unlock_sock(sk);
	sock_put(sk);
}

static void tcp_probe_timer(struct sock *sk)
{
264
	struct inet_connection_sock *icsk = inet_csk(sk);
L
Linus Torvalds 已提交
265 266 267
	struct tcp_sock *tp = tcp_sk(sk);
	int max_probes;

268
	if (tp->packets_out || !tcp_send_head(sk)) {
269
		icsk->icsk_probes_out = 0;
L
Linus Torvalds 已提交
270 271 272 273 274 275 276 277 278 279
		return;
	}

	/* *WARNING* RFC 1122 forbids this
	 *
	 * It doesn't AFAIK, because we kill the retransmit timer -AK
	 *
	 * FIXME: We ought not to do it, Solaris 2.5 actually has fixing
	 * this behaviour in Solaris down as a bug fix. [AC]
	 *
280
	 * Let me to explain. icsk_probes_out is zeroed by incoming ACKs
L
Linus Torvalds 已提交
281 282 283 284 285 286 287 288 289 290
	 * even if they advertise zero window. Hence, connection is killed only
	 * if we received no ACKs for normal connection timeout. It is not killed
	 * only because window stays zero for some time, window may be zero
	 * until armageddon and even later. We are in full accordance
	 * with RFCs, only probe timer combines both retransmission timeout
	 * and probe timeout in one bottle.				--ANK
	 */
	max_probes = sysctl_tcp_retries2;

	if (sock_flag(sk, SOCK_DEAD)) {
291
		const int alive = ((icsk->icsk_rto << icsk->icsk_backoff) < TCP_RTO_MAX);
292

L
Linus Torvalds 已提交
293 294
		max_probes = tcp_orphan_retries(sk, alive);

295
		if (tcp_out_of_resources(sk, alive || icsk->icsk_probes_out <= max_probes))
L
Linus Torvalds 已提交
296 297 298
			return;
	}

299
	if (icsk->icsk_probes_out > max_probes) {
L
Linus Torvalds 已提交
300 301 302 303 304 305 306
		tcp_write_err(sk);
	} else {
		/* Only send another probe if we didn't close things up. */
		tcp_send_probe0(sk);
	}
}

307 308 309 310 311 312 313 314 315 316 317 318 319 320
/*
 *	Timer for Fast Open socket to retransmit SYNACK. Note that the
 *	sk here is the child socket, not the parent (listener) socket.
 */
static void tcp_fastopen_synack_timer(struct sock *sk)
{
	struct inet_connection_sock *icsk = inet_csk(sk);
	int max_retries = icsk->icsk_syn_retries ? :
	    sysctl_tcp_synack_retries + 1; /* add one more retry for fastopen */
	struct request_sock *req;

	req = tcp_sk(sk)->fastopen_rsk;
	req->rsk_ops->syn_ack_timeout(sk, req);

321
	if (req->num_timeout >= max_retries) {
322 323 324 325 326 327 328 329
		tcp_write_err(sk);
		return;
	}
	/* XXX (TFO) - Unlike regular SYN-ACK retransmit, we ignore error
	 * returned from rtx_syn_ack() to make it more persistent like
	 * regular retransmit because if the child socket has been accepted
	 * it's not good to give up too easily.
	 */
330 331
	inet_rtx_syn_ack(sk, req);
	req->num_timeout++;
332
	inet_csk_reset_xmit_timer(sk, ICSK_TIME_RETRANS,
333
			  TCP_TIMEOUT_INIT << req->num_timeout, TCP_RTO_MAX);
334 335
}

L
Linus Torvalds 已提交
336 337 338 339
/*
 *	The TCP retransmit timer.
 */

340
void tcp_retransmit_timer(struct sock *sk)
L
Linus Torvalds 已提交
341 342
{
	struct tcp_sock *tp = tcp_sk(sk);
343
	struct inet_connection_sock *icsk = inet_csk(sk);
L
Linus Torvalds 已提交
344

345
	if (tp->fastopen_rsk) {
346 347
		WARN_ON_ONCE(sk->sk_state != TCP_SYN_RECV &&
			     sk->sk_state != TCP_FIN_WAIT1);
348 349 350 351 352 353
		tcp_fastopen_synack_timer(sk);
		/* Before we receive ACK to our SYN-ACK don't retransmit
		 * anything else (e.g., data or FIN segments).
		 */
		return;
	}
L
Linus Torvalds 已提交
354 355 356
	if (!tp->packets_out)
		goto out;

357
	WARN_ON(tcp_write_queue_empty(sk));
L
Linus Torvalds 已提交
358 359 360 361 362 363 364 365

	if (!tp->snd_wnd && !sock_flag(sk, SOCK_DEAD) &&
	    !((1 << sk->sk_state) & (TCPF_SYN_SENT | TCPF_SYN_RECV))) {
		/* Receiver dastardly shrinks window. Our retransmits
		 * become zero probes, but we should not timeout this
		 * connection. If the socket is an orphan, time it out,
		 * we cannot allow such beasts to hang infinitely.
		 */
366 367
		struct inet_sock *inet = inet_sk(sk);
		if (sk->sk_family == AF_INET) {
368 369 370 371
			LIMIT_NETDEBUG(KERN_DEBUG pr_fmt("Peer %pI4:%u/%u unexpectedly shrunk window %u:%u (repaired)\n"),
				       &inet->inet_daddr,
				       ntohs(inet->inet_dport), inet->inet_num,
				       tp->snd_una, tp->snd_nxt);
L
Linus Torvalds 已提交
372
		}
E
Eric Dumazet 已提交
373
#if IS_ENABLED(CONFIG_IPV6)
374 375
		else if (sk->sk_family == AF_INET6) {
			struct ipv6_pinfo *np = inet6_sk(sk);
376 377 378 379
			LIMIT_NETDEBUG(KERN_DEBUG pr_fmt("Peer %pI6:%u/%u unexpectedly shrunk window %u:%u (repaired)\n"),
				       &np->daddr,
				       ntohs(inet->inet_dport), inet->inet_num,
				       tp->snd_una, tp->snd_nxt);
380
		}
L
Linus Torvalds 已提交
381 382 383 384 385 386
#endif
		if (tcp_time_stamp - tp->rcv_tstamp > TCP_RTO_MAX) {
			tcp_write_err(sk);
			goto out;
		}
		tcp_enter_loss(sk, 0);
387
		tcp_retransmit_skb(sk, tcp_write_queue_head(sk));
L
Linus Torvalds 已提交
388 389 390 391 392 393 394
		__sk_dst_reset(sk);
		goto out_reset_timer;
	}

	if (tcp_write_timeout(sk))
		goto out;

395
	if (icsk->icsk_retransmits == 0) {
396 397
		int mib_idx;

398
		if (icsk->icsk_ca_state == TCP_CA_Recovery) {
399 400 401 402
			if (tcp_is_sack(tp))
				mib_idx = LINUX_MIB_TCPSACKRECOVERYFAIL;
			else
				mib_idx = LINUX_MIB_TCPRENORECOVERYFAIL;
403
		} else if (icsk->icsk_ca_state == TCP_CA_Loss) {
404
			mib_idx = LINUX_MIB_TCPLOSSFAILURES;
405 406 407 408 409 410
		} else if ((icsk->icsk_ca_state == TCP_CA_Disorder) ||
			   tp->sacked_out) {
			if (tcp_is_sack(tp))
				mib_idx = LINUX_MIB_TCPSACKFAILURES;
			else
				mib_idx = LINUX_MIB_TCPRENOFAILURES;
L
Linus Torvalds 已提交
411
		} else {
412
			mib_idx = LINUX_MIB_TCPTIMEOUTS;
L
Linus Torvalds 已提交
413
		}
414
		NET_INC_STATS_BH(sock_net(sk), mib_idx);
L
Linus Torvalds 已提交
415 416 417 418 419 420 421 422
	}

	if (tcp_use_frto(sk)) {
		tcp_enter_frto(sk);
	} else {
		tcp_enter_loss(sk, 0);
	}

423
	if (tcp_retransmit_skb(sk, tcp_write_queue_head(sk)) > 0) {
L
Linus Torvalds 已提交
424 425 426
		/* Retransmission failed because of local congestion,
		 * do not backoff.
		 */
427 428 429
		if (!icsk->icsk_retransmits)
			icsk->icsk_retransmits = 1;
		inet_csk_reset_xmit_timer(sk, ICSK_TIME_RETRANS,
430 431
					  min(icsk->icsk_rto, TCP_RESOURCE_PROBE_INTERVAL),
					  TCP_RTO_MAX);
L
Linus Torvalds 已提交
432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449
		goto out;
	}

	/* Increase the timeout each time we retransmit.  Note that
	 * we do not increase the rtt estimate.  rto is initialized
	 * from rtt, but increases here.  Jacobson (SIGCOMM 88) suggests
	 * that doubling rto each time is the least we can get away with.
	 * In KA9Q, Karn uses this for the first few times, and then
	 * goes to quadratic.  netBSD doubles, but only goes up to *64,
	 * and clamps at 1 to 64 sec afterwards.  Note that 120 sec is
	 * defined in the protocol as the maximum possible RTT.  I guess
	 * we'll have to use something other than TCP to talk to the
	 * University of Mars.
	 *
	 * PAWS allows us longer timeouts and large windows, so once
	 * implemented ftp to mars will work nicely. We will have to fix
	 * the 120 second clamps though!
	 */
450 451
	icsk->icsk_backoff++;
	icsk->icsk_retransmits++;
L
Linus Torvalds 已提交
452 453

out_reset_timer:
A
Andreas Petlund 已提交
454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472
	/* If stream is thin, use linear timeouts. Since 'icsk_backoff' is
	 * used to reset timer, set to 0. Recalculate 'icsk_rto' as this
	 * might be increased if the stream oscillates between thin and thick,
	 * thus the old value might already be too high compared to the value
	 * set by 'tcp_set_rto' in tcp_input.c which resets the rto without
	 * backoff. Limit to TCP_THIN_LINEAR_RETRIES before initiating
	 * exponential backoff behaviour to avoid continue hammering
	 * linear-timeout retransmissions into a black hole
	 */
	if (sk->sk_state == TCP_ESTABLISHED &&
	    (tp->thin_lto || sysctl_tcp_thin_linear_timeouts) &&
	    tcp_stream_is_thin(tp) &&
	    icsk->icsk_retransmits <= TCP_THIN_LINEAR_RETRIES) {
		icsk->icsk_backoff = 0;
		icsk->icsk_rto = min(__tcp_set_rto(tp), TCP_RTO_MAX);
	} else {
		/* Use normal (exponential) backoff */
		icsk->icsk_rto = min(icsk->icsk_rto << 1, TCP_RTO_MAX);
	}
473
	inet_csk_reset_xmit_timer(sk, ICSK_TIME_RETRANS, icsk->icsk_rto, TCP_RTO_MAX);
474
	if (retransmits_timed_out(sk, sysctl_tcp_retries1 + 1, 0, 0))
L
Linus Torvalds 已提交
475 476 477 478 479
		__sk_dst_reset(sk);

out:;
}

480
void tcp_write_timer_handler(struct sock *sk)
L
Linus Torvalds 已提交
481
{
482
	struct inet_connection_sock *icsk = inet_csk(sk);
L
Linus Torvalds 已提交
483 484
	int event;

485
	if (sk->sk_state == TCP_CLOSE || !icsk->icsk_pending)
L
Linus Torvalds 已提交
486 487
		goto out;

488 489
	if (time_after(icsk->icsk_timeout, jiffies)) {
		sk_reset_timer(sk, &icsk->icsk_retransmit_timer, icsk->icsk_timeout);
L
Linus Torvalds 已提交
490 491 492
		goto out;
	}

493
	event = icsk->icsk_pending;
L
Linus Torvalds 已提交
494 495

	switch (event) {
N
Nandita Dukkipati 已提交
496 497 498 499 500 501
	case ICSK_TIME_EARLY_RETRANS:
		tcp_resume_early_retransmit(sk);
		break;
	case ICSK_TIME_LOSS_PROBE:
		tcp_send_loss_probe(sk);
		break;
502
	case ICSK_TIME_RETRANS:
N
Nandita Dukkipati 已提交
503
		icsk->icsk_pending = 0;
L
Linus Torvalds 已提交
504 505
		tcp_retransmit_timer(sk);
		break;
506
	case ICSK_TIME_PROBE0:
N
Nandita Dukkipati 已提交
507
		icsk->icsk_pending = 0;
L
Linus Torvalds 已提交
508 509 510 511 512
		tcp_probe_timer(sk);
		break;
	}

out:
513
	sk_mem_reclaim(sk);
514 515 516 517 518 519 520 521 522 523 524
}

static void tcp_write_timer(unsigned long data)
{
	struct sock *sk = (struct sock *)data;

	bh_lock_sock(sk);
	if (!sock_owned_by_user(sk)) {
		tcp_write_timer_handler(sk);
	} else {
		/* deleguate our work to tcp_release_cb() */
525 526
		if (!test_and_set_bit(TCP_WRITE_TIMER_DEFERRED, &tcp_sk(sk)->tsq_flags))
			sock_hold(sk);
527
	}
L
Linus Torvalds 已提交
528 529 530 531
	bh_unlock_sock(sk);
	sock_put(sk);
}

532 533 534 535 536 537
/*
 *	Timer for listening sockets
 */

static void tcp_synack_timer(struct sock *sk)
{
538 539
	inet_csk_reqsk_queue_prune(sk, TCP_SYNQ_INTERVAL,
				   TCP_TIMEOUT_INIT, TCP_RTO_MAX);
L
Linus Torvalds 已提交
540 541
}

542 543 544 545 546 547
void tcp_syn_ack_timeout(struct sock *sk, struct request_sock *req)
{
	NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPTIMEOUTS);
}
EXPORT_SYMBOL(tcp_syn_ack_timeout);

L
Linus Torvalds 已提交
548 549 550 551 552 553
void tcp_set_keepalive(struct sock *sk, int val)
{
	if ((1 << sk->sk_state) & (TCPF_CLOSE | TCPF_LISTEN))
		return;

	if (val && !sock_flag(sk, SOCK_KEEPOPEN))
554
		inet_csk_reset_keepalive_timer(sk, keepalive_time_when(tcp_sk(sk)));
L
Linus Torvalds 已提交
555
	else if (!val)
556
		inet_csk_delete_keepalive_timer(sk);
L
Linus Torvalds 已提交
557 558 559 560 561 562
}


static void tcp_keepalive_timer (unsigned long data)
{
	struct sock *sk = (struct sock *) data;
563
	struct inet_connection_sock *icsk = inet_csk(sk);
L
Linus Torvalds 已提交
564
	struct tcp_sock *tp = tcp_sk(sk);
565
	u32 elapsed;
L
Linus Torvalds 已提交
566 567 568 569

	/* Only process if socket is not in use. */
	bh_lock_sock(sk);
	if (sock_owned_by_user(sk)) {
570
		/* Try again later. */
571
		inet_csk_reset_keepalive_timer (sk, HZ/20);
L
Linus Torvalds 已提交
572 573 574 575 576 577 578 579 580 581
		goto out;
	}

	if (sk->sk_state == TCP_LISTEN) {
		tcp_synack_timer(sk);
		goto out;
	}

	if (sk->sk_state == TCP_FIN_WAIT2 && sock_flag(sk, SOCK_DEAD)) {
		if (tp->linger2 >= 0) {
582
			const int tmo = tcp_fin_time(sk) - TCP_TIMEWAIT_LEN;
L
Linus Torvalds 已提交
583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598

			if (tmo > 0) {
				tcp_time_wait(sk, TCP_FIN_WAIT2, tmo);
				goto out;
			}
		}
		tcp_send_active_reset(sk, GFP_ATOMIC);
		goto death;
	}

	if (!sock_flag(sk, SOCK_KEEPOPEN) || sk->sk_state == TCP_CLOSE)
		goto out;

	elapsed = keepalive_time_when(tp);

	/* It is alive without keepalive 8) */
599
	if (tp->packets_out || tcp_send_head(sk))
L
Linus Torvalds 已提交
600 601
		goto resched;

602
	elapsed = keepalive_time_elapsed(tp);
L
Linus Torvalds 已提交
603 604

	if (elapsed >= keepalive_time_when(tp)) {
605 606 607 608 609 610 611 612
		/* If the TCP_USER_TIMEOUT option is enabled, use that
		 * to determine when to timeout instead.
		 */
		if ((icsk->icsk_user_timeout != 0 &&
		    elapsed >= icsk->icsk_user_timeout &&
		    icsk->icsk_probes_out > 0) ||
		    (icsk->icsk_user_timeout == 0 &&
		    icsk->icsk_probes_out >= keepalive_probes(tp))) {
L
Linus Torvalds 已提交
613 614 615 616 617
			tcp_send_active_reset(sk, GFP_ATOMIC);
			tcp_write_err(sk);
			goto out;
		}
		if (tcp_write_wakeup(sk) <= 0) {
618
			icsk->icsk_probes_out++;
L
Linus Torvalds 已提交
619 620 621 622 623 624 625 626 627 628 629 630
			elapsed = keepalive_intvl_when(tp);
		} else {
			/* If keepalive was lost due to local congestion,
			 * try harder.
			 */
			elapsed = TCP_RESOURCE_PROBE_INTERVAL;
		}
	} else {
		/* It is tp->rcv_tstamp + keepalive_time_when(tp) */
		elapsed = keepalive_time_when(tp) - elapsed;
	}

631
	sk_mem_reclaim(sk);
L
Linus Torvalds 已提交
632 633

resched:
634
	inet_csk_reset_keepalive_timer (sk, elapsed);
L
Linus Torvalds 已提交
635 636
	goto out;

637
death:
L
Linus Torvalds 已提交
638 639 640 641 642 643
	tcp_done(sk);

out:
	bh_unlock_sock(sk);
	sock_put(sk);
}
644 645 646 647 648 649 650

void tcp_init_xmit_timers(struct sock *sk)
{
	inet_csk_init_xmit_timers(sk, &tcp_write_timer, &tcp_delack_timer,
				  &tcp_keepalive_timer);
}
EXPORT_SYMBOL(tcp_init_xmit_timers);