xref: /openbmc/linux/net/ipv4/tcp_timer.c (revision 588b48ca)
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  *		Implementation of the Transmission Control Protocol(TCP).
7  *
8  * Authors:	Ross Biro
9  *		Fred N. van Kempen, <waltje@uWalt.NL.Mugnet.ORG>
10  *		Mark Evans, <evansmp@uhura.aston.ac.uk>
11  *		Corey Minyard <wf-rch!minyard@relay.EU.net>
12  *		Florian La Roche, <flla@stud.uni-sb.de>
13  *		Charles Hedrick, <hedrick@klinzhai.rutgers.edu>
14  *		Linus Torvalds, <torvalds@cs.helsinki.fi>
15  *		Alan Cox, <gw4pts@gw4pts.ampr.org>
16  *		Matthew Dillon, <dillon@apollo.west.oic.com>
17  *		Arnt Gulbrandsen, <agulbra@nvg.unit.no>
18  *		Jorge Cwik, <jorge@laser.satlink.net>
19  */
20 
21 #include <linux/module.h>
22 #include <linux/gfp.h>
23 #include <net/tcp.h>
24 
25 int sysctl_tcp_syn_retries __read_mostly = TCP_SYN_RETRIES;
26 int sysctl_tcp_synack_retries __read_mostly = TCP_SYNACK_RETRIES;
27 int sysctl_tcp_keepalive_time __read_mostly = TCP_KEEPALIVE_TIME;
28 int sysctl_tcp_keepalive_probes __read_mostly = TCP_KEEPALIVE_PROBES;
29 int sysctl_tcp_keepalive_intvl __read_mostly = TCP_KEEPALIVE_INTVL;
30 int sysctl_tcp_retries1 __read_mostly = TCP_RETR1;
31 int sysctl_tcp_retries2 __read_mostly = TCP_RETR2;
32 int sysctl_tcp_orphan_retries __read_mostly;
33 int sysctl_tcp_thin_linear_timeouts __read_mostly;
34 
35 static void tcp_write_err(struct sock *sk)
36 {
37 	sk->sk_err = sk->sk_err_soft ? : ETIMEDOUT;
38 	sk->sk_error_report(sk);
39 
40 	tcp_done(sk);
41 	NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPABORTONTIMEOUT);
42 }
43 
44 /* Do not allow orphaned sockets to eat all our resources.
45  * This is direct violation of TCP specs, but it is required
46  * to prevent DoS attacks. It is called when a retransmission timeout
47  * or zero probe timeout occurs on orphaned socket.
48  *
49  * Criteria is still not confirmed experimentally and may change.
50  * We kill the socket, if:
51  * 1. If number of orphaned sockets exceeds an administratively configured
52  *    limit.
53  * 2. If we have strong memory pressure.
54  */
55 static int tcp_out_of_resources(struct sock *sk, int do_reset)
56 {
57 	struct tcp_sock *tp = tcp_sk(sk);
58 	int shift = 0;
59 
60 	/* If peer does not open window for long time, or did not transmit
61 	 * anything for long time, penalize it. */
62 	if ((s32)(tcp_time_stamp - tp->lsndtime) > 2*TCP_RTO_MAX || !do_reset)
63 		shift++;
64 
65 	/* If some dubious ICMP arrived, penalize even more. */
66 	if (sk->sk_err_soft)
67 		shift++;
68 
69 	if (tcp_check_oom(sk, shift)) {
70 		/* Catch exceptional cases, when connection requires reset.
71 		 *      1. Last segment was sent recently. */
72 		if ((s32)(tcp_time_stamp - tp->lsndtime) <= TCP_TIMEWAIT_LEN ||
73 		    /*  2. Window is closed. */
74 		    (!tp->snd_wnd && !tp->packets_out))
75 			do_reset = 1;
76 		if (do_reset)
77 			tcp_send_active_reset(sk, GFP_ATOMIC);
78 		tcp_done(sk);
79 		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPABORTONMEMORY);
80 		return 1;
81 	}
82 	return 0;
83 }
84 
85 /* Calculate maximal number or retries on an orphaned socket. */
86 static int tcp_orphan_retries(struct sock *sk, int alive)
87 {
88 	int retries = sysctl_tcp_orphan_retries; /* May be zero. */
89 
90 	/* We know from an ICMP that something is wrong. */
91 	if (sk->sk_err_soft && !alive)
92 		retries = 0;
93 
94 	/* However, if socket sent something recently, select some safe
95 	 * number of retries. 8 corresponds to >100 seconds with minimal
96 	 * RTO of 200msec. */
97 	if (retries == 0 && alive)
98 		retries = 8;
99 	return retries;
100 }
101 
102 static void tcp_mtu_probing(struct inet_connection_sock *icsk, struct sock *sk)
103 {
104 	/* Black hole detection */
105 	if (sysctl_tcp_mtu_probing) {
106 		if (!icsk->icsk_mtup.enabled) {
107 			icsk->icsk_mtup.enabled = 1;
108 			tcp_sync_mss(sk, icsk->icsk_pmtu_cookie);
109 		} else {
110 			struct tcp_sock *tp = tcp_sk(sk);
111 			int mss;
112 
113 			mss = tcp_mtu_to_mss(sk, icsk->icsk_mtup.search_low) >> 1;
114 			mss = min(sysctl_tcp_base_mss, mss);
115 			mss = max(mss, 68 - tp->tcp_header_len);
116 			icsk->icsk_mtup.search_low = tcp_mss_to_mtu(sk, mss);
117 			tcp_sync_mss(sk, icsk->icsk_pmtu_cookie);
118 		}
119 	}
120 }
121 
122 /* This function calculates a "timeout" which is equivalent to the timeout of a
123  * TCP connection after "boundary" unsuccessful, exponentially backed-off
124  * retransmissions with an initial RTO of TCP_RTO_MIN or TCP_TIMEOUT_INIT if
125  * 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 	if (!inet_csk(sk)->icsk_retransmits)
136 		return false;
137 
138 	if (unlikely(!tcp_sk(sk)->retrans_stamp))
139 		start_ts = TCP_SKB_CB(tcp_write_queue_head(sk))->when;
140 	else
141 		start_ts = tcp_sk(sk)->retrans_stamp;
142 
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 				(boundary - linear_backoff_thresh) * TCP_RTO_MAX;
151 	}
152 	return (tcp_time_stamp - start_ts) >= timeout;
153 }
154 
155 /* A write timeout has occurred. Process the after effects. */
156 static int tcp_write_timeout(struct sock *sk)
157 {
158 	struct inet_connection_sock *icsk = inet_csk(sk);
159 	struct tcp_sock *tp = tcp_sk(sk);
160 	int retry_until;
161 	bool do_reset, syn_set = false;
162 
163 	if ((1 << sk->sk_state) & (TCPF_SYN_SENT | TCPF_SYN_RECV)) {
164 		if (icsk->icsk_retransmits) {
165 			dst_negative_advice(sk);
166 			if (tp->syn_fastopen || tp->syn_data)
167 				tcp_fastopen_cache_set(sk, 0, NULL, true);
168 			if (tp->syn_data)
169 				NET_INC_STATS_BH(sock_net(sk),
170 						 LINUX_MIB_TCPFASTOPENACTIVEFAIL);
171 		}
172 		retry_until = icsk->icsk_syn_retries ? : sysctl_tcp_syn_retries;
173 		syn_set = true;
174 	} else {
175 		if (retransmits_timed_out(sk, sysctl_tcp_retries1, 0, 0)) {
176 			/* Black hole detection */
177 			tcp_mtu_probing(icsk, sk);
178 
179 			dst_negative_advice(sk);
180 		}
181 
182 		retry_until = sysctl_tcp_retries2;
183 		if (sock_flag(sk, SOCK_DEAD)) {
184 			const int alive = (icsk->icsk_rto < TCP_RTO_MAX);
185 
186 			retry_until = tcp_orphan_retries(sk, alive);
187 			do_reset = alive ||
188 				!retransmits_timed_out(sk, retry_until, 0, 0);
189 
190 			if (tcp_out_of_resources(sk, do_reset))
191 				return 1;
192 		}
193 	}
194 
195 	if (retransmits_timed_out(sk, retry_until,
196 				  syn_set ? 0 : icsk->icsk_user_timeout, syn_set)) {
197 		/* Has it gone just too far? */
198 		tcp_write_err(sk);
199 		return 1;
200 	}
201 	return 0;
202 }
203 
204 void tcp_delack_timer_handler(struct sock *sk)
205 {
206 	struct tcp_sock *tp = tcp_sk(sk);
207 	struct inet_connection_sock *icsk = inet_csk(sk);
208 
209 	sk_mem_reclaim_partial(sk);
210 
211 	if (sk->sk_state == TCP_CLOSE || !(icsk->icsk_ack.pending & ICSK_ACK_TIMER))
212 		goto out;
213 
214 	if (time_after(icsk->icsk_ack.timeout, jiffies)) {
215 		sk_reset_timer(sk, &icsk->icsk_delack_timer, icsk->icsk_ack.timeout);
216 		goto out;
217 	}
218 	icsk->icsk_ack.pending &= ~ICSK_ACK_TIMER;
219 
220 	if (!skb_queue_empty(&tp->ucopy.prequeue)) {
221 		struct sk_buff *skb;
222 
223 		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPSCHEDULERFAILED);
224 
225 		while ((skb = __skb_dequeue(&tp->ucopy.prequeue)) != NULL)
226 			sk_backlog_rcv(sk, skb);
227 
228 		tp->ucopy.memory = 0;
229 	}
230 
231 	if (inet_csk_ack_scheduled(sk)) {
232 		if (!icsk->icsk_ack.pingpong) {
233 			/* Delayed ACK missed: inflate ATO. */
234 			icsk->icsk_ack.ato = min(icsk->icsk_ack.ato << 1, icsk->icsk_rto);
235 		} else {
236 			/* Delayed ACK missed: leave pingpong mode and
237 			 * deflate ATO.
238 			 */
239 			icsk->icsk_ack.pingpong = 0;
240 			icsk->icsk_ack.ato      = TCP_ATO_MIN;
241 		}
242 		tcp_send_ack(sk);
243 		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_DELAYEDACKS);
244 	}
245 
246 out:
247 	if (sk_under_memory_pressure(sk))
248 		sk_mem_reclaim(sk);
249 }
250 
251 static void tcp_delack_timer(unsigned long data)
252 {
253 	struct sock *sk = (struct sock *)data;
254 
255 	bh_lock_sock(sk);
256 	if (!sock_owned_by_user(sk)) {
257 		tcp_delack_timer_handler(sk);
258 	} else {
259 		inet_csk(sk)->icsk_ack.blocked = 1;
260 		NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_DELAYEDACKLOCKED);
261 		/* deleguate our work to tcp_release_cb() */
262 		if (!test_and_set_bit(TCP_DELACK_TIMER_DEFERRED, &tcp_sk(sk)->tsq_flags))
263 			sock_hold(sk);
264 	}
265 	bh_unlock_sock(sk);
266 	sock_put(sk);
267 }
268 
269 static void tcp_probe_timer(struct sock *sk)
270 {
271 	struct inet_connection_sock *icsk = inet_csk(sk);
272 	struct tcp_sock *tp = tcp_sk(sk);
273 	int max_probes;
274 
275 	if (tp->packets_out || !tcp_send_head(sk)) {
276 		icsk->icsk_probes_out = 0;
277 		return;
278 	}
279 
280 	/* *WARNING* RFC 1122 forbids this
281 	 *
282 	 * It doesn't AFAIK, because we kill the retransmit timer -AK
283 	 *
284 	 * FIXME: We ought not to do it, Solaris 2.5 actually has fixing
285 	 * this behaviour in Solaris down as a bug fix. [AC]
286 	 *
287 	 * Let me to explain. icsk_probes_out is zeroed by incoming ACKs
288 	 * even if they advertise zero window. Hence, connection is killed only
289 	 * if we received no ACKs for normal connection timeout. It is not killed
290 	 * only because window stays zero for some time, window may be zero
291 	 * until armageddon and even later. We are in full accordance
292 	 * with RFCs, only probe timer combines both retransmission timeout
293 	 * and probe timeout in one bottle.				--ANK
294 	 */
295 	max_probes = sysctl_tcp_retries2;
296 
297 	if (sock_flag(sk, SOCK_DEAD)) {
298 		const int alive = ((icsk->icsk_rto << icsk->icsk_backoff) < TCP_RTO_MAX);
299 
300 		max_probes = tcp_orphan_retries(sk, alive);
301 
302 		if (tcp_out_of_resources(sk, alive || icsk->icsk_probes_out <= max_probes))
303 			return;
304 	}
305 
306 	if (icsk->icsk_probes_out > max_probes) {
307 		tcp_write_err(sk);
308 	} else {
309 		/* Only send another probe if we didn't close things up. */
310 		tcp_send_probe0(sk);
311 	}
312 }
313 
314 /*
315  *	Timer for Fast Open socket to retransmit SYNACK. Note that the
316  *	sk here is the child socket, not the parent (listener) socket.
317  */
318 static void tcp_fastopen_synack_timer(struct sock *sk)
319 {
320 	struct inet_connection_sock *icsk = inet_csk(sk);
321 	int max_retries = icsk->icsk_syn_retries ? :
322 	    sysctl_tcp_synack_retries + 1; /* add one more retry for fastopen */
323 	struct request_sock *req;
324 
325 	req = tcp_sk(sk)->fastopen_rsk;
326 	req->rsk_ops->syn_ack_timeout(sk, req);
327 
328 	if (req->num_timeout >= max_retries) {
329 		tcp_write_err(sk);
330 		return;
331 	}
332 	/* XXX (TFO) - Unlike regular SYN-ACK retransmit, we ignore error
333 	 * returned from rtx_syn_ack() to make it more persistent like
334 	 * regular retransmit because if the child socket has been accepted
335 	 * it's not good to give up too easily.
336 	 */
337 	inet_rtx_syn_ack(sk, req);
338 	req->num_timeout++;
339 	inet_csk_reset_xmit_timer(sk, ICSK_TIME_RETRANS,
340 			  TCP_TIMEOUT_INIT << req->num_timeout, TCP_RTO_MAX);
341 }
342 
343 /*
344  *	The TCP retransmit timer.
345  */
346 
347 void tcp_retransmit_timer(struct sock *sk)
348 {
349 	struct tcp_sock *tp = tcp_sk(sk);
350 	struct inet_connection_sock *icsk = inet_csk(sk);
351 
352 	if (tp->fastopen_rsk) {
353 		WARN_ON_ONCE(sk->sk_state != TCP_SYN_RECV &&
354 			     sk->sk_state != TCP_FIN_WAIT1);
355 		tcp_fastopen_synack_timer(sk);
356 		/* Before we receive ACK to our SYN-ACK don't retransmit
357 		 * anything else (e.g., data or FIN segments).
358 		 */
359 		return;
360 	}
361 	if (!tp->packets_out)
362 		goto out;
363 
364 	WARN_ON(tcp_write_queue_empty(sk));
365 
366 	tp->tlp_high_seq = 0;
367 
368 	if (!tp->snd_wnd && !sock_flag(sk, SOCK_DEAD) &&
369 	    !((1 << sk->sk_state) & (TCPF_SYN_SENT | TCPF_SYN_RECV))) {
370 		/* Receiver dastardly shrinks window. Our retransmits
371 		 * become zero probes, but we should not timeout this
372 		 * connection. If the socket is an orphan, time it out,
373 		 * we cannot allow such beasts to hang infinitely.
374 		 */
375 		struct inet_sock *inet = inet_sk(sk);
376 		if (sk->sk_family == AF_INET) {
377 			LIMIT_NETDEBUG(KERN_DEBUG pr_fmt("Peer %pI4:%u/%u unexpectedly shrunk window %u:%u (repaired)\n"),
378 				       &inet->inet_daddr,
379 				       ntohs(inet->inet_dport), inet->inet_num,
380 				       tp->snd_una, tp->snd_nxt);
381 		}
382 #if IS_ENABLED(CONFIG_IPV6)
383 		else if (sk->sk_family == AF_INET6) {
384 			LIMIT_NETDEBUG(KERN_DEBUG pr_fmt("Peer %pI6:%u/%u unexpectedly shrunk window %u:%u (repaired)\n"),
385 				       &sk->sk_v6_daddr,
386 				       ntohs(inet->inet_dport), inet->inet_num,
387 				       tp->snd_una, tp->snd_nxt);
388 		}
389 #endif
390 		if (tcp_time_stamp - tp->rcv_tstamp > TCP_RTO_MAX) {
391 			tcp_write_err(sk);
392 			goto out;
393 		}
394 		tcp_enter_loss(sk);
395 		tcp_retransmit_skb(sk, tcp_write_queue_head(sk));
396 		__sk_dst_reset(sk);
397 		goto out_reset_timer;
398 	}
399 
400 	if (tcp_write_timeout(sk))
401 		goto out;
402 
403 	if (icsk->icsk_retransmits == 0) {
404 		int mib_idx;
405 
406 		if (icsk->icsk_ca_state == TCP_CA_Recovery) {
407 			if (tcp_is_sack(tp))
408 				mib_idx = LINUX_MIB_TCPSACKRECOVERYFAIL;
409 			else
410 				mib_idx = LINUX_MIB_TCPRENORECOVERYFAIL;
411 		} else if (icsk->icsk_ca_state == TCP_CA_Loss) {
412 			mib_idx = LINUX_MIB_TCPLOSSFAILURES;
413 		} else if ((icsk->icsk_ca_state == TCP_CA_Disorder) ||
414 			   tp->sacked_out) {
415 			if (tcp_is_sack(tp))
416 				mib_idx = LINUX_MIB_TCPSACKFAILURES;
417 			else
418 				mib_idx = LINUX_MIB_TCPRENOFAILURES;
419 		} else {
420 			mib_idx = LINUX_MIB_TCPTIMEOUTS;
421 		}
422 		NET_INC_STATS_BH(sock_net(sk), mib_idx);
423 	}
424 
425 	tcp_enter_loss(sk);
426 
427 	if (tcp_retransmit_skb(sk, tcp_write_queue_head(sk)) > 0) {
428 		/* Retransmission failed because of local congestion,
429 		 * do not backoff.
430 		 */
431 		if (!icsk->icsk_retransmits)
432 			icsk->icsk_retransmits = 1;
433 		inet_csk_reset_xmit_timer(sk, ICSK_TIME_RETRANS,
434 					  min(icsk->icsk_rto, TCP_RESOURCE_PROBE_INTERVAL),
435 					  TCP_RTO_MAX);
436 		goto out;
437 	}
438 
439 	/* Increase the timeout each time we retransmit.  Note that
440 	 * we do not increase the rtt estimate.  rto is initialized
441 	 * from rtt, but increases here.  Jacobson (SIGCOMM 88) suggests
442 	 * that doubling rto each time is the least we can get away with.
443 	 * In KA9Q, Karn uses this for the first few times, and then
444 	 * goes to quadratic.  netBSD doubles, but only goes up to *64,
445 	 * and clamps at 1 to 64 sec afterwards.  Note that 120 sec is
446 	 * defined in the protocol as the maximum possible RTT.  I guess
447 	 * we'll have to use something other than TCP to talk to the
448 	 * University of Mars.
449 	 *
450 	 * PAWS allows us longer timeouts and large windows, so once
451 	 * implemented ftp to mars will work nicely. We will have to fix
452 	 * the 120 second clamps though!
453 	 */
454 	icsk->icsk_backoff++;
455 	icsk->icsk_retransmits++;
456 
457 out_reset_timer:
458 	/* If stream is thin, use linear timeouts. Since 'icsk_backoff' is
459 	 * used to reset timer, set to 0. Recalculate 'icsk_rto' as this
460 	 * might be increased if the stream oscillates between thin and thick,
461 	 * thus the old value might already be too high compared to the value
462 	 * set by 'tcp_set_rto' in tcp_input.c which resets the rto without
463 	 * backoff. Limit to TCP_THIN_LINEAR_RETRIES before initiating
464 	 * exponential backoff behaviour to avoid continue hammering
465 	 * linear-timeout retransmissions into a black hole
466 	 */
467 	if (sk->sk_state == TCP_ESTABLISHED &&
468 	    (tp->thin_lto || sysctl_tcp_thin_linear_timeouts) &&
469 	    tcp_stream_is_thin(tp) &&
470 	    icsk->icsk_retransmits <= TCP_THIN_LINEAR_RETRIES) {
471 		icsk->icsk_backoff = 0;
472 		icsk->icsk_rto = min(__tcp_set_rto(tp), TCP_RTO_MAX);
473 	} else {
474 		/* Use normal (exponential) backoff */
475 		icsk->icsk_rto = min(icsk->icsk_rto << 1, TCP_RTO_MAX);
476 	}
477 	inet_csk_reset_xmit_timer(sk, ICSK_TIME_RETRANS, icsk->icsk_rto, TCP_RTO_MAX);
478 	if (retransmits_timed_out(sk, sysctl_tcp_retries1 + 1, 0, 0))
479 		__sk_dst_reset(sk);
480 
481 out:;
482 }
483 
484 void tcp_write_timer_handler(struct sock *sk)
485 {
486 	struct inet_connection_sock *icsk = inet_csk(sk);
487 	int event;
488 
489 	if (sk->sk_state == TCP_CLOSE || !icsk->icsk_pending)
490 		goto out;
491 
492 	if (time_after(icsk->icsk_timeout, jiffies)) {
493 		sk_reset_timer(sk, &icsk->icsk_retransmit_timer, icsk->icsk_timeout);
494 		goto out;
495 	}
496 
497 	event = icsk->icsk_pending;
498 
499 	switch (event) {
500 	case ICSK_TIME_EARLY_RETRANS:
501 		tcp_resume_early_retransmit(sk);
502 		break;
503 	case ICSK_TIME_LOSS_PROBE:
504 		tcp_send_loss_probe(sk);
505 		break;
506 	case ICSK_TIME_RETRANS:
507 		icsk->icsk_pending = 0;
508 		tcp_retransmit_timer(sk);
509 		break;
510 	case ICSK_TIME_PROBE0:
511 		icsk->icsk_pending = 0;
512 		tcp_probe_timer(sk);
513 		break;
514 	}
515 
516 out:
517 	sk_mem_reclaim(sk);
518 }
519 
520 static void tcp_write_timer(unsigned long data)
521 {
522 	struct sock *sk = (struct sock *)data;
523 
524 	bh_lock_sock(sk);
525 	if (!sock_owned_by_user(sk)) {
526 		tcp_write_timer_handler(sk);
527 	} else {
528 		/* deleguate our work to tcp_release_cb() */
529 		if (!test_and_set_bit(TCP_WRITE_TIMER_DEFERRED, &tcp_sk(sk)->tsq_flags))
530 			sock_hold(sk);
531 	}
532 	bh_unlock_sock(sk);
533 	sock_put(sk);
534 }
535 
536 /*
537  *	Timer for listening sockets
538  */
539 
540 static void tcp_synack_timer(struct sock *sk)
541 {
542 	inet_csk_reqsk_queue_prune(sk, TCP_SYNQ_INTERVAL,
543 				   TCP_TIMEOUT_INIT, TCP_RTO_MAX);
544 }
545 
546 void tcp_syn_ack_timeout(struct sock *sk, struct request_sock *req)
547 {
548 	NET_INC_STATS_BH(sock_net(sk), LINUX_MIB_TCPTIMEOUTS);
549 }
550 EXPORT_SYMBOL(tcp_syn_ack_timeout);
551 
552 void tcp_set_keepalive(struct sock *sk, int val)
553 {
554 	if ((1 << sk->sk_state) & (TCPF_CLOSE | TCPF_LISTEN))
555 		return;
556 
557 	if (val && !sock_flag(sk, SOCK_KEEPOPEN))
558 		inet_csk_reset_keepalive_timer(sk, keepalive_time_when(tcp_sk(sk)));
559 	else if (!val)
560 		inet_csk_delete_keepalive_timer(sk);
561 }
562 
563 
564 static void tcp_keepalive_timer (unsigned long data)
565 {
566 	struct sock *sk = (struct sock *) data;
567 	struct inet_connection_sock *icsk = inet_csk(sk);
568 	struct tcp_sock *tp = tcp_sk(sk);
569 	u32 elapsed;
570 
571 	/* Only process if socket is not in use. */
572 	bh_lock_sock(sk);
573 	if (sock_owned_by_user(sk)) {
574 		/* Try again later. */
575 		inet_csk_reset_keepalive_timer (sk, HZ/20);
576 		goto out;
577 	}
578 
579 	if (sk->sk_state == TCP_LISTEN) {
580 		tcp_synack_timer(sk);
581 		goto out;
582 	}
583 
584 	if (sk->sk_state == TCP_FIN_WAIT2 && sock_flag(sk, SOCK_DEAD)) {
585 		if (tp->linger2 >= 0) {
586 			const int tmo = tcp_fin_time(sk) - TCP_TIMEWAIT_LEN;
587 
588 			if (tmo > 0) {
589 				tcp_time_wait(sk, TCP_FIN_WAIT2, tmo);
590 				goto out;
591 			}
592 		}
593 		tcp_send_active_reset(sk, GFP_ATOMIC);
594 		goto death;
595 	}
596 
597 	if (!sock_flag(sk, SOCK_KEEPOPEN) || sk->sk_state == TCP_CLOSE)
598 		goto out;
599 
600 	elapsed = keepalive_time_when(tp);
601 
602 	/* It is alive without keepalive 8) */
603 	if (tp->packets_out || tcp_send_head(sk))
604 		goto resched;
605 
606 	elapsed = keepalive_time_elapsed(tp);
607 
608 	if (elapsed >= keepalive_time_when(tp)) {
609 		/* If the TCP_USER_TIMEOUT option is enabled, use that
610 		 * to determine when to timeout instead.
611 		 */
612 		if ((icsk->icsk_user_timeout != 0 &&
613 		    elapsed >= icsk->icsk_user_timeout &&
614 		    icsk->icsk_probes_out > 0) ||
615 		    (icsk->icsk_user_timeout == 0 &&
616 		    icsk->icsk_probes_out >= keepalive_probes(tp))) {
617 			tcp_send_active_reset(sk, GFP_ATOMIC);
618 			tcp_write_err(sk);
619 			goto out;
620 		}
621 		if (tcp_write_wakeup(sk) <= 0) {
622 			icsk->icsk_probes_out++;
623 			elapsed = keepalive_intvl_when(tp);
624 		} else {
625 			/* If keepalive was lost due to local congestion,
626 			 * try harder.
627 			 */
628 			elapsed = TCP_RESOURCE_PROBE_INTERVAL;
629 		}
630 	} else {
631 		/* It is tp->rcv_tstamp + keepalive_time_when(tp) */
632 		elapsed = keepalive_time_when(tp) - elapsed;
633 	}
634 
635 	sk_mem_reclaim(sk);
636 
637 resched:
638 	inet_csk_reset_keepalive_timer (sk, elapsed);
639 	goto out;
640 
641 death:
642 	tcp_done(sk);
643 
644 out:
645 	bh_unlock_sock(sk);
646 	sock_put(sk);
647 }
648 
649 void tcp_init_xmit_timers(struct sock *sk)
650 {
651 	inet_csk_init_xmit_timers(sk, &tcp_write_timer, &tcp_delack_timer,
652 				  &tcp_keepalive_timer);
653 }
654 EXPORT_SYMBOL(tcp_init_xmit_timers);
655