xref: /netbsd-src/sys/netinet/tcp_timer.c (revision 313c6c94c424eed90c7b7e494aa83308a0a5d0ce)
1 /*	$NetBSD: tcp_timer.c,v 1.84 2008/11/10 01:06:43 uebayasi Exp $	*/
2 
3 /*
4  * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
5  * All rights reserved.
6  *
7  * Redistribution and use in source and binary forms, with or without
8  * modification, are permitted provided that the following conditions
9  * are met:
10  * 1. Redistributions of source code must retain the above copyright
11  *    notice, this list of conditions and the following disclaimer.
12  * 2. Redistributions in binary form must reproduce the above copyright
13  *    notice, this list of conditions and the following disclaimer in the
14  *    documentation and/or other materials provided with the distribution.
15  * 3. Neither the name of the project nor the names of its contributors
16  *    may be used to endorse or promote products derived from this software
17  *    without specific prior written permission.
18  *
19  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
20  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
23  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29  * SUCH DAMAGE.
30  */
31 
32 /*-
33  * Copyright (c) 1997, 1998, 2001, 2005 The NetBSD Foundation, Inc.
34  * All rights reserved.
35  *
36  * This code is derived from software contributed to The NetBSD Foundation
37  * by Jason R. Thorpe and Kevin M. Lahey of the Numerical Aerospace Simulation
38  * Facility, NASA Ames Research Center.
39  * This code is derived from software contributed to The NetBSD Foundation
40  * by Charles M. Hannum.
41  *
42  * Redistribution and use in source and binary forms, with or without
43  * modification, are permitted provided that the following conditions
44  * are met:
45  * 1. Redistributions of source code must retain the above copyright
46  *    notice, this list of conditions and the following disclaimer.
47  * 2. Redistributions in binary form must reproduce the above copyright
48  *    notice, this list of conditions and the following disclaimer in the
49  *    documentation and/or other materials provided with the distribution.
50  *
51  * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
52  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
53  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
54  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
55  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
56  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
57  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
58  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
59  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
60  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
61  * POSSIBILITY OF SUCH DAMAGE.
62  */
63 
64 /*
65  * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1995
66  *	The Regents of the University of California.  All rights reserved.
67  *
68  * Redistribution and use in source and binary forms, with or without
69  * modification, are permitted provided that the following conditions
70  * are met:
71  * 1. Redistributions of source code must retain the above copyright
72  *    notice, this list of conditions and the following disclaimer.
73  * 2. Redistributions in binary form must reproduce the above copyright
74  *    notice, this list of conditions and the following disclaimer in the
75  *    documentation and/or other materials provided with the distribution.
76  * 3. Neither the name of the University nor the names of its contributors
77  *    may be used to endorse or promote products derived from this software
78  *    without specific prior written permission.
79  *
80  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
81  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
82  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
83  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
84  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
85  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
86  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
87  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
88  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
89  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
90  * SUCH DAMAGE.
91  *
92  *	@(#)tcp_timer.c	8.2 (Berkeley) 5/24/95
93  */
94 
95 #include <sys/cdefs.h>
96 __KERNEL_RCSID(0, "$NetBSD: tcp_timer.c,v 1.84 2008/11/10 01:06:43 uebayasi Exp $");
97 
98 #include "opt_inet.h"
99 #include "opt_tcp_debug.h"
100 
101 #include <sys/param.h>
102 #include <sys/systm.h>
103 #include <sys/malloc.h>
104 #include <sys/mbuf.h>
105 #include <sys/socket.h>
106 #include <sys/socketvar.h>
107 #include <sys/protosw.h>
108 #include <sys/errno.h>
109 #include <sys/kernel.h>
110 
111 #include <net/if.h>
112 #include <net/route.h>
113 
114 #include <netinet/in.h>
115 #include <netinet/in_systm.h>
116 #include <netinet/ip.h>
117 #include <netinet/in_pcb.h>
118 #include <netinet/ip_var.h>
119 #include <netinet/ip_icmp.h>
120 
121 #ifdef INET6
122 #ifndef INET
123 #include <netinet/in.h>
124 #endif
125 #include <netinet/ip6.h>
126 #include <netinet6/in6_pcb.h>
127 #endif
128 
129 #include <netinet/tcp.h>
130 #include <netinet/tcp_fsm.h>
131 #include <netinet/tcp_seq.h>
132 #include <netinet/tcp_timer.h>
133 #include <netinet/tcp_var.h>
134 #include <netinet/tcp_private.h>
135 #include <netinet/tcp_congctl.h>
136 #include <netinet/tcpip.h>
137 #ifdef TCP_DEBUG
138 #include <netinet/tcp_debug.h>
139 #endif
140 
141 /*
142  * Various tunable timer parameters.  These are initialized in tcp_init(),
143  * unless they are patched.
144  */
145 u_int	tcp_keepinit = 0;
146 u_int	tcp_keepidle = 0;
147 u_int	tcp_keepintvl = 0;
148 u_int	tcp_keepcnt = 0;		/* max idle probes */
149 
150 int	tcp_maxpersistidle = 0;		/* max idle time in persist */
151 
152 /*
153  * Time to delay the ACK.  This is initialized in tcp_init(), unless
154  * its patched.
155  */
156 int	tcp_delack_ticks = 0;
157 
158 void	tcp_timer_rexmt(void *);
159 void	tcp_timer_persist(void *);
160 void	tcp_timer_keep(void *);
161 void	tcp_timer_2msl(void *);
162 
163 const tcp_timer_func_t tcp_timer_funcs[TCPT_NTIMERS] = {
164 	tcp_timer_rexmt,
165 	tcp_timer_persist,
166 	tcp_timer_keep,
167 	tcp_timer_2msl,
168 };
169 
170 /*
171  * Timer state initialization, called from tcp_init().
172  */
173 void
174 tcp_timer_init(void)
175 {
176 
177 	if (tcp_keepinit == 0)
178 		tcp_keepinit = TCPTV_KEEP_INIT;
179 
180 	if (tcp_keepidle == 0)
181 		tcp_keepidle = TCPTV_KEEP_IDLE;
182 
183 	if (tcp_keepintvl == 0)
184 		tcp_keepintvl = TCPTV_KEEPINTVL;
185 
186 	if (tcp_keepcnt == 0)
187 		tcp_keepcnt = TCPTV_KEEPCNT;
188 
189 	if (tcp_maxpersistidle == 0)
190 		tcp_maxpersistidle = TCPTV_KEEP_IDLE;
191 
192 	if (tcp_delack_ticks == 0)
193 		tcp_delack_ticks = TCP_DELACK_TICKS;
194 }
195 
196 /*
197  * Callout to process delayed ACKs for a TCPCB.
198  */
199 void
200 tcp_delack(void *arg)
201 {
202 	struct tcpcb *tp = arg;
203 
204 	/*
205 	 * If tcp_output() wasn't able to transmit the ACK
206 	 * for whatever reason, it will restart the delayed
207 	 * ACK callout.
208 	 */
209 
210 	mutex_enter(softnet_lock);
211 	if ((tp->t_flags & (TF_DEAD | TF_DELACK)) != TF_DELACK) {
212 		mutex_exit(softnet_lock);
213 		return;
214 	}
215 	if (!callout_expired(&tp->t_delack_ch)) {
216 		mutex_exit(softnet_lock);
217 		return;
218 	}
219 
220 	tp->t_flags |= TF_ACKNOW;
221 	KERNEL_LOCK(1, NULL);
222 	(void) tcp_output(tp);
223 	KERNEL_UNLOCK_ONE(NULL);
224 	mutex_exit(softnet_lock);
225 }
226 
227 /*
228  * Tcp protocol timeout routine called every 500 ms.
229  * Updates the timers in all active tcb's and
230  * causes finite state machine actions if timers expire.
231  */
232 void
233 tcp_slowtimo(void)
234 {
235 
236 	mutex_enter(softnet_lock);
237 	tcp_iss_seq += TCP_ISSINCR;			/* increment iss */
238 	tcp_now++;					/* for timestamps */
239 	mutex_exit(softnet_lock);
240 }
241 
242 /*
243  * Cancel all timers for TCP tp.
244  */
245 void
246 tcp_canceltimers(struct tcpcb *tp)
247 {
248 	int i;
249 
250 	for (i = 0; i < TCPT_NTIMERS; i++)
251 		TCP_TIMER_DISARM(tp, i);
252 }
253 
254 const int	tcp_backoff[TCP_MAXRXTSHIFT + 1] =
255     { 1, 2, 4, 8, 16, 32, 64, 64, 64, 64, 64, 64, 64 };
256 
257 const int	tcp_totbackoff = 511;	/* sum of tcp_backoff[] */
258 
259 /*
260  * TCP timer processing.
261  */
262 
263 void
264 tcp_timer_rexmt(void *arg)
265 {
266 	struct tcpcb *tp = arg;
267 	uint32_t rto;
268 #ifdef TCP_DEBUG
269 	struct socket *so = NULL;
270 	short ostate;
271 #endif
272 
273 	mutex_enter(softnet_lock);
274 	if ((tp->t_flags & TF_DEAD) != 0) {
275 		mutex_exit(softnet_lock);
276 		return;
277 	}
278 	if (!callout_expired(&tp->t_timer[TCPT_REXMT])) {
279 		mutex_exit(softnet_lock);
280 		return;
281 	}
282 
283 	KERNEL_LOCK(1, NULL);
284 	if ((tp->t_flags & TF_PMTUD_PEND) && tp->t_inpcb &&
285 	    SEQ_GEQ(tp->t_pmtud_th_seq, tp->snd_una) &&
286 	    SEQ_LT(tp->t_pmtud_th_seq, (int)(tp->snd_una + tp->t_ourmss))) {
287 		extern struct sockaddr_in icmpsrc;
288 		struct icmp icmp;
289 
290 		tp->t_flags &= ~TF_PMTUD_PEND;
291 
292 		/* XXX create fake icmp message with relevant entries */
293 		icmp.icmp_nextmtu = tp->t_pmtud_nextmtu;
294 		icmp.icmp_ip.ip_len = tp->t_pmtud_ip_len;
295 		icmp.icmp_ip.ip_hl = tp->t_pmtud_ip_hl;
296 		icmpsrc.sin_addr = tp->t_inpcb->inp_faddr;
297 		icmp_mtudisc(&icmp, icmpsrc.sin_addr);
298 
299 		/*
300 		 * Notify all connections to the same peer about
301 		 * new mss and trigger retransmit.
302 		 */
303 		in_pcbnotifyall(&tcbtable, icmpsrc.sin_addr, EMSGSIZE,
304 		    tcp_mtudisc);
305 		KERNEL_UNLOCK_ONE(NULL);
306 		mutex_exit(softnet_lock);
307  		return;
308  	}
309 #ifdef TCP_DEBUG
310 #ifdef INET
311 	if (tp->t_inpcb)
312 		so = tp->t_inpcb->inp_socket;
313 #endif
314 #ifdef INET6
315 	if (tp->t_in6pcb)
316 		so = tp->t_in6pcb->in6p_socket;
317 #endif
318 	ostate = tp->t_state;
319 #endif /* TCP_DEBUG */
320 
321 	/*
322 	 * Clear the SACK scoreboard, reset FACK estimate.
323 	 */
324 	tcp_free_sackholes(tp);
325 	tp->snd_fack = tp->snd_una;
326 
327 	/*
328 	 * Retransmission timer went off.  Message has not
329 	 * been acked within retransmit interval.  Back off
330 	 * to a longer retransmit interval and retransmit one segment.
331 	 */
332 
333 	if (++tp->t_rxtshift > TCP_MAXRXTSHIFT) {
334 		tp->t_rxtshift = TCP_MAXRXTSHIFT;
335 		TCP_STATINC(TCP_STAT_TIMEOUTDROP);
336 		tp = tcp_drop(tp, tp->t_softerror ?
337 		    tp->t_softerror : ETIMEDOUT);
338 		goto out;
339 	}
340 	TCP_STATINC(TCP_STAT_REXMTTIMEO);
341 	rto = TCP_REXMTVAL(tp);
342 	if (rto < tp->t_rttmin)
343 		rto = tp->t_rttmin;
344 	TCPT_RANGESET(tp->t_rxtcur, rto * tcp_backoff[tp->t_rxtshift],
345 	    tp->t_rttmin, TCPTV_REXMTMAX);
346 	TCP_TIMER_ARM(tp, TCPT_REXMT, tp->t_rxtcur);
347 
348 	/*
349 	 * If we are losing and we are trying path MTU discovery,
350 	 * try turning it off.  This will avoid black holes in
351 	 * the network which suppress or fail to send "packet
352 	 * too big" ICMP messages.  We should ideally do
353 	 * lots more sophisticated searching to find the right
354 	 * value here...
355 	 */
356 	if (tp->t_mtudisc && tp->t_rxtshift > TCP_MAXRXTSHIFT / 6) {
357 		TCP_STATINC(TCP_STAT_PMTUBLACKHOLE);
358 
359 #ifdef INET
360 		/* try turning PMTUD off */
361 		if (tp->t_inpcb)
362 			tp->t_mtudisc = 0;
363 #endif
364 #ifdef INET6
365 		/* try using IPv6 minimum MTU */
366 		if (tp->t_in6pcb)
367 			tp->t_mtudisc = 0;
368 #endif
369 
370 		/* XXX: more sophisticated Black hole recovery code? */
371 	}
372 
373 	/*
374 	 * If losing, let the lower level know and try for
375 	 * a better route.  Also, if we backed off this far,
376 	 * our srtt estimate is probably bogus.  Clobber it
377 	 * so we'll take the next rtt measurement as our srtt;
378 	 * move the current srtt into rttvar to keep the current
379 	 * retransmit times until then.
380 	 */
381 	if (tp->t_rxtshift > TCP_MAXRXTSHIFT / 4) {
382 #ifdef INET
383 		if (tp->t_inpcb)
384 			in_losing(tp->t_inpcb);
385 #endif
386 #ifdef INET6
387 		if (tp->t_in6pcb)
388 			in6_losing(tp->t_in6pcb);
389 #endif
390 		tp->t_rttvar += (tp->t_srtt >> TCP_RTT_SHIFT);
391 		tp->t_srtt = 0;
392 	}
393 	tp->snd_nxt = tp->snd_una;
394 	tp->snd_high = tp->snd_max;
395 	/*
396 	 * If timing a segment in this window, stop the timer.
397 	 */
398 	tp->t_rtttime = 0;
399 	/*
400 	 * Remember if we are retransmitting a SYN, because if
401 	 * we do, set the initial congestion window must be set
402 	 * to 1 segment.
403 	 */
404 	if (tp->t_state == TCPS_SYN_SENT)
405 		tp->t_flags |= TF_SYN_REXMT;
406 
407 	/*
408 	 * Adjust congestion control parameters.
409 	 */
410 	tp->t_congctl->slow_retransmit(tp);
411 
412 	(void) tcp_output(tp);
413 
414  out:
415 #ifdef TCP_DEBUG
416 	if (tp && so->so_options & SO_DEBUG)
417 		tcp_trace(TA_USER, ostate, tp, NULL,
418 		    PRU_SLOWTIMO | (TCPT_REXMT << 8));
419 #endif
420 	KERNEL_UNLOCK_ONE(NULL);
421 	mutex_exit(softnet_lock);
422 }
423 
424 void
425 tcp_timer_persist(void *arg)
426 {
427 	struct tcpcb *tp = arg;
428 	uint32_t rto;
429 #ifdef TCP_DEBUG
430 	struct socket *so = NULL;
431 	short ostate;
432 #endif
433 
434 	mutex_enter(softnet_lock);
435 	if ((tp->t_flags & TF_DEAD) != 0) {
436 		mutex_exit(softnet_lock);
437 		return;
438 	}
439 	if (!callout_expired(&tp->t_timer[TCPT_PERSIST])) {
440 		mutex_exit(softnet_lock);
441 		return;
442 	}
443 
444 	KERNEL_LOCK(1, NULL);
445 #ifdef TCP_DEBUG
446 #ifdef INET
447 	if (tp->t_inpcb)
448 		so = tp->t_inpcb->inp_socket;
449 #endif
450 #ifdef INET6
451 	if (tp->t_in6pcb)
452 		so = tp->t_in6pcb->in6p_socket;
453 #endif
454 
455 	ostate = tp->t_state;
456 #endif /* TCP_DEBUG */
457 
458 	/*
459 	 * Persistance timer into zero window.
460 	 * Force a byte to be output, if possible.
461 	 */
462 
463 	/*
464 	 * Hack: if the peer is dead/unreachable, we do not
465 	 * time out if the window is closed.  After a full
466 	 * backoff, drop the connection if the idle time
467 	 * (no responses to probes) reaches the maximum
468 	 * backoff that we would use if retransmitting.
469 	 */
470 	rto = TCP_REXMTVAL(tp);
471 	if (rto < tp->t_rttmin)
472 		rto = tp->t_rttmin;
473 	if (tp->t_rxtshift == TCP_MAXRXTSHIFT &&
474 	    ((tcp_now - tp->t_rcvtime) >= tcp_maxpersistidle ||
475 	    (tcp_now - tp->t_rcvtime) >= rto * tcp_totbackoff)) {
476 		TCP_STATINC(TCP_STAT_PERSISTDROPS);
477 		tp = tcp_drop(tp, ETIMEDOUT);
478 		goto out;
479 	}
480 	TCP_STATINC(TCP_STAT_PERSISTTIMEO);
481 	tcp_setpersist(tp);
482 	tp->t_force = 1;
483 	(void) tcp_output(tp);
484 	tp->t_force = 0;
485 
486  out:
487 #ifdef TCP_DEBUG
488 	if (tp && so->so_options & SO_DEBUG)
489 		tcp_trace(TA_USER, ostate, tp, NULL,
490 		    PRU_SLOWTIMO | (TCPT_PERSIST << 8));
491 #endif
492 	KERNEL_UNLOCK_ONE(NULL);
493 	mutex_exit(softnet_lock);
494 }
495 
496 void
497 tcp_timer_keep(void *arg)
498 {
499 	struct tcpcb *tp = arg;
500 	struct socket *so = NULL;	/* Quell compiler warning */
501 #ifdef TCP_DEBUG
502 	short ostate;
503 #endif
504 
505 	mutex_enter(softnet_lock);
506 	if ((tp->t_flags & TF_DEAD) != 0) {
507 		mutex_exit(softnet_lock);
508 		return;
509 	}
510 	if (!callout_expired(&tp->t_timer[TCPT_KEEP])) {
511 		mutex_exit(softnet_lock);
512 		return;
513 	}
514 
515 	KERNEL_LOCK(1, NULL);
516 
517 #ifdef TCP_DEBUG
518 	ostate = tp->t_state;
519 #endif /* TCP_DEBUG */
520 
521 	/*
522 	 * Keep-alive timer went off; send something
523 	 * or drop connection if idle for too long.
524 	 */
525 
526 	TCP_STATINC(TCP_STAT_KEEPTIMEO);
527 	if (TCPS_HAVEESTABLISHED(tp->t_state) == 0)
528 		goto dropit;
529 #ifdef INET
530 	if (tp->t_inpcb)
531 		so = tp->t_inpcb->inp_socket;
532 #endif
533 #ifdef INET6
534 	if (tp->t_in6pcb)
535 		so = tp->t_in6pcb->in6p_socket;
536 #endif
537 	KASSERT(so != NULL);
538 	if (so->so_options & SO_KEEPALIVE &&
539 	    tp->t_state <= TCPS_CLOSE_WAIT) {
540 	    	if ((tp->t_maxidle > 0) &&
541 		    ((tcp_now - tp->t_rcvtime) >=
542 		     tp->t_keepidle + tp->t_maxidle))
543 			goto dropit;
544 		/*
545 		 * Send a packet designed to force a response
546 		 * if the peer is up and reachable:
547 		 * either an ACK if the connection is still alive,
548 		 * or an RST if the peer has closed the connection
549 		 * due to timeout or reboot.
550 		 * Using sequence number tp->snd_una-1
551 		 * causes the transmitted zero-length segment
552 		 * to lie outside the receive window;
553 		 * by the protocol spec, this requires the
554 		 * correspondent TCP to respond.
555 		 */
556 		TCP_STATINC(TCP_STAT_KEEPPROBE);
557 		if (tcp_compat_42) {
558 			/*
559 			 * The keepalive packet must have nonzero
560 			 * length to get a 4.2 host to respond.
561 			 */
562 			(void)tcp_respond(tp, tp->t_template,
563 			    (struct mbuf *)NULL, NULL, tp->rcv_nxt - 1,
564 			    tp->snd_una - 1, 0);
565 		} else {
566 			(void)tcp_respond(tp, tp->t_template,
567 			    (struct mbuf *)NULL, NULL, tp->rcv_nxt,
568 			    tp->snd_una - 1, 0);
569 		}
570 		TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepintvl);
571 	} else
572 		TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepidle);
573 
574 #ifdef TCP_DEBUG
575 	if (tp && so->so_options & SO_DEBUG)
576 		tcp_trace(TA_USER, ostate, tp, NULL,
577 		    PRU_SLOWTIMO | (TCPT_KEEP << 8));
578 #endif
579 	KERNEL_UNLOCK_ONE(NULL);
580 	mutex_exit(softnet_lock);
581 	return;
582 
583  dropit:
584 	TCP_STATINC(TCP_STAT_KEEPDROPS);
585 	(void) tcp_drop(tp, ETIMEDOUT);
586 	KERNEL_UNLOCK_ONE(NULL);
587 	mutex_exit(softnet_lock);
588 }
589 
590 void
591 tcp_timer_2msl(void *arg)
592 {
593 	struct tcpcb *tp = arg;
594 #ifdef TCP_DEBUG
595 	struct socket *so = NULL;
596 	short ostate;
597 #endif
598 
599 	mutex_enter(softnet_lock);
600 	if ((tp->t_flags & TF_DEAD) != 0) {
601 		mutex_exit(softnet_lock);
602 		return;
603 	}
604 	if (!callout_expired(&tp->t_timer[TCPT_2MSL])) {
605 		mutex_exit(softnet_lock);
606 		return;
607 	}
608 
609 	/*
610 	 * 2 MSL timeout went off, clear the SACK scoreboard, reset
611 	 * the FACK estimate.
612 	 */
613 	KERNEL_LOCK(1, NULL);
614 	tcp_free_sackholes(tp);
615 	tp->snd_fack = tp->snd_una;
616 
617 #ifdef TCP_DEBUG
618 #ifdef INET
619 	if (tp->t_inpcb)
620 		so = tp->t_inpcb->inp_socket;
621 #endif
622 #ifdef INET6
623 	if (tp->t_in6pcb)
624 		so = tp->t_in6pcb->in6p_socket;
625 #endif
626 
627 	ostate = tp->t_state;
628 #endif /* TCP_DEBUG */
629 
630 	/*
631 	 * 2 MSL timeout in shutdown went off.  If we're closed but
632 	 * still waiting for peer to close and connection has been idle
633 	 * too long, or if 2MSL time is up from TIME_WAIT, delete connection
634 	 * control block.  Otherwise, check again in a bit.
635 	 */
636 	if (tp->t_state != TCPS_TIME_WAIT &&
637 	    ((tp->t_maxidle == 0) ||
638 	    ((tcp_now - tp->t_rcvtime) <= tp->t_maxidle)))
639 	    TCP_TIMER_ARM(tp, TCPT_2MSL, tp->t_keepintvl);
640 	else
641 		tp = tcp_close(tp);
642 
643 #ifdef TCP_DEBUG
644 	if (tp && so->so_options & SO_DEBUG)
645 		tcp_trace(TA_USER, ostate, tp, NULL,
646 		    PRU_SLOWTIMO | (TCPT_2MSL << 8));
647 #endif
648 	mutex_exit(softnet_lock);
649 	KERNEL_UNLOCK_ONE(NULL);
650 }
651