xref: /netbsd-src/sys/netinet/tcp_input.c (revision 975a152cfcdb39ae6e496af647af0c7275ca0b61)
1 /*	$NetBSD: tcp_input.c,v 1.328 2013/08/29 17:49:20 rmind 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   1.1 (NRL) 17 January 1995
34  *
35  * NRL grants permission for redistribution and use in source and binary
36  * forms, with or without modification, of the software and documentation
37  * created at NRL provided that the following conditions are met:
38  *
39  * 1. Redistributions of source code must retain the above copyright
40  *    notice, this list of conditions and the following disclaimer.
41  * 2. Redistributions in binary form must reproduce the above copyright
42  *    notice, this list of conditions and the following disclaimer in the
43  *    documentation and/or other materials provided with the distribution.
44  * 3. All advertising materials mentioning features or use of this software
45  *    must display the following acknowledgements:
46  *      This product includes software developed by the University of
47  *      California, Berkeley and its contributors.
48  *      This product includes software developed at the Information
49  *      Technology Division, US Naval Research Laboratory.
50  * 4. Neither the name of the NRL nor the names of its contributors
51  *    may be used to endorse or promote products derived from this software
52  *    without specific prior written permission.
53  *
54  * THE SOFTWARE PROVIDED BY NRL IS PROVIDED BY NRL AND CONTRIBUTORS ``AS
55  * IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
56  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A
57  * PARTICULAR PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL NRL OR
58  * CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
59  * EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
60  * PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR
61  * PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF
62  * LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING
63  * NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS
64  * SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
65  *
66  * The views and conclusions contained in the software and documentation
67  * are those of the authors and should not be interpreted as representing
68  * official policies, either expressed or implied, of the US Naval
69  * Research Laboratory (NRL).
70  */
71 
72 /*-
73  * Copyright (c) 1997, 1998, 1999, 2001, 2005, 2006,
74  * 2011 The NetBSD Foundation, Inc.
75  * All rights reserved.
76  *
77  * This code is derived from software contributed to The NetBSD Foundation
78  * by Coyote Point Systems, Inc.
79  * This code is derived from software contributed to The NetBSD Foundation
80  * by Jason R. Thorpe and Kevin M. Lahey of the Numerical Aerospace Simulation
81  * Facility, NASA Ames Research Center.
82  * This code is derived from software contributed to The NetBSD Foundation
83  * by Charles M. Hannum.
84  * This code is derived from software contributed to The NetBSD Foundation
85  * by Rui Paulo.
86  *
87  * Redistribution and use in source and binary forms, with or without
88  * modification, are permitted provided that the following conditions
89  * are met:
90  * 1. Redistributions of source code must retain the above copyright
91  *    notice, this list of conditions and the following disclaimer.
92  * 2. Redistributions in binary form must reproduce the above copyright
93  *    notice, this list of conditions and the following disclaimer in the
94  *    documentation and/or other materials provided with the distribution.
95  *
96  * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
97  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
98  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
99  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
100  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
101  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
102  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
103  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
104  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
105  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
106  * POSSIBILITY OF SUCH DAMAGE.
107  */
108 
109 /*
110  * Copyright (c) 1982, 1986, 1988, 1990, 1993, 1994, 1995
111  *	The Regents of the University of California.  All rights reserved.
112  *
113  * Redistribution and use in source and binary forms, with or without
114  * modification, are permitted provided that the following conditions
115  * are met:
116  * 1. Redistributions of source code must retain the above copyright
117  *    notice, this list of conditions and the following disclaimer.
118  * 2. Redistributions in binary form must reproduce the above copyright
119  *    notice, this list of conditions and the following disclaimer in the
120  *    documentation and/or other materials provided with the distribution.
121  * 3. Neither the name of the University nor the names of its contributors
122  *    may be used to endorse or promote products derived from this software
123  *    without specific prior written permission.
124  *
125  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
126  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
127  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
128  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
129  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
130  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
131  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
132  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
133  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
134  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
135  * SUCH DAMAGE.
136  *
137  *	@(#)tcp_input.c	8.12 (Berkeley) 5/24/95
138  */
139 
140 /*
141  *	TODO list for SYN cache stuff:
142  *
143  *	Find room for a "state" field, which is needed to keep a
144  *	compressed state for TIME_WAIT TCBs.  It's been noted already
145  *	that this is fairly important for very high-volume web and
146  *	mail servers, which use a large number of short-lived
147  *	connections.
148  */
149 
150 #include <sys/cdefs.h>
151 __KERNEL_RCSID(0, "$NetBSD: tcp_input.c,v 1.328 2013/08/29 17:49:20 rmind Exp $");
152 
153 #include "opt_inet.h"
154 #include "opt_ipsec.h"
155 #include "opt_inet_csum.h"
156 #include "opt_tcp_debug.h"
157 
158 #include <sys/param.h>
159 #include <sys/systm.h>
160 #include <sys/malloc.h>
161 #include <sys/mbuf.h>
162 #include <sys/protosw.h>
163 #include <sys/socket.h>
164 #include <sys/socketvar.h>
165 #include <sys/errno.h>
166 #include <sys/syslog.h>
167 #include <sys/pool.h>
168 #include <sys/domain.h>
169 #include <sys/kernel.h>
170 #ifdef TCP_SIGNATURE
171 #include <sys/md5.h>
172 #endif
173 #include <sys/lwp.h> /* for lwp0 */
174 #include <sys/cprng.h>
175 
176 #include <net/if.h>
177 #include <net/route.h>
178 #include <net/if_types.h>
179 
180 #include <netinet/in.h>
181 #include <netinet/in_systm.h>
182 #include <netinet/ip.h>
183 #include <netinet/in_pcb.h>
184 #include <netinet/in_var.h>
185 #include <netinet/ip_var.h>
186 #include <netinet/in_offload.h>
187 
188 #ifdef INET6
189 #ifndef INET
190 #include <netinet/in.h>
191 #endif
192 #include <netinet/ip6.h>
193 #include <netinet6/ip6_var.h>
194 #include <netinet6/in6_pcb.h>
195 #include <netinet6/ip6_var.h>
196 #include <netinet6/in6_var.h>
197 #include <netinet/icmp6.h>
198 #include <netinet6/nd6.h>
199 #ifdef TCP_SIGNATURE
200 #include <netinet6/scope6_var.h>
201 #endif
202 #endif
203 
204 #ifndef INET6
205 /* always need ip6.h for IP6_EXTHDR_GET */
206 #include <netinet/ip6.h>
207 #endif
208 
209 #include <netinet/tcp.h>
210 #include <netinet/tcp_fsm.h>
211 #include <netinet/tcp_seq.h>
212 #include <netinet/tcp_timer.h>
213 #include <netinet/tcp_var.h>
214 #include <netinet/tcp_private.h>
215 #include <netinet/tcpip.h>
216 #include <netinet/tcp_congctl.h>
217 #include <netinet/tcp_debug.h>
218 
219 #ifdef INET6
220 #include "faith.h"
221 #if defined(NFAITH) && NFAITH > 0
222 #include <net/if_faith.h>
223 #endif
224 #endif	/* INET6 */
225 
226 #ifdef IPSEC
227 #include <netipsec/ipsec.h>
228 #include <netipsec/ipsec_var.h>
229 #include <netipsec/ipsec_private.h>
230 #include <netipsec/key.h>
231 #ifdef INET6
232 #include <netipsec/ipsec6.h>
233 #endif
234 #endif	/* IPSEC*/
235 
236 #include <netinet/tcp_vtw.h>
237 
238 int	tcprexmtthresh = 3;
239 int	tcp_log_refused;
240 
241 int	tcp_do_autorcvbuf = 1;
242 int	tcp_autorcvbuf_inc = 16 * 1024;
243 int	tcp_autorcvbuf_max = 256 * 1024;
244 int	tcp_msl = (TCPTV_MSL / PR_SLOWHZ);
245 
246 static int tcp_rst_ppslim_count = 0;
247 static struct timeval tcp_rst_ppslim_last;
248 static int tcp_ackdrop_ppslim_count = 0;
249 static struct timeval tcp_ackdrop_ppslim_last;
250 
251 #define TCP_PAWS_IDLE	(24U * 24 * 60 * 60 * PR_SLOWHZ)
252 
253 /* for modulo comparisons of timestamps */
254 #define TSTMP_LT(a,b)	((int)((a)-(b)) < 0)
255 #define TSTMP_GEQ(a,b)	((int)((a)-(b)) >= 0)
256 
257 /*
258  * Neighbor Discovery, Neighbor Unreachability Detection Upper layer hint.
259  */
260 #ifdef INET6
261 static inline void
262 nd6_hint(struct tcpcb *tp)
263 {
264 	struct rtentry *rt;
265 
266 	if (tp != NULL && tp->t_in6pcb != NULL && tp->t_family == AF_INET6 &&
267 	    (rt = rtcache_validate(&tp->t_in6pcb->in6p_route)) != NULL)
268 		nd6_nud_hint(rt, NULL, 0);
269 }
270 #else
271 static inline void
272 nd6_hint(struct tcpcb *tp)
273 {
274 }
275 #endif
276 
277 /*
278  * Compute ACK transmission behavior.  Delay the ACK unless
279  * we have already delayed an ACK (must send an ACK every two segments).
280  * We also ACK immediately if we received a PUSH and the ACK-on-PUSH
281  * option is enabled.
282  */
283 static void
284 tcp_setup_ack(struct tcpcb *tp, const struct tcphdr *th)
285 {
286 
287 	if (tp->t_flags & TF_DELACK ||
288 	    (tcp_ack_on_push && th->th_flags & TH_PUSH))
289 		tp->t_flags |= TF_ACKNOW;
290 	else
291 		TCP_SET_DELACK(tp);
292 }
293 
294 static void
295 icmp_check(struct tcpcb *tp, const struct tcphdr *th, int acked)
296 {
297 
298 	/*
299 	 * If we had a pending ICMP message that refers to data that have
300 	 * just been acknowledged, disregard the recorded ICMP message.
301 	 */
302 	if ((tp->t_flags & TF_PMTUD_PEND) &&
303 	    SEQ_GT(th->th_ack, tp->t_pmtud_th_seq))
304 		tp->t_flags &= ~TF_PMTUD_PEND;
305 
306 	/*
307 	 * Keep track of the largest chunk of data
308 	 * acknowledged since last PMTU update
309 	 */
310 	if (tp->t_pmtud_mss_acked < acked)
311 		tp->t_pmtud_mss_acked = acked;
312 }
313 
314 /*
315  * Convert TCP protocol fields to host order for easier processing.
316  */
317 static void
318 tcp_fields_to_host(struct tcphdr *th)
319 {
320 
321 	NTOHL(th->th_seq);
322 	NTOHL(th->th_ack);
323 	NTOHS(th->th_win);
324 	NTOHS(th->th_urp);
325 }
326 
327 /*
328  * ... and reverse the above.
329  */
330 static void
331 tcp_fields_to_net(struct tcphdr *th)
332 {
333 
334 	HTONL(th->th_seq);
335 	HTONL(th->th_ack);
336 	HTONS(th->th_win);
337 	HTONS(th->th_urp);
338 }
339 
340 #ifdef TCP_CSUM_COUNTERS
341 #include <sys/device.h>
342 
343 #if defined(INET)
344 extern struct evcnt tcp_hwcsum_ok;
345 extern struct evcnt tcp_hwcsum_bad;
346 extern struct evcnt tcp_hwcsum_data;
347 extern struct evcnt tcp_swcsum;
348 #endif /* defined(INET) */
349 #if defined(INET6)
350 extern struct evcnt tcp6_hwcsum_ok;
351 extern struct evcnt tcp6_hwcsum_bad;
352 extern struct evcnt tcp6_hwcsum_data;
353 extern struct evcnt tcp6_swcsum;
354 #endif /* defined(INET6) */
355 
356 #define	TCP_CSUM_COUNTER_INCR(ev)	(ev)->ev_count++
357 
358 #else
359 
360 #define	TCP_CSUM_COUNTER_INCR(ev)	/* nothing */
361 
362 #endif /* TCP_CSUM_COUNTERS */
363 
364 #ifdef TCP_REASS_COUNTERS
365 #include <sys/device.h>
366 
367 extern struct evcnt tcp_reass_;
368 extern struct evcnt tcp_reass_empty;
369 extern struct evcnt tcp_reass_iteration[8];
370 extern struct evcnt tcp_reass_prependfirst;
371 extern struct evcnt tcp_reass_prepend;
372 extern struct evcnt tcp_reass_insert;
373 extern struct evcnt tcp_reass_inserttail;
374 extern struct evcnt tcp_reass_append;
375 extern struct evcnt tcp_reass_appendtail;
376 extern struct evcnt tcp_reass_overlaptail;
377 extern struct evcnt tcp_reass_overlapfront;
378 extern struct evcnt tcp_reass_segdup;
379 extern struct evcnt tcp_reass_fragdup;
380 
381 #define	TCP_REASS_COUNTER_INCR(ev)	(ev)->ev_count++
382 
383 #else
384 
385 #define	TCP_REASS_COUNTER_INCR(ev)	/* nothing */
386 
387 #endif /* TCP_REASS_COUNTERS */
388 
389 static int tcp_reass(struct tcpcb *, const struct tcphdr *, struct mbuf *,
390     int *);
391 static int tcp_dooptions(struct tcpcb *, const u_char *, int,
392     struct tcphdr *, struct mbuf *, int, struct tcp_opt_info *);
393 
394 #ifdef INET
395 static void tcp4_log_refused(const struct ip *, const struct tcphdr *);
396 #endif
397 #ifdef INET6
398 static void tcp6_log_refused(const struct ip6_hdr *, const struct tcphdr *);
399 #endif
400 
401 #define	TRAVERSE(x) while ((x)->m_next) (x) = (x)->m_next
402 
403 #if defined(MBUFTRACE)
404 struct mowner tcp_reass_mowner = MOWNER_INIT("tcp", "reass");
405 #endif /* defined(MBUFTRACE) */
406 
407 static struct pool tcpipqent_pool;
408 
409 void
410 tcpipqent_init(void)
411 {
412 
413 	pool_init(&tcpipqent_pool, sizeof(struct ipqent), 0, 0, 0, "tcpipqepl",
414 	    NULL, IPL_VM);
415 }
416 
417 struct ipqent *
418 tcpipqent_alloc(void)
419 {
420 	struct ipqent *ipqe;
421 	int s;
422 
423 	s = splvm();
424 	ipqe = pool_get(&tcpipqent_pool, PR_NOWAIT);
425 	splx(s);
426 
427 	return ipqe;
428 }
429 
430 void
431 tcpipqent_free(struct ipqent *ipqe)
432 {
433 	int s;
434 
435 	s = splvm();
436 	pool_put(&tcpipqent_pool, ipqe);
437 	splx(s);
438 }
439 
440 static int
441 tcp_reass(struct tcpcb *tp, const struct tcphdr *th, struct mbuf *m, int *tlen)
442 {
443 	struct ipqent *p, *q, *nq, *tiqe = NULL;
444 	struct socket *so = NULL;
445 	int pkt_flags;
446 	tcp_seq pkt_seq;
447 	unsigned pkt_len;
448 	u_long rcvpartdupbyte = 0;
449 	u_long rcvoobyte;
450 #ifdef TCP_REASS_COUNTERS
451 	u_int count = 0;
452 #endif
453 	uint64_t *tcps;
454 
455 	if (tp->t_inpcb)
456 		so = tp->t_inpcb->inp_socket;
457 #ifdef INET6
458 	else if (tp->t_in6pcb)
459 		so = tp->t_in6pcb->in6p_socket;
460 #endif
461 
462 	TCP_REASS_LOCK_CHECK(tp);
463 
464 	/*
465 	 * Call with th==0 after become established to
466 	 * force pre-ESTABLISHED data up to user socket.
467 	 */
468 	if (th == 0)
469 		goto present;
470 
471 	m_claimm(m, &tcp_reass_mowner);
472 
473 	rcvoobyte = *tlen;
474 	/*
475 	 * Copy these to local variables because the tcpiphdr
476 	 * gets munged while we are collapsing mbufs.
477 	 */
478 	pkt_seq = th->th_seq;
479 	pkt_len = *tlen;
480 	pkt_flags = th->th_flags;
481 
482 	TCP_REASS_COUNTER_INCR(&tcp_reass_);
483 
484 	if ((p = TAILQ_LAST(&tp->segq, ipqehead)) != NULL) {
485 		/*
486 		 * When we miss a packet, the vast majority of time we get
487 		 * packets that follow it in order.  So optimize for that.
488 		 */
489 		if (pkt_seq == p->ipqe_seq + p->ipqe_len) {
490 			p->ipqe_len += pkt_len;
491 			p->ipqe_flags |= pkt_flags;
492 			m_cat(p->ipre_mlast, m);
493 			TRAVERSE(p->ipre_mlast);
494 			m = NULL;
495 			tiqe = p;
496 			TAILQ_REMOVE(&tp->timeq, p, ipqe_timeq);
497 			TCP_REASS_COUNTER_INCR(&tcp_reass_appendtail);
498 			goto skip_replacement;
499 		}
500 		/*
501 		 * While we're here, if the pkt is completely beyond
502 		 * anything we have, just insert it at the tail.
503 		 */
504 		if (SEQ_GT(pkt_seq, p->ipqe_seq + p->ipqe_len)) {
505 			TCP_REASS_COUNTER_INCR(&tcp_reass_inserttail);
506 			goto insert_it;
507 		}
508 	}
509 
510 	q = TAILQ_FIRST(&tp->segq);
511 
512 	if (q != NULL) {
513 		/*
514 		 * If this segment immediately precedes the first out-of-order
515 		 * block, simply slap the segment in front of it and (mostly)
516 		 * skip the complicated logic.
517 		 */
518 		if (pkt_seq + pkt_len == q->ipqe_seq) {
519 			q->ipqe_seq = pkt_seq;
520 			q->ipqe_len += pkt_len;
521 			q->ipqe_flags |= pkt_flags;
522 			m_cat(m, q->ipqe_m);
523 			q->ipqe_m = m;
524 			q->ipre_mlast = m; /* last mbuf may have changed */
525 			TRAVERSE(q->ipre_mlast);
526 			tiqe = q;
527 			TAILQ_REMOVE(&tp->timeq, q, ipqe_timeq);
528 			TCP_REASS_COUNTER_INCR(&tcp_reass_prependfirst);
529 			goto skip_replacement;
530 		}
531 	} else {
532 		TCP_REASS_COUNTER_INCR(&tcp_reass_empty);
533 	}
534 
535 	/*
536 	 * Find a segment which begins after this one does.
537 	 */
538 	for (p = NULL; q != NULL; q = nq) {
539 		nq = TAILQ_NEXT(q, ipqe_q);
540 #ifdef TCP_REASS_COUNTERS
541 		count++;
542 #endif
543 		/*
544 		 * If the received segment is just right after this
545 		 * fragment, merge the two together and then check
546 		 * for further overlaps.
547 		 */
548 		if (q->ipqe_seq + q->ipqe_len == pkt_seq) {
549 #ifdef TCPREASS_DEBUG
550 			printf("tcp_reass[%p]: concat %u:%u(%u) to %u:%u(%u)\n",
551 			       tp, pkt_seq, pkt_seq + pkt_len, pkt_len,
552 			       q->ipqe_seq, q->ipqe_seq + q->ipqe_len, q->ipqe_len);
553 #endif
554 			pkt_len += q->ipqe_len;
555 			pkt_flags |= q->ipqe_flags;
556 			pkt_seq = q->ipqe_seq;
557 			m_cat(q->ipre_mlast, m);
558 			TRAVERSE(q->ipre_mlast);
559 			m = q->ipqe_m;
560 			TCP_REASS_COUNTER_INCR(&tcp_reass_append);
561 			goto free_ipqe;
562 		}
563 		/*
564 		 * If the received segment is completely past this
565 		 * fragment, we need to go the next fragment.
566 		 */
567 		if (SEQ_LT(q->ipqe_seq + q->ipqe_len, pkt_seq)) {
568 			p = q;
569 			continue;
570 		}
571 		/*
572 		 * If the fragment is past the received segment,
573 		 * it (or any following) can't be concatenated.
574 		 */
575 		if (SEQ_GT(q->ipqe_seq, pkt_seq + pkt_len)) {
576 			TCP_REASS_COUNTER_INCR(&tcp_reass_insert);
577 			break;
578 		}
579 
580 		/*
581 		 * We've received all the data in this segment before.
582 		 * mark it as a duplicate and return.
583 		 */
584 		if (SEQ_LEQ(q->ipqe_seq, pkt_seq) &&
585 		    SEQ_GEQ(q->ipqe_seq + q->ipqe_len, pkt_seq + pkt_len)) {
586 			tcps = TCP_STAT_GETREF();
587 			tcps[TCP_STAT_RCVDUPPACK]++;
588 			tcps[TCP_STAT_RCVDUPBYTE] += pkt_len;
589 			TCP_STAT_PUTREF();
590 			tcp_new_dsack(tp, pkt_seq, pkt_len);
591 			m_freem(m);
592 			if (tiqe != NULL) {
593 				tcpipqent_free(tiqe);
594 			}
595 			TCP_REASS_COUNTER_INCR(&tcp_reass_segdup);
596 			goto out;
597 		}
598 		/*
599 		 * Received segment completely overlaps this fragment
600 		 * so we drop the fragment (this keeps the temporal
601 		 * ordering of segments correct).
602 		 */
603 		if (SEQ_GEQ(q->ipqe_seq, pkt_seq) &&
604 		    SEQ_LEQ(q->ipqe_seq + q->ipqe_len, pkt_seq + pkt_len)) {
605 			rcvpartdupbyte += q->ipqe_len;
606 			m_freem(q->ipqe_m);
607 			TCP_REASS_COUNTER_INCR(&tcp_reass_fragdup);
608 			goto free_ipqe;
609 		}
610 		/*
611 		 * RX'ed segment extends past the end of the
612 		 * fragment.  Drop the overlapping bytes.  Then
613 		 * merge the fragment and segment then treat as
614 		 * a longer received packet.
615 		 */
616 		if (SEQ_LT(q->ipqe_seq, pkt_seq) &&
617 		    SEQ_GT(q->ipqe_seq + q->ipqe_len, pkt_seq))  {
618 			int overlap = q->ipqe_seq + q->ipqe_len - pkt_seq;
619 #ifdef TCPREASS_DEBUG
620 			printf("tcp_reass[%p]: trim starting %d bytes of %u:%u(%u)\n",
621 			       tp, overlap,
622 			       pkt_seq, pkt_seq + pkt_len, pkt_len);
623 #endif
624 			m_adj(m, overlap);
625 			rcvpartdupbyte += overlap;
626 			m_cat(q->ipre_mlast, m);
627 			TRAVERSE(q->ipre_mlast);
628 			m = q->ipqe_m;
629 			pkt_seq = q->ipqe_seq;
630 			pkt_len += q->ipqe_len - overlap;
631 			rcvoobyte -= overlap;
632 			TCP_REASS_COUNTER_INCR(&tcp_reass_overlaptail);
633 			goto free_ipqe;
634 		}
635 		/*
636 		 * RX'ed segment extends past the front of the
637 		 * fragment.  Drop the overlapping bytes on the
638 		 * received packet.  The packet will then be
639 		 * contatentated with this fragment a bit later.
640 		 */
641 		if (SEQ_GT(q->ipqe_seq, pkt_seq) &&
642 		    SEQ_LT(q->ipqe_seq, pkt_seq + pkt_len))  {
643 			int overlap = pkt_seq + pkt_len - q->ipqe_seq;
644 #ifdef TCPREASS_DEBUG
645 			printf("tcp_reass[%p]: trim trailing %d bytes of %u:%u(%u)\n",
646 			       tp, overlap,
647 			       pkt_seq, pkt_seq + pkt_len, pkt_len);
648 #endif
649 			m_adj(m, -overlap);
650 			pkt_len -= overlap;
651 			rcvpartdupbyte += overlap;
652 			TCP_REASS_COUNTER_INCR(&tcp_reass_overlapfront);
653 			rcvoobyte -= overlap;
654 		}
655 		/*
656 		 * If the received segment immediates precedes this
657 		 * fragment then tack the fragment onto this segment
658 		 * and reinsert the data.
659 		 */
660 		if (q->ipqe_seq == pkt_seq + pkt_len) {
661 #ifdef TCPREASS_DEBUG
662 			printf("tcp_reass[%p]: append %u:%u(%u) to %u:%u(%u)\n",
663 			       tp, q->ipqe_seq, q->ipqe_seq + q->ipqe_len, q->ipqe_len,
664 			       pkt_seq, pkt_seq + pkt_len, pkt_len);
665 #endif
666 			pkt_len += q->ipqe_len;
667 			pkt_flags |= q->ipqe_flags;
668 			m_cat(m, q->ipqe_m);
669 			TAILQ_REMOVE(&tp->segq, q, ipqe_q);
670 			TAILQ_REMOVE(&tp->timeq, q, ipqe_timeq);
671 			tp->t_segqlen--;
672 			KASSERT(tp->t_segqlen >= 0);
673 			KASSERT(tp->t_segqlen != 0 ||
674 			    (TAILQ_EMPTY(&tp->segq) &&
675 			    TAILQ_EMPTY(&tp->timeq)));
676 			if (tiqe == NULL) {
677 				tiqe = q;
678 			} else {
679 				tcpipqent_free(q);
680 			}
681 			TCP_REASS_COUNTER_INCR(&tcp_reass_prepend);
682 			break;
683 		}
684 		/*
685 		 * If the fragment is before the segment, remember it.
686 		 * When this loop is terminated, p will contain the
687 		 * pointer to fragment that is right before the received
688 		 * segment.
689 		 */
690 		if (SEQ_LEQ(q->ipqe_seq, pkt_seq))
691 			p = q;
692 
693 		continue;
694 
695 		/*
696 		 * This is a common operation.  It also will allow
697 		 * to save doing a malloc/free in most instances.
698 		 */
699 	  free_ipqe:
700 		TAILQ_REMOVE(&tp->segq, q, ipqe_q);
701 		TAILQ_REMOVE(&tp->timeq, q, ipqe_timeq);
702 		tp->t_segqlen--;
703 		KASSERT(tp->t_segqlen >= 0);
704 		KASSERT(tp->t_segqlen != 0 ||
705 		    (TAILQ_EMPTY(&tp->segq) && TAILQ_EMPTY(&tp->timeq)));
706 		if (tiqe == NULL) {
707 			tiqe = q;
708 		} else {
709 			tcpipqent_free(q);
710 		}
711 	}
712 
713 #ifdef TCP_REASS_COUNTERS
714 	if (count > 7)
715 		TCP_REASS_COUNTER_INCR(&tcp_reass_iteration[0]);
716 	else if (count > 0)
717 		TCP_REASS_COUNTER_INCR(&tcp_reass_iteration[count]);
718 #endif
719 
720     insert_it:
721 
722 	/*
723 	 * Allocate a new queue entry since the received segment did not
724 	 * collapse onto any other out-of-order block; thus we are allocating
725 	 * a new block.  If it had collapsed, tiqe would not be NULL and
726 	 * we would be reusing it.
727 	 * XXX If we can't, just drop the packet.  XXX
728 	 */
729 	if (tiqe == NULL) {
730 		tiqe = tcpipqent_alloc();
731 		if (tiqe == NULL) {
732 			TCP_STATINC(TCP_STAT_RCVMEMDROP);
733 			m_freem(m);
734 			goto out;
735 		}
736 	}
737 
738 	/*
739 	 * Update the counters.
740 	 */
741 	tcps = TCP_STAT_GETREF();
742 	tcps[TCP_STAT_RCVOOPACK]++;
743 	tcps[TCP_STAT_RCVOOBYTE] += rcvoobyte;
744 	if (rcvpartdupbyte) {
745 	    tcps[TCP_STAT_RCVPARTDUPPACK]++;
746 	    tcps[TCP_STAT_RCVPARTDUPBYTE] += rcvpartdupbyte;
747 	}
748 	TCP_STAT_PUTREF();
749 
750 	/*
751 	 * Insert the new fragment queue entry into both queues.
752 	 */
753 	tiqe->ipqe_m = m;
754 	tiqe->ipre_mlast = m;
755 	tiqe->ipqe_seq = pkt_seq;
756 	tiqe->ipqe_len = pkt_len;
757 	tiqe->ipqe_flags = pkt_flags;
758 	if (p == NULL) {
759 		TAILQ_INSERT_HEAD(&tp->segq, tiqe, ipqe_q);
760 #ifdef TCPREASS_DEBUG
761 		if (tiqe->ipqe_seq != tp->rcv_nxt)
762 			printf("tcp_reass[%p]: insert %u:%u(%u) at front\n",
763 			       tp, pkt_seq, pkt_seq + pkt_len, pkt_len);
764 #endif
765 	} else {
766 		TAILQ_INSERT_AFTER(&tp->segq, p, tiqe, ipqe_q);
767 #ifdef TCPREASS_DEBUG
768 		printf("tcp_reass[%p]: insert %u:%u(%u) after %u:%u(%u)\n",
769 		       tp, pkt_seq, pkt_seq + pkt_len, pkt_len,
770 		       p->ipqe_seq, p->ipqe_seq + p->ipqe_len, p->ipqe_len);
771 #endif
772 	}
773 	tp->t_segqlen++;
774 
775 skip_replacement:
776 
777 	TAILQ_INSERT_HEAD(&tp->timeq, tiqe, ipqe_timeq);
778 
779 present:
780 	/*
781 	 * Present data to user, advancing rcv_nxt through
782 	 * completed sequence space.
783 	 */
784 	if (TCPS_HAVEESTABLISHED(tp->t_state) == 0)
785 		goto out;
786 	q = TAILQ_FIRST(&tp->segq);
787 	if (q == NULL || q->ipqe_seq != tp->rcv_nxt)
788 		goto out;
789 	if (tp->t_state == TCPS_SYN_RECEIVED && q->ipqe_len)
790 		goto out;
791 
792 	tp->rcv_nxt += q->ipqe_len;
793 	pkt_flags = q->ipqe_flags & TH_FIN;
794 	nd6_hint(tp);
795 
796 	TAILQ_REMOVE(&tp->segq, q, ipqe_q);
797 	TAILQ_REMOVE(&tp->timeq, q, ipqe_timeq);
798 	tp->t_segqlen--;
799 	KASSERT(tp->t_segqlen >= 0);
800 	KASSERT(tp->t_segqlen != 0 ||
801 	    (TAILQ_EMPTY(&tp->segq) && TAILQ_EMPTY(&tp->timeq)));
802 	if (so->so_state & SS_CANTRCVMORE)
803 		m_freem(q->ipqe_m);
804 	else
805 		sbappendstream(&so->so_rcv, q->ipqe_m);
806 	tcpipqent_free(q);
807 	TCP_REASS_UNLOCK(tp);
808 	sorwakeup(so);
809 	return (pkt_flags);
810 out:
811 	TCP_REASS_UNLOCK(tp);
812 	return (0);
813 }
814 
815 #ifdef INET6
816 int
817 tcp6_input(struct mbuf **mp, int *offp, int proto)
818 {
819 	struct mbuf *m = *mp;
820 
821 	/*
822 	 * draft-itojun-ipv6-tcp-to-anycast
823 	 * better place to put this in?
824 	 */
825 	if (m->m_flags & M_ANYCAST6) {
826 		struct ip6_hdr *ip6;
827 		if (m->m_len < sizeof(struct ip6_hdr)) {
828 			if ((m = m_pullup(m, sizeof(struct ip6_hdr))) == NULL) {
829 				TCP_STATINC(TCP_STAT_RCVSHORT);
830 				return IPPROTO_DONE;
831 			}
832 		}
833 		ip6 = mtod(m, struct ip6_hdr *);
834 		icmp6_error(m, ICMP6_DST_UNREACH, ICMP6_DST_UNREACH_ADDR,
835 		    (char *)&ip6->ip6_dst - (char *)ip6);
836 		return IPPROTO_DONE;
837 	}
838 
839 	tcp_input(m, *offp, proto);
840 	return IPPROTO_DONE;
841 }
842 #endif
843 
844 #ifdef INET
845 static void
846 tcp4_log_refused(const struct ip *ip, const struct tcphdr *th)
847 {
848 	char src[4*sizeof "123"];
849 	char dst[4*sizeof "123"];
850 
851 	if (ip) {
852 		strlcpy(src, inet_ntoa(ip->ip_src), sizeof(src));
853 		strlcpy(dst, inet_ntoa(ip->ip_dst), sizeof(dst));
854 	}
855 	else {
856 		strlcpy(src, "(unknown)", sizeof(src));
857 		strlcpy(dst, "(unknown)", sizeof(dst));
858 	}
859 	log(LOG_INFO,
860 	    "Connection attempt to TCP %s:%d from %s:%d\n",
861 	    dst, ntohs(th->th_dport),
862 	    src, ntohs(th->th_sport));
863 }
864 #endif
865 
866 #ifdef INET6
867 static void
868 tcp6_log_refused(const struct ip6_hdr *ip6, const struct tcphdr *th)
869 {
870 	char src[INET6_ADDRSTRLEN];
871 	char dst[INET6_ADDRSTRLEN];
872 
873 	if (ip6) {
874 		strlcpy(src, ip6_sprintf(&ip6->ip6_src), sizeof(src));
875 		strlcpy(dst, ip6_sprintf(&ip6->ip6_dst), sizeof(dst));
876 	}
877 	else {
878 		strlcpy(src, "(unknown v6)", sizeof(src));
879 		strlcpy(dst, "(unknown v6)", sizeof(dst));
880 	}
881 	log(LOG_INFO,
882 	    "Connection attempt to TCP [%s]:%d from [%s]:%d\n",
883 	    dst, ntohs(th->th_dport),
884 	    src, ntohs(th->th_sport));
885 }
886 #endif
887 
888 /*
889  * Checksum extended TCP header and data.
890  */
891 int
892 tcp_input_checksum(int af, struct mbuf *m, const struct tcphdr *th,
893     int toff, int off, int tlen)
894 {
895 
896 	/*
897 	 * XXX it's better to record and check if this mbuf is
898 	 * already checked.
899 	 */
900 
901 	switch (af) {
902 #ifdef INET
903 	case AF_INET:
904 		switch (m->m_pkthdr.csum_flags &
905 			((m->m_pkthdr.rcvif->if_csum_flags_rx & M_CSUM_TCPv4) |
906 			 M_CSUM_TCP_UDP_BAD | M_CSUM_DATA)) {
907 		case M_CSUM_TCPv4|M_CSUM_TCP_UDP_BAD:
908 			TCP_CSUM_COUNTER_INCR(&tcp_hwcsum_bad);
909 			goto badcsum;
910 
911 		case M_CSUM_TCPv4|M_CSUM_DATA: {
912 			u_int32_t hw_csum = m->m_pkthdr.csum_data;
913 
914 			TCP_CSUM_COUNTER_INCR(&tcp_hwcsum_data);
915 			if (m->m_pkthdr.csum_flags & M_CSUM_NO_PSEUDOHDR) {
916 				const struct ip *ip =
917 				    mtod(m, const struct ip *);
918 
919 				hw_csum = in_cksum_phdr(ip->ip_src.s_addr,
920 				    ip->ip_dst.s_addr,
921 				    htons(hw_csum + tlen + off + IPPROTO_TCP));
922 			}
923 			if ((hw_csum ^ 0xffff) != 0)
924 				goto badcsum;
925 			break;
926 		}
927 
928 		case M_CSUM_TCPv4:
929 			/* Checksum was okay. */
930 			TCP_CSUM_COUNTER_INCR(&tcp_hwcsum_ok);
931 			break;
932 
933 		default:
934 			/*
935 			 * Must compute it ourselves.  Maybe skip checksum
936 			 * on loopback interfaces.
937 			 */
938 			if (__predict_true(!(m->m_pkthdr.rcvif->if_flags &
939 					     IFF_LOOPBACK) ||
940 					   tcp_do_loopback_cksum)) {
941 				TCP_CSUM_COUNTER_INCR(&tcp_swcsum);
942 				if (in4_cksum(m, IPPROTO_TCP, toff,
943 					      tlen + off) != 0)
944 					goto badcsum;
945 			}
946 			break;
947 		}
948 		break;
949 #endif /* INET4 */
950 
951 #ifdef INET6
952 	case AF_INET6:
953 		switch (m->m_pkthdr.csum_flags &
954 			((m->m_pkthdr.rcvif->if_csum_flags_rx & M_CSUM_TCPv6) |
955 			 M_CSUM_TCP_UDP_BAD | M_CSUM_DATA)) {
956 		case M_CSUM_TCPv6|M_CSUM_TCP_UDP_BAD:
957 			TCP_CSUM_COUNTER_INCR(&tcp6_hwcsum_bad);
958 			goto badcsum;
959 
960 #if 0 /* notyet */
961 		case M_CSUM_TCPv6|M_CSUM_DATA:
962 #endif
963 
964 		case M_CSUM_TCPv6:
965 			/* Checksum was okay. */
966 			TCP_CSUM_COUNTER_INCR(&tcp6_hwcsum_ok);
967 			break;
968 
969 		default:
970 			/*
971 			 * Must compute it ourselves.  Maybe skip checksum
972 			 * on loopback interfaces.
973 			 */
974 			if (__predict_true((m->m_flags & M_LOOP) == 0 ||
975 			    tcp_do_loopback_cksum)) {
976 				TCP_CSUM_COUNTER_INCR(&tcp6_swcsum);
977 				if (in6_cksum(m, IPPROTO_TCP, toff,
978 				    tlen + off) != 0)
979 					goto badcsum;
980 			}
981 		}
982 		break;
983 #endif /* INET6 */
984 	}
985 
986 	return 0;
987 
988 badcsum:
989 	TCP_STATINC(TCP_STAT_RCVBADSUM);
990 	return -1;
991 }
992 
993 /* When a packet arrives addressed to a vestigial tcpbp, we
994  * nevertheless have to respond to it per the spec.
995  */
996 static void tcp_vtw_input(struct tcphdr *th, vestigial_inpcb_t *vp,
997 			  struct mbuf *m, int tlen, int multicast)
998 {
999 	int		tiflags;
1000 	int		todrop, dupseg;
1001 	uint32_t	t_flags = 0;
1002 	uint64_t	*tcps;
1003 
1004 	tiflags = th->th_flags;
1005 	todrop  = vp->rcv_nxt - th->th_seq;
1006 	dupseg  = false;
1007 
1008 	if (todrop > 0) {
1009 		if (tiflags & TH_SYN) {
1010 			tiflags &= ~TH_SYN;
1011 			++th->th_seq;
1012 			if (th->th_urp > 1)
1013 				--th->th_urp;
1014 			else {
1015 				tiflags &= ~TH_URG;
1016 				th->th_urp = 0;
1017 			}
1018 			--todrop;
1019 		}
1020 		if (todrop > tlen ||
1021 		    (todrop == tlen && (tiflags & TH_FIN) == 0)) {
1022 			/*
1023 			 * Any valid FIN or RST must be to the left of the
1024 			 * window.  At this point the FIN or RST must be a
1025 			 * duplicate or out of sequence; drop it.
1026 			 */
1027 			if (tiflags & TH_RST)
1028 				goto drop;
1029 			tiflags &= ~(TH_FIN|TH_RST);
1030 			/*
1031 			 * Send an ACK to resynchronize and drop any data.
1032 			 * But keep on processing for RST or ACK.
1033 			 */
1034 			t_flags |= TF_ACKNOW;
1035 			todrop = tlen;
1036 			dupseg = true;
1037 			tcps = TCP_STAT_GETREF();
1038 			tcps[TCP_STAT_RCVDUPPACK] += 1;
1039 			tcps[TCP_STAT_RCVDUPBYTE] += todrop;
1040 			TCP_STAT_PUTREF();
1041 		} else if ((tiflags & TH_RST)
1042 			   && th->th_seq != vp->rcv_nxt) {
1043 			/*
1044 			 * Test for reset before adjusting the sequence
1045 			 * number for overlapping data.
1046 			 */
1047 			goto dropafterack_ratelim;
1048 		} else {
1049 			tcps = TCP_STAT_GETREF();
1050 			tcps[TCP_STAT_RCVPARTDUPPACK] += 1;
1051 			tcps[TCP_STAT_RCVPARTDUPBYTE] += todrop;
1052 			TCP_STAT_PUTREF();
1053 		}
1054 
1055 //		tcp_new_dsack(tp, th->th_seq, todrop);
1056 //		hdroptlen += todrop;	/*drop from head afterwards*/
1057 
1058 		th->th_seq += todrop;
1059 		tlen -= todrop;
1060 
1061 		if (th->th_urp > todrop)
1062 			th->th_urp -= todrop;
1063 		else {
1064 			tiflags &= ~TH_URG;
1065 			th->th_urp = 0;
1066 		}
1067 	}
1068 
1069 	/*
1070 	 * If new data are received on a connection after the
1071 	 * user processes are gone, then RST the other end.
1072 	 */
1073 	if (tlen) {
1074 		TCP_STATINC(TCP_STAT_RCVAFTERCLOSE);
1075 		goto dropwithreset;
1076 	}
1077 
1078 	/*
1079 	 * If segment ends after window, drop trailing data
1080 	 * (and PUSH and FIN); if nothing left, just ACK.
1081 	 */
1082 	todrop = (th->th_seq + tlen) - (vp->rcv_nxt+vp->rcv_wnd);
1083 
1084 	if (todrop > 0) {
1085 		TCP_STATINC(TCP_STAT_RCVPACKAFTERWIN);
1086 		if (todrop >= tlen) {
1087 			/*
1088 			 * The segment actually starts after the window.
1089 			 * th->th_seq + tlen - vp->rcv_nxt - vp->rcv_wnd >= tlen
1090 			 * th->th_seq - vp->rcv_nxt - vp->rcv_wnd >= 0
1091 			 * th->th_seq >= vp->rcv_nxt + vp->rcv_wnd
1092 			 */
1093 			TCP_STATADD(TCP_STAT_RCVBYTEAFTERWIN, tlen);
1094 			/*
1095 			 * If a new connection request is received
1096 			 * while in TIME_WAIT, drop the old connection
1097 			 * and start over if the sequence numbers
1098 			 * are above the previous ones.
1099 			 */
1100 			if ((tiflags & TH_SYN)
1101 			    && SEQ_GT(th->th_seq, vp->rcv_nxt)) {
1102 				/* We only support this in the !NOFDREF case, which
1103 				 * is to say: not here.
1104 				 */
1105 				goto dropwithreset;;
1106 			}
1107 			/*
1108 			 * If window is closed can only take segments at
1109 			 * window edge, and have to drop data and PUSH from
1110 			 * incoming segments.  Continue processing, but
1111 			 * remember to ack.  Otherwise, drop segment
1112 			 * and (if not RST) ack.
1113 			 */
1114 			if (vp->rcv_wnd == 0 && th->th_seq == vp->rcv_nxt) {
1115 				t_flags |= TF_ACKNOW;
1116 				TCP_STATINC(TCP_STAT_RCVWINPROBE);
1117 			} else
1118 				goto dropafterack;
1119 		} else
1120 			TCP_STATADD(TCP_STAT_RCVBYTEAFTERWIN, todrop);
1121 		m_adj(m, -todrop);
1122 		tlen -= todrop;
1123 		tiflags &= ~(TH_PUSH|TH_FIN);
1124 	}
1125 
1126 	if (tiflags & TH_RST) {
1127 		if (th->th_seq != vp->rcv_nxt)
1128 			goto dropafterack_ratelim;
1129 
1130 		vtw_del(vp->ctl, vp->vtw);
1131 		goto drop;
1132 	}
1133 
1134 	/*
1135 	 * If the ACK bit is off we drop the segment and return.
1136 	 */
1137 	if ((tiflags & TH_ACK) == 0) {
1138 		if (t_flags & TF_ACKNOW)
1139 			goto dropafterack;
1140 		else
1141 			goto drop;
1142 	}
1143 
1144 	/*
1145 	 * In TIME_WAIT state the only thing that should arrive
1146 	 * is a retransmission of the remote FIN.  Acknowledge
1147 	 * it and restart the finack timer.
1148 	 */
1149 	vtw_restart(vp);
1150 	goto dropafterack;
1151 
1152 dropafterack:
1153 	/*
1154 	 * Generate an ACK dropping incoming segment if it occupies
1155 	 * sequence space, where the ACK reflects our state.
1156 	 */
1157 	if (tiflags & TH_RST)
1158 		goto drop;
1159 	goto dropafterack2;
1160 
1161 dropafterack_ratelim:
1162 	/*
1163 	 * We may want to rate-limit ACKs against SYN/RST attack.
1164 	 */
1165 	if (ppsratecheck(&tcp_ackdrop_ppslim_last, &tcp_ackdrop_ppslim_count,
1166 			 tcp_ackdrop_ppslim) == 0) {
1167 		/* XXX stat */
1168 		goto drop;
1169 	}
1170 	/* ...fall into dropafterack2... */
1171 
1172 dropafterack2:
1173 	(void)tcp_respond(0, m, m, th, th->th_seq + tlen, th->th_ack,
1174 			  TH_ACK);
1175 	return;
1176 
1177 dropwithreset:
1178 	/*
1179 	 * Generate a RST, dropping incoming segment.
1180 	 * Make ACK acceptable to originator of segment.
1181 	 */
1182 	if (tiflags & TH_RST)
1183 		goto drop;
1184 
1185 	if (tiflags & TH_ACK)
1186 		tcp_respond(0, m, m, th, (tcp_seq)0, th->th_ack, TH_RST);
1187 	else {
1188 		if (tiflags & TH_SYN)
1189 			++tlen;
1190 		(void)tcp_respond(0, m, m, th, th->th_seq + tlen, (tcp_seq)0,
1191 				  TH_RST|TH_ACK);
1192 	}
1193 	return;
1194 drop:
1195 	m_freem(m);
1196 }
1197 
1198 /*
1199  * TCP input routine, follows pages 65-76 of RFC 793 very closely.
1200  */
1201 void
1202 tcp_input(struct mbuf *m, ...)
1203 {
1204 	struct tcphdr *th;
1205 	struct ip *ip;
1206 	struct inpcb *inp;
1207 #ifdef INET6
1208 	struct ip6_hdr *ip6;
1209 	struct in6pcb *in6p;
1210 #endif
1211 	u_int8_t *optp = NULL;
1212 	int optlen = 0;
1213 	int len, tlen, toff, hdroptlen = 0;
1214 	struct tcpcb *tp = 0;
1215 	int tiflags;
1216 	struct socket *so = NULL;
1217 	int todrop, acked, ourfinisacked, needoutput = 0;
1218 	bool dupseg;
1219 #ifdef TCP_DEBUG
1220 	short ostate = 0;
1221 #endif
1222 	u_long tiwin;
1223 	struct tcp_opt_info opti;
1224 	int off, iphlen;
1225 	va_list ap;
1226 	int af;		/* af on the wire */
1227 	struct mbuf *tcp_saveti = NULL;
1228 	uint32_t ts_rtt;
1229 	uint8_t iptos;
1230 	uint64_t *tcps;
1231 	vestigial_inpcb_t vestige;
1232 
1233 	vestige.valid = 0;
1234 
1235 	MCLAIM(m, &tcp_rx_mowner);
1236 	va_start(ap, m);
1237 	toff = va_arg(ap, int);
1238 	(void)va_arg(ap, int);		/* ignore value, advance ap */
1239 	va_end(ap);
1240 
1241 	TCP_STATINC(TCP_STAT_RCVTOTAL);
1242 
1243 	memset(&opti, 0, sizeof(opti));
1244 	opti.ts_present = 0;
1245 	opti.maxseg = 0;
1246 
1247 	/*
1248 	 * RFC1122 4.2.3.10, p. 104: discard bcast/mcast SYN.
1249 	 *
1250 	 * TCP is, by definition, unicast, so we reject all
1251 	 * multicast outright.
1252 	 *
1253 	 * Note, there are additional src/dst address checks in
1254 	 * the AF-specific code below.
1255 	 */
1256 	if (m->m_flags & (M_BCAST|M_MCAST)) {
1257 		/* XXX stat */
1258 		goto drop;
1259 	}
1260 #ifdef INET6
1261 	if (m->m_flags & M_ANYCAST6) {
1262 		/* XXX stat */
1263 		goto drop;
1264 	}
1265 #endif
1266 
1267 	/*
1268 	 * Get IP and TCP header.
1269 	 * Note: IP leaves IP header in first mbuf.
1270 	 */
1271 	ip = mtod(m, struct ip *);
1272 	switch (ip->ip_v) {
1273 #ifdef INET
1274 	case 4:
1275 #ifdef INET6
1276 		ip6 = NULL;
1277 #endif
1278 		af = AF_INET;
1279 		iphlen = sizeof(struct ip);
1280 		IP6_EXTHDR_GET(th, struct tcphdr *, m, toff,
1281 			sizeof(struct tcphdr));
1282 		if (th == NULL) {
1283 			TCP_STATINC(TCP_STAT_RCVSHORT);
1284 			return;
1285 		}
1286 		/* We do the checksum after PCB lookup... */
1287 		len = ntohs(ip->ip_len);
1288 		tlen = len - toff;
1289 		iptos = ip->ip_tos;
1290 		break;
1291 #endif
1292 #ifdef INET6
1293 	case 6:
1294 		ip = NULL;
1295 		iphlen = sizeof(struct ip6_hdr);
1296 		af = AF_INET6;
1297 		ip6 = mtod(m, struct ip6_hdr *);
1298 		IP6_EXTHDR_GET(th, struct tcphdr *, m, toff,
1299 			sizeof(struct tcphdr));
1300 		if (th == NULL) {
1301 			TCP_STATINC(TCP_STAT_RCVSHORT);
1302 			return;
1303 		}
1304 
1305 		/* Be proactive about malicious use of IPv4 mapped address */
1306 		if (IN6_IS_ADDR_V4MAPPED(&ip6->ip6_src) ||
1307 		    IN6_IS_ADDR_V4MAPPED(&ip6->ip6_dst)) {
1308 			/* XXX stat */
1309 			goto drop;
1310 		}
1311 
1312 		/*
1313 		 * Be proactive about unspecified IPv6 address in source.
1314 		 * As we use all-zero to indicate unbounded/unconnected pcb,
1315 		 * unspecified IPv6 address can be used to confuse us.
1316 		 *
1317 		 * Note that packets with unspecified IPv6 destination is
1318 		 * already dropped in ip6_input.
1319 		 */
1320 		if (IN6_IS_ADDR_UNSPECIFIED(&ip6->ip6_src)) {
1321 			/* XXX stat */
1322 			goto drop;
1323 		}
1324 
1325 		/*
1326 		 * Make sure destination address is not multicast.
1327 		 * Source address checked in ip6_input().
1328 		 */
1329 		if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst)) {
1330 			/* XXX stat */
1331 			goto drop;
1332 		}
1333 
1334 		/* We do the checksum after PCB lookup... */
1335 		len = m->m_pkthdr.len;
1336 		tlen = len - toff;
1337 		iptos = (ntohl(ip6->ip6_flow) >> 20) & 0xff;
1338 		break;
1339 #endif
1340 	default:
1341 		m_freem(m);
1342 		return;
1343 	}
1344 
1345 	KASSERT(TCP_HDR_ALIGNED_P(th));
1346 
1347 	/*
1348 	 * Check that TCP offset makes sense,
1349 	 * pull out TCP options and adjust length.		XXX
1350 	 */
1351 	off = th->th_off << 2;
1352 	if (off < sizeof (struct tcphdr) || off > tlen) {
1353 		TCP_STATINC(TCP_STAT_RCVBADOFF);
1354 		goto drop;
1355 	}
1356 	tlen -= off;
1357 
1358 	/*
1359 	 * tcp_input() has been modified to use tlen to mean the TCP data
1360 	 * length throughout the function.  Other functions can use
1361 	 * m->m_pkthdr.len as the basis for calculating the TCP data length.
1362 	 * rja
1363 	 */
1364 
1365 	if (off > sizeof (struct tcphdr)) {
1366 		IP6_EXTHDR_GET(th, struct tcphdr *, m, toff, off);
1367 		if (th == NULL) {
1368 			TCP_STATINC(TCP_STAT_RCVSHORT);
1369 			return;
1370 		}
1371 		/*
1372 		 * NOTE: ip/ip6 will not be affected by m_pulldown()
1373 		 * (as they're before toff) and we don't need to update those.
1374 		 */
1375 		KASSERT(TCP_HDR_ALIGNED_P(th));
1376 		optlen = off - sizeof (struct tcphdr);
1377 		optp = ((u_int8_t *)th) + sizeof(struct tcphdr);
1378 		/*
1379 		 * Do quick retrieval of timestamp options ("options
1380 		 * prediction?").  If timestamp is the only option and it's
1381 		 * formatted as recommended in RFC 1323 appendix A, we
1382 		 * quickly get the values now and not bother calling
1383 		 * tcp_dooptions(), etc.
1384 		 */
1385 		if ((optlen == TCPOLEN_TSTAMP_APPA ||
1386 		     (optlen > TCPOLEN_TSTAMP_APPA &&
1387 			optp[TCPOLEN_TSTAMP_APPA] == TCPOPT_EOL)) &&
1388 		     *(u_int32_t *)optp == htonl(TCPOPT_TSTAMP_HDR) &&
1389 		     (th->th_flags & TH_SYN) == 0) {
1390 			opti.ts_present = 1;
1391 			opti.ts_val = ntohl(*(u_int32_t *)(optp + 4));
1392 			opti.ts_ecr = ntohl(*(u_int32_t *)(optp + 8));
1393 			optp = NULL;	/* we've parsed the options */
1394 		}
1395 	}
1396 	tiflags = th->th_flags;
1397 
1398 	/*
1399 	 * Locate pcb for segment.
1400 	 */
1401 findpcb:
1402 	inp = NULL;
1403 #ifdef INET6
1404 	in6p = NULL;
1405 #endif
1406 	switch (af) {
1407 #ifdef INET
1408 	case AF_INET:
1409 		inp = in_pcblookup_connect(&tcbtable, ip->ip_src, th->th_sport,
1410 					   ip->ip_dst, th->th_dport,
1411 					   &vestige);
1412 		if (inp == 0 && !vestige.valid) {
1413 			TCP_STATINC(TCP_STAT_PCBHASHMISS);
1414 			inp = in_pcblookup_bind(&tcbtable, ip->ip_dst, th->th_dport);
1415 		}
1416 #ifdef INET6
1417 		if (inp == 0 && !vestige.valid) {
1418 			struct in6_addr s, d;
1419 
1420 			/* mapped addr case */
1421 			memset(&s, 0, sizeof(s));
1422 			s.s6_addr16[5] = htons(0xffff);
1423 			bcopy(&ip->ip_src, &s.s6_addr32[3], sizeof(ip->ip_src));
1424 			memset(&d, 0, sizeof(d));
1425 			d.s6_addr16[5] = htons(0xffff);
1426 			bcopy(&ip->ip_dst, &d.s6_addr32[3], sizeof(ip->ip_dst));
1427 			in6p = in6_pcblookup_connect(&tcbtable, &s,
1428 						     th->th_sport, &d, th->th_dport,
1429 						     0, &vestige);
1430 			if (in6p == 0 && !vestige.valid) {
1431 				TCP_STATINC(TCP_STAT_PCBHASHMISS);
1432 				in6p = in6_pcblookup_bind(&tcbtable, &d,
1433 				    th->th_dport, 0);
1434 			}
1435 		}
1436 #endif
1437 #ifndef INET6
1438 		if (inp == 0 && !vestige.valid)
1439 #else
1440 		if (inp == 0 && in6p == 0 && !vestige.valid)
1441 #endif
1442 		{
1443 			TCP_STATINC(TCP_STAT_NOPORT);
1444 			if (tcp_log_refused &&
1445 			    (tiflags & (TH_RST|TH_ACK|TH_SYN)) == TH_SYN) {
1446 				tcp4_log_refused(ip, th);
1447 			}
1448 			tcp_fields_to_host(th);
1449 			goto dropwithreset_ratelim;
1450 		}
1451 #if defined(IPSEC)
1452 		if (inp && (inp->inp_socket->so_options & SO_ACCEPTCONN) == 0 &&
1453 		    ipsec4_in_reject(m, inp)) {
1454 			IPSEC_STATINC(IPSEC_STAT_IN_POLVIO);
1455 			goto drop;
1456 		}
1457 #ifdef INET6
1458 		else if (in6p &&
1459 		    (in6p->in6p_socket->so_options & SO_ACCEPTCONN) == 0 &&
1460 		    ipsec6_in_reject_so(m, in6p->in6p_socket)) {
1461 			IPSEC_STATINC(IPSEC_STAT_IN_POLVIO);
1462 			goto drop;
1463 		}
1464 #endif
1465 #endif /*IPSEC*/
1466 		break;
1467 #endif /*INET*/
1468 #ifdef INET6
1469 	case AF_INET6:
1470 	    {
1471 		int faith;
1472 
1473 #if defined(NFAITH) && NFAITH > 0
1474 		faith = faithprefix(&ip6->ip6_dst);
1475 #else
1476 		faith = 0;
1477 #endif
1478 		in6p = in6_pcblookup_connect(&tcbtable, &ip6->ip6_src,
1479 					     th->th_sport, &ip6->ip6_dst, th->th_dport, faith, &vestige);
1480 		if (!in6p && !vestige.valid) {
1481 			TCP_STATINC(TCP_STAT_PCBHASHMISS);
1482 			in6p = in6_pcblookup_bind(&tcbtable, &ip6->ip6_dst,
1483 				th->th_dport, faith);
1484 		}
1485 		if (!in6p && !vestige.valid) {
1486 			TCP_STATINC(TCP_STAT_NOPORT);
1487 			if (tcp_log_refused &&
1488 			    (tiflags & (TH_RST|TH_ACK|TH_SYN)) == TH_SYN) {
1489 				tcp6_log_refused(ip6, th);
1490 			}
1491 			tcp_fields_to_host(th);
1492 			goto dropwithreset_ratelim;
1493 		}
1494 #if defined(IPSEC)
1495 		if (in6p
1496 		    && (in6p->in6p_socket->so_options & SO_ACCEPTCONN) == 0
1497 		    && ipsec6_in_reject(m, in6p)) {
1498 			IPSEC6_STATINC(IPSEC_STAT_IN_POLVIO);
1499 			goto drop;
1500 		}
1501 #endif /*IPSEC*/
1502 		break;
1503 	    }
1504 #endif
1505 	}
1506 
1507 	/*
1508 	 * If the state is CLOSED (i.e., TCB does not exist) then
1509 	 * all data in the incoming segment is discarded.
1510 	 * If the TCB exists but is in CLOSED state, it is embryonic,
1511 	 * but should either do a listen or a connect soon.
1512 	 */
1513 	tp = NULL;
1514 	so = NULL;
1515 	if (inp) {
1516 		/* Check the minimum TTL for socket. */
1517 		if (ip->ip_ttl < inp->inp_ip_minttl)
1518 			goto drop;
1519 
1520 		tp = intotcpcb(inp);
1521 		so = inp->inp_socket;
1522 	}
1523 #ifdef INET6
1524 	else if (in6p) {
1525 		tp = in6totcpcb(in6p);
1526 		so = in6p->in6p_socket;
1527 	}
1528 #endif
1529 	else if (vestige.valid) {
1530 		int mc = 0;
1531 
1532 		/* We do not support the resurrection of vtw tcpcps.
1533 		 */
1534 		if (tcp_input_checksum(af, m, th, toff, off, tlen))
1535 			goto badcsum;
1536 
1537 		switch (af) {
1538 #ifdef INET6
1539 		case AF_INET6:
1540 			mc = IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst);
1541 			break;
1542 #endif
1543 
1544 		case AF_INET:
1545 			mc = (IN_MULTICAST(ip->ip_dst.s_addr)
1546 			      || in_broadcast(ip->ip_dst, m->m_pkthdr.rcvif));
1547 			break;
1548 		}
1549 
1550 		tcp_fields_to_host(th);
1551 		tcp_vtw_input(th, &vestige, m, tlen, mc);
1552 		m = 0;
1553 		goto drop;
1554 	}
1555 
1556 	if (tp == 0) {
1557 		tcp_fields_to_host(th);
1558 		goto dropwithreset_ratelim;
1559 	}
1560 	if (tp->t_state == TCPS_CLOSED)
1561 		goto drop;
1562 
1563 	KASSERT(so->so_lock == softnet_lock);
1564 	KASSERT(solocked(so));
1565 
1566 	/*
1567 	 * Checksum extended TCP header and data.
1568 	 */
1569 	if (tcp_input_checksum(af, m, th, toff, off, tlen))
1570 		goto badcsum;
1571 
1572 	tcp_fields_to_host(th);
1573 
1574 	/* Unscale the window into a 32-bit value. */
1575 	if ((tiflags & TH_SYN) == 0)
1576 		tiwin = th->th_win << tp->snd_scale;
1577 	else
1578 		tiwin = th->th_win;
1579 
1580 #ifdef INET6
1581 	/* save packet options if user wanted */
1582 	if (in6p && (in6p->in6p_flags & IN6P_CONTROLOPTS)) {
1583 		if (in6p->in6p_options) {
1584 			m_freem(in6p->in6p_options);
1585 			in6p->in6p_options = 0;
1586 		}
1587 		KASSERT(ip6 != NULL);
1588 		ip6_savecontrol(in6p, &in6p->in6p_options, ip6, m);
1589 	}
1590 #endif
1591 
1592 	if (so->so_options & (SO_DEBUG|SO_ACCEPTCONN)) {
1593 		union syn_cache_sa src;
1594 		union syn_cache_sa dst;
1595 
1596 		memset(&src, 0, sizeof(src));
1597 		memset(&dst, 0, sizeof(dst));
1598 		switch (af) {
1599 #ifdef INET
1600 		case AF_INET:
1601 			src.sin.sin_len = sizeof(struct sockaddr_in);
1602 			src.sin.sin_family = AF_INET;
1603 			src.sin.sin_addr = ip->ip_src;
1604 			src.sin.sin_port = th->th_sport;
1605 
1606 			dst.sin.sin_len = sizeof(struct sockaddr_in);
1607 			dst.sin.sin_family = AF_INET;
1608 			dst.sin.sin_addr = ip->ip_dst;
1609 			dst.sin.sin_port = th->th_dport;
1610 			break;
1611 #endif
1612 #ifdef INET6
1613 		case AF_INET6:
1614 			src.sin6.sin6_len = sizeof(struct sockaddr_in6);
1615 			src.sin6.sin6_family = AF_INET6;
1616 			src.sin6.sin6_addr = ip6->ip6_src;
1617 			src.sin6.sin6_port = th->th_sport;
1618 
1619 			dst.sin6.sin6_len = sizeof(struct sockaddr_in6);
1620 			dst.sin6.sin6_family = AF_INET6;
1621 			dst.sin6.sin6_addr = ip6->ip6_dst;
1622 			dst.sin6.sin6_port = th->th_dport;
1623 			break;
1624 #endif /* INET6 */
1625 		default:
1626 			goto badsyn;	/*sanity*/
1627 		}
1628 
1629 		if (so->so_options & SO_DEBUG) {
1630 #ifdef TCP_DEBUG
1631 			ostate = tp->t_state;
1632 #endif
1633 
1634 			tcp_saveti = NULL;
1635 			if (iphlen + sizeof(struct tcphdr) > MHLEN)
1636 				goto nosave;
1637 
1638 			if (m->m_len > iphlen && (m->m_flags & M_EXT) == 0) {
1639 				tcp_saveti = m_copym(m, 0, iphlen, M_DONTWAIT);
1640 				if (!tcp_saveti)
1641 					goto nosave;
1642 			} else {
1643 				MGETHDR(tcp_saveti, M_DONTWAIT, MT_HEADER);
1644 				if (!tcp_saveti)
1645 					goto nosave;
1646 				MCLAIM(m, &tcp_mowner);
1647 				tcp_saveti->m_len = iphlen;
1648 				m_copydata(m, 0, iphlen,
1649 				    mtod(tcp_saveti, void *));
1650 			}
1651 
1652 			if (M_TRAILINGSPACE(tcp_saveti) < sizeof(struct tcphdr)) {
1653 				m_freem(tcp_saveti);
1654 				tcp_saveti = NULL;
1655 			} else {
1656 				tcp_saveti->m_len += sizeof(struct tcphdr);
1657 				memcpy(mtod(tcp_saveti, char *) + iphlen, th,
1658 				    sizeof(struct tcphdr));
1659 			}
1660 	nosave:;
1661 		}
1662 		if (so->so_options & SO_ACCEPTCONN) {
1663 			if ((tiflags & (TH_RST|TH_ACK|TH_SYN)) != TH_SYN) {
1664 				if (tiflags & TH_RST) {
1665 					syn_cache_reset(&src.sa, &dst.sa, th);
1666 				} else if ((tiflags & (TH_ACK|TH_SYN)) ==
1667 				    (TH_ACK|TH_SYN)) {
1668 					/*
1669 					 * Received a SYN,ACK.  This should
1670 					 * never happen while we are in
1671 					 * LISTEN.  Send an RST.
1672 					 */
1673 					goto badsyn;
1674 				} else if (tiflags & TH_ACK) {
1675 					so = syn_cache_get(&src.sa, &dst.sa,
1676 						th, toff, tlen, so, m);
1677 					if (so == NULL) {
1678 						/*
1679 						 * We don't have a SYN for
1680 						 * this ACK; send an RST.
1681 						 */
1682 						goto badsyn;
1683 					} else if (so ==
1684 					    (struct socket *)(-1)) {
1685 						/*
1686 						 * We were unable to create
1687 						 * the connection.  If the
1688 						 * 3-way handshake was
1689 						 * completed, and RST has
1690 						 * been sent to the peer.
1691 						 * Since the mbuf might be
1692 						 * in use for the reply,
1693 						 * do not free it.
1694 						 */
1695 						m = NULL;
1696 					} else {
1697 						/*
1698 						 * We have created a
1699 						 * full-blown connection.
1700 						 */
1701 						tp = NULL;
1702 						inp = NULL;
1703 #ifdef INET6
1704 						in6p = NULL;
1705 #endif
1706 						switch (so->so_proto->pr_domain->dom_family) {
1707 #ifdef INET
1708 						case AF_INET:
1709 							inp = sotoinpcb(so);
1710 							tp = intotcpcb(inp);
1711 							break;
1712 #endif
1713 #ifdef INET6
1714 						case AF_INET6:
1715 							in6p = sotoin6pcb(so);
1716 							tp = in6totcpcb(in6p);
1717 							break;
1718 #endif
1719 						}
1720 						if (tp == NULL)
1721 							goto badsyn;	/*XXX*/
1722 						tiwin <<= tp->snd_scale;
1723 						goto after_listen;
1724 					}
1725 				} else {
1726 					/*
1727 					 * None of RST, SYN or ACK was set.
1728 					 * This is an invalid packet for a
1729 					 * TCB in LISTEN state.  Send a RST.
1730 					 */
1731 					goto badsyn;
1732 				}
1733 			} else {
1734 				/*
1735 				 * Received a SYN.
1736 				 *
1737 				 * RFC1122 4.2.3.10, p. 104: discard bcast/mcast SYN
1738 				 */
1739 				if (m->m_flags & (M_BCAST|M_MCAST))
1740 					goto drop;
1741 
1742 				switch (af) {
1743 #ifdef INET6
1744 				case AF_INET6:
1745 					if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst))
1746 						goto drop;
1747 					break;
1748 #endif /* INET6 */
1749 				case AF_INET:
1750 					if (IN_MULTICAST(ip->ip_dst.s_addr) ||
1751 					    in_broadcast(ip->ip_dst, m->m_pkthdr.rcvif))
1752 						goto drop;
1753 				break;
1754 				}
1755 
1756 #ifdef INET6
1757 				/*
1758 				 * If deprecated address is forbidden, we do
1759 				 * not accept SYN to deprecated interface
1760 				 * address to prevent any new inbound
1761 				 * connection from getting established.
1762 				 * When we do not accept SYN, we send a TCP
1763 				 * RST, with deprecated source address (instead
1764 				 * of dropping it).  We compromise it as it is
1765 				 * much better for peer to send a RST, and
1766 				 * RST will be the final packet for the
1767 				 * exchange.
1768 				 *
1769 				 * If we do not forbid deprecated addresses, we
1770 				 * accept the SYN packet.  RFC2462 does not
1771 				 * suggest dropping SYN in this case.
1772 				 * If we decipher RFC2462 5.5.4, it says like
1773 				 * this:
1774 				 * 1. use of deprecated addr with existing
1775 				 *    communication is okay - "SHOULD continue
1776 				 *    to be used"
1777 				 * 2. use of it with new communication:
1778 				 *   (2a) "SHOULD NOT be used if alternate
1779 				 *        address with sufficient scope is
1780 				 *        available"
1781 				 *   (2b) nothing mentioned otherwise.
1782 				 * Here we fall into (2b) case as we have no
1783 				 * choice in our source address selection - we
1784 				 * must obey the peer.
1785 				 *
1786 				 * The wording in RFC2462 is confusing, and
1787 				 * there are multiple description text for
1788 				 * deprecated address handling - worse, they
1789 				 * are not exactly the same.  I believe 5.5.4
1790 				 * is the best one, so we follow 5.5.4.
1791 				 */
1792 				if (af == AF_INET6 && !ip6_use_deprecated) {
1793 					struct in6_ifaddr *ia6;
1794 					if ((ia6 = in6ifa_ifpwithaddr(m->m_pkthdr.rcvif,
1795 					    &ip6->ip6_dst)) &&
1796 					    (ia6->ia6_flags & IN6_IFF_DEPRECATED)) {
1797 						tp = NULL;
1798 						goto dropwithreset;
1799 					}
1800 				}
1801 #endif
1802 
1803 #if defined(IPSEC)
1804 				switch (af) {
1805 #ifdef INET
1806 				case AF_INET:
1807 					if (ipsec4_in_reject_so(m, so)) {
1808 						IPSEC_STATINC(IPSEC_STAT_IN_POLVIO);
1809 						tp = NULL;
1810 						goto dropwithreset;
1811 					}
1812 					break;
1813 #endif
1814 #ifdef INET6
1815 				case AF_INET6:
1816 					if (ipsec6_in_reject_so(m, so)) {
1817 						IPSEC6_STATINC(IPSEC_STAT_IN_POLVIO);
1818 						tp = NULL;
1819 						goto dropwithreset;
1820 					}
1821 					break;
1822 #endif /*INET6*/
1823 				}
1824 #endif /*IPSEC*/
1825 
1826 				/*
1827 				 * LISTEN socket received a SYN
1828 				 * from itself?  This can't possibly
1829 				 * be valid; drop the packet.
1830 				 */
1831 				if (th->th_sport == th->th_dport) {
1832 					int i;
1833 
1834 					switch (af) {
1835 #ifdef INET
1836 					case AF_INET:
1837 						i = in_hosteq(ip->ip_src, ip->ip_dst);
1838 						break;
1839 #endif
1840 #ifdef INET6
1841 					case AF_INET6:
1842 						i = IN6_ARE_ADDR_EQUAL(&ip6->ip6_src, &ip6->ip6_dst);
1843 						break;
1844 #endif
1845 					default:
1846 						i = 1;
1847 					}
1848 					if (i) {
1849 						TCP_STATINC(TCP_STAT_BADSYN);
1850 						goto drop;
1851 					}
1852 				}
1853 
1854 				/*
1855 				 * SYN looks ok; create compressed TCP
1856 				 * state for it.
1857 				 */
1858 				if (so->so_qlen <= so->so_qlimit &&
1859 				    syn_cache_add(&src.sa, &dst.sa, th, tlen,
1860 						so, m, optp, optlen, &opti))
1861 					m = NULL;
1862 			}
1863 			goto drop;
1864 		}
1865 	}
1866 
1867 after_listen:
1868 #ifdef DIAGNOSTIC
1869 	/*
1870 	 * Should not happen now that all embryonic connections
1871 	 * are handled with compressed state.
1872 	 */
1873 	if (tp->t_state == TCPS_LISTEN)
1874 		panic("tcp_input: TCPS_LISTEN");
1875 #endif
1876 
1877 	/*
1878 	 * Segment received on connection.
1879 	 * Reset idle time and keep-alive timer.
1880 	 */
1881 	tp->t_rcvtime = tcp_now;
1882 	if (TCPS_HAVEESTABLISHED(tp->t_state))
1883 		TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepidle);
1884 
1885 	/*
1886 	 * Process options.
1887 	 */
1888 #ifdef TCP_SIGNATURE
1889 	if (optp || (tp->t_flags & TF_SIGNATURE))
1890 #else
1891 	if (optp)
1892 #endif
1893 		if (tcp_dooptions(tp, optp, optlen, th, m, toff, &opti) < 0)
1894 			goto drop;
1895 
1896 	if (TCP_SACK_ENABLED(tp)) {
1897 		tcp_del_sackholes(tp, th);
1898 	}
1899 
1900 	if (TCP_ECN_ALLOWED(tp)) {
1901 		if (tiflags & TH_CWR) {
1902 			tp->t_flags &= ~TF_ECN_SND_ECE;
1903 		}
1904 		switch (iptos & IPTOS_ECN_MASK) {
1905 		case IPTOS_ECN_CE:
1906 			tp->t_flags |= TF_ECN_SND_ECE;
1907 			TCP_STATINC(TCP_STAT_ECN_CE);
1908 			break;
1909 		case IPTOS_ECN_ECT0:
1910 			TCP_STATINC(TCP_STAT_ECN_ECT);
1911 			break;
1912 		case IPTOS_ECN_ECT1:
1913 			/* XXX: ignore for now -- rpaulo */
1914 			break;
1915 		}
1916 		/*
1917 		 * Congestion experienced.
1918 		 * Ignore if we are already trying to recover.
1919 		 */
1920 		if ((tiflags & TH_ECE) && SEQ_GEQ(tp->snd_una, tp->snd_recover))
1921 			tp->t_congctl->cong_exp(tp);
1922 	}
1923 
1924 	if (opti.ts_present && opti.ts_ecr) {
1925 		/*
1926 		 * Calculate the RTT from the returned time stamp and the
1927 		 * connection's time base.  If the time stamp is later than
1928 		 * the current time, or is extremely old, fall back to non-1323
1929 		 * RTT calculation.  Since ts_rtt is unsigned, we can test both
1930 		 * at the same time.
1931 		 *
1932 		 * Note that ts_rtt is in units of slow ticks (500
1933 		 * ms).  Since most earthbound RTTs are < 500 ms,
1934 		 * observed values will have large quantization noise.
1935 		 * Our smoothed RTT is then the fraction of observed
1936 		 * samples that are 1 tick instead of 0 (times 500
1937 		 * ms).
1938 		 *
1939 		 * ts_rtt is increased by 1 to denote a valid sample,
1940 		 * with 0 indicating an invalid measurement.  This
1941 		 * extra 1 must be removed when ts_rtt is used, or
1942 		 * else an an erroneous extra 500 ms will result.
1943 		 */
1944 		ts_rtt = TCP_TIMESTAMP(tp) - opti.ts_ecr + 1;
1945 		if (ts_rtt > TCP_PAWS_IDLE)
1946 			ts_rtt = 0;
1947 	} else {
1948 		ts_rtt = 0;
1949 	}
1950 
1951 	/*
1952 	 * Header prediction: check for the two common cases
1953 	 * of a uni-directional data xfer.  If the packet has
1954 	 * no control flags, is in-sequence, the window didn't
1955 	 * change and we're not retransmitting, it's a
1956 	 * candidate.  If the length is zero and the ack moved
1957 	 * forward, we're the sender side of the xfer.  Just
1958 	 * free the data acked & wake any higher level process
1959 	 * that was blocked waiting for space.  If the length
1960 	 * is non-zero and the ack didn't move, we're the
1961 	 * receiver side.  If we're getting packets in-order
1962 	 * (the reassembly queue is empty), add the data to
1963 	 * the socket buffer and note that we need a delayed ack.
1964 	 */
1965 	if (tp->t_state == TCPS_ESTABLISHED &&
1966 	    (tiflags & (TH_SYN|TH_FIN|TH_RST|TH_URG|TH_ECE|TH_CWR|TH_ACK))
1967 	        == TH_ACK &&
1968 	    (!opti.ts_present || TSTMP_GEQ(opti.ts_val, tp->ts_recent)) &&
1969 	    th->th_seq == tp->rcv_nxt &&
1970 	    tiwin && tiwin == tp->snd_wnd &&
1971 	    tp->snd_nxt == tp->snd_max) {
1972 
1973 		/*
1974 		 * If last ACK falls within this segment's sequence numbers,
1975 		 * record the timestamp.
1976 		 * NOTE that the test is modified according to the latest
1977 		 * proposal of the tcplw@cray.com list (Braden 1993/04/26).
1978 		 *
1979 		 * note that we already know
1980 		 *	TSTMP_GEQ(opti.ts_val, tp->ts_recent)
1981 		 */
1982 		if (opti.ts_present &&
1983 		    SEQ_LEQ(th->th_seq, tp->last_ack_sent)) {
1984 			tp->ts_recent_age = tcp_now;
1985 			tp->ts_recent = opti.ts_val;
1986 		}
1987 
1988 		if (tlen == 0) {
1989 			/* Ack prediction. */
1990 			if (SEQ_GT(th->th_ack, tp->snd_una) &&
1991 			    SEQ_LEQ(th->th_ack, tp->snd_max) &&
1992 			    tp->snd_cwnd >= tp->snd_wnd &&
1993 			    tp->t_partialacks < 0) {
1994 				/*
1995 				 * this is a pure ack for outstanding data.
1996 				 */
1997 				if (ts_rtt)
1998 					tcp_xmit_timer(tp, ts_rtt - 1);
1999 				else if (tp->t_rtttime &&
2000 				    SEQ_GT(th->th_ack, tp->t_rtseq))
2001 					tcp_xmit_timer(tp,
2002 					  tcp_now - tp->t_rtttime);
2003 				acked = th->th_ack - tp->snd_una;
2004 				tcps = TCP_STAT_GETREF();
2005 				tcps[TCP_STAT_PREDACK]++;
2006 				tcps[TCP_STAT_RCVACKPACK]++;
2007 				tcps[TCP_STAT_RCVACKBYTE] += acked;
2008 				TCP_STAT_PUTREF();
2009 				nd6_hint(tp);
2010 
2011 				if (acked > (tp->t_lastoff - tp->t_inoff))
2012 					tp->t_lastm = NULL;
2013 				sbdrop(&so->so_snd, acked);
2014 				tp->t_lastoff -= acked;
2015 
2016 				icmp_check(tp, th, acked);
2017 
2018 				tp->snd_una = th->th_ack;
2019 				tp->snd_fack = tp->snd_una;
2020 				if (SEQ_LT(tp->snd_high, tp->snd_una))
2021 					tp->snd_high = tp->snd_una;
2022 				m_freem(m);
2023 
2024 				/*
2025 				 * If all outstanding data are acked, stop
2026 				 * retransmit timer, otherwise restart timer
2027 				 * using current (possibly backed-off) value.
2028 				 * If process is waiting for space,
2029 				 * wakeup/selnotify/signal.  If data
2030 				 * are ready to send, let tcp_output
2031 				 * decide between more output or persist.
2032 				 */
2033 				if (tp->snd_una == tp->snd_max)
2034 					TCP_TIMER_DISARM(tp, TCPT_REXMT);
2035 				else if (TCP_TIMER_ISARMED(tp,
2036 				    TCPT_PERSIST) == 0)
2037 					TCP_TIMER_ARM(tp, TCPT_REXMT,
2038 					    tp->t_rxtcur);
2039 
2040 				sowwakeup(so);
2041 				if (so->so_snd.sb_cc) {
2042 					KERNEL_LOCK(1, NULL);
2043 					(void) tcp_output(tp);
2044 					KERNEL_UNLOCK_ONE(NULL);
2045 				}
2046 				if (tcp_saveti)
2047 					m_freem(tcp_saveti);
2048 				return;
2049 			}
2050 		} else if (th->th_ack == tp->snd_una &&
2051 		    TAILQ_FIRST(&tp->segq) == NULL &&
2052 		    tlen <= sbspace(&so->so_rcv)) {
2053 			int newsize = 0;	/* automatic sockbuf scaling */
2054 
2055 			/*
2056 			 * this is a pure, in-sequence data packet
2057 			 * with nothing on the reassembly queue and
2058 			 * we have enough buffer space to take it.
2059 			 */
2060 			tp->rcv_nxt += tlen;
2061 			tcps = TCP_STAT_GETREF();
2062 			tcps[TCP_STAT_PREDDAT]++;
2063 			tcps[TCP_STAT_RCVPACK]++;
2064 			tcps[TCP_STAT_RCVBYTE] += tlen;
2065 			TCP_STAT_PUTREF();
2066 			nd6_hint(tp);
2067 
2068 		/*
2069 		 * Automatic sizing enables the performance of large buffers
2070 		 * and most of the efficiency of small ones by only allocating
2071 		 * space when it is needed.
2072 		 *
2073 		 * On the receive side the socket buffer memory is only rarely
2074 		 * used to any significant extent.  This allows us to be much
2075 		 * more aggressive in scaling the receive socket buffer.  For
2076 		 * the case that the buffer space is actually used to a large
2077 		 * extent and we run out of kernel memory we can simply drop
2078 		 * the new segments; TCP on the sender will just retransmit it
2079 		 * later.  Setting the buffer size too big may only consume too
2080 		 * much kernel memory if the application doesn't read() from
2081 		 * the socket or packet loss or reordering makes use of the
2082 		 * reassembly queue.
2083 		 *
2084 		 * The criteria to step up the receive buffer one notch are:
2085 		 *  1. the number of bytes received during the time it takes
2086 		 *     one timestamp to be reflected back to us (the RTT);
2087 		 *  2. received bytes per RTT is within seven eighth of the
2088 		 *     current socket buffer size;
2089 		 *  3. receive buffer size has not hit maximal automatic size;
2090 		 *
2091 		 * This algorithm does one step per RTT at most and only if
2092 		 * we receive a bulk stream w/o packet losses or reorderings.
2093 		 * Shrinking the buffer during idle times is not necessary as
2094 		 * it doesn't consume any memory when idle.
2095 		 *
2096 		 * TODO: Only step up if the application is actually serving
2097 		 * the buffer to better manage the socket buffer resources.
2098 		 */
2099 			if (tcp_do_autorcvbuf &&
2100 			    opti.ts_ecr &&
2101 			    (so->so_rcv.sb_flags & SB_AUTOSIZE)) {
2102 				if (opti.ts_ecr > tp->rfbuf_ts &&
2103 				    opti.ts_ecr - tp->rfbuf_ts < PR_SLOWHZ) {
2104 					if (tp->rfbuf_cnt >
2105 					    (so->so_rcv.sb_hiwat / 8 * 7) &&
2106 					    so->so_rcv.sb_hiwat <
2107 					    tcp_autorcvbuf_max) {
2108 						newsize =
2109 						    min(so->so_rcv.sb_hiwat +
2110 						    tcp_autorcvbuf_inc,
2111 						    tcp_autorcvbuf_max);
2112 					}
2113 					/* Start over with next RTT. */
2114 					tp->rfbuf_ts = 0;
2115 					tp->rfbuf_cnt = 0;
2116 				} else
2117 					tp->rfbuf_cnt += tlen;	/* add up */
2118 			}
2119 
2120 			/*
2121 			 * Drop TCP, IP headers and TCP options then add data
2122 			 * to socket buffer.
2123 			 */
2124 			if (so->so_state & SS_CANTRCVMORE)
2125 				m_freem(m);
2126 			else {
2127 				/*
2128 				 * Set new socket buffer size.
2129 				 * Give up when limit is reached.
2130 				 */
2131 				if (newsize)
2132 					if (!sbreserve(&so->so_rcv,
2133 					    newsize, so))
2134 						so->so_rcv.sb_flags &= ~SB_AUTOSIZE;
2135 				m_adj(m, toff + off);
2136 				sbappendstream(&so->so_rcv, m);
2137 			}
2138 			sorwakeup(so);
2139 			tcp_setup_ack(tp, th);
2140 			if (tp->t_flags & TF_ACKNOW) {
2141 				KERNEL_LOCK(1, NULL);
2142 				(void) tcp_output(tp);
2143 				KERNEL_UNLOCK_ONE(NULL);
2144 			}
2145 			if (tcp_saveti)
2146 				m_freem(tcp_saveti);
2147 			return;
2148 		}
2149 	}
2150 
2151 	/*
2152 	 * Compute mbuf offset to TCP data segment.
2153 	 */
2154 	hdroptlen = toff + off;
2155 
2156 	/*
2157 	 * Calculate amount of space in receive window,
2158 	 * and then do TCP input processing.
2159 	 * Receive window is amount of space in rcv queue,
2160 	 * but not less than advertised window.
2161 	 */
2162 	{ int win;
2163 
2164 	win = sbspace(&so->so_rcv);
2165 	if (win < 0)
2166 		win = 0;
2167 	tp->rcv_wnd = imax(win, (int)(tp->rcv_adv - tp->rcv_nxt));
2168 	}
2169 
2170 	/* Reset receive buffer auto scaling when not in bulk receive mode. */
2171 	tp->rfbuf_ts = 0;
2172 	tp->rfbuf_cnt = 0;
2173 
2174 	switch (tp->t_state) {
2175 	/*
2176 	 * If the state is SYN_SENT:
2177 	 *	if seg contains an ACK, but not for our SYN, drop the input.
2178 	 *	if seg contains a RST, then drop the connection.
2179 	 *	if seg does not contain SYN, then drop it.
2180 	 * Otherwise this is an acceptable SYN segment
2181 	 *	initialize tp->rcv_nxt and tp->irs
2182 	 *	if seg contains ack then advance tp->snd_una
2183 	 *	if seg contains a ECE and ECN support is enabled, the stream
2184 	 *	    is ECN capable.
2185 	 *	if SYN has been acked change to ESTABLISHED else SYN_RCVD state
2186 	 *	arrange for segment to be acked (eventually)
2187 	 *	continue processing rest of data/controls, beginning with URG
2188 	 */
2189 	case TCPS_SYN_SENT:
2190 		if ((tiflags & TH_ACK) &&
2191 		    (SEQ_LEQ(th->th_ack, tp->iss) ||
2192 		     SEQ_GT(th->th_ack, tp->snd_max)))
2193 			goto dropwithreset;
2194 		if (tiflags & TH_RST) {
2195 			if (tiflags & TH_ACK)
2196 				tp = tcp_drop(tp, ECONNREFUSED);
2197 			goto drop;
2198 		}
2199 		if ((tiflags & TH_SYN) == 0)
2200 			goto drop;
2201 		if (tiflags & TH_ACK) {
2202 			tp->snd_una = th->th_ack;
2203 			if (SEQ_LT(tp->snd_nxt, tp->snd_una))
2204 				tp->snd_nxt = tp->snd_una;
2205 			if (SEQ_LT(tp->snd_high, tp->snd_una))
2206 				tp->snd_high = tp->snd_una;
2207 			TCP_TIMER_DISARM(tp, TCPT_REXMT);
2208 
2209 			if ((tiflags & TH_ECE) && tcp_do_ecn) {
2210 				tp->t_flags |= TF_ECN_PERMIT;
2211 				TCP_STATINC(TCP_STAT_ECN_SHS);
2212 			}
2213 
2214 		}
2215 		tp->irs = th->th_seq;
2216 		tcp_rcvseqinit(tp);
2217 		tp->t_flags |= TF_ACKNOW;
2218 		tcp_mss_from_peer(tp, opti.maxseg);
2219 
2220 		/*
2221 		 * Initialize the initial congestion window.  If we
2222 		 * had to retransmit the SYN, we must initialize cwnd
2223 		 * to 1 segment (i.e. the Loss Window).
2224 		 */
2225 		if (tp->t_flags & TF_SYN_REXMT)
2226 			tp->snd_cwnd = tp->t_peermss;
2227 		else {
2228 			int ss = tcp_init_win;
2229 #ifdef INET
2230 			if (inp != NULL && in_localaddr(inp->inp_faddr))
2231 				ss = tcp_init_win_local;
2232 #endif
2233 #ifdef INET6
2234 			if (in6p != NULL && in6_localaddr(&in6p->in6p_faddr))
2235 				ss = tcp_init_win_local;
2236 #endif
2237 			tp->snd_cwnd = TCP_INITIAL_WINDOW(ss, tp->t_peermss);
2238 		}
2239 
2240 		tcp_rmx_rtt(tp);
2241 		if (tiflags & TH_ACK) {
2242 			TCP_STATINC(TCP_STAT_CONNECTS);
2243 			/*
2244 			 * move tcp_established before soisconnected
2245 			 * because upcall handler can drive tcp_output
2246 			 * functionality.
2247 			 * XXX we might call soisconnected at the end of
2248 			 * all processing
2249 			 */
2250 			tcp_established(tp);
2251 			soisconnected(so);
2252 			/* Do window scaling on this connection? */
2253 			if ((tp->t_flags & (TF_RCVD_SCALE|TF_REQ_SCALE)) ==
2254 			    (TF_RCVD_SCALE|TF_REQ_SCALE)) {
2255 				tp->snd_scale = tp->requested_s_scale;
2256 				tp->rcv_scale = tp->request_r_scale;
2257 			}
2258 			TCP_REASS_LOCK(tp);
2259 			(void) tcp_reass(tp, NULL, NULL, &tlen);
2260 			/*
2261 			 * if we didn't have to retransmit the SYN,
2262 			 * use its rtt as our initial srtt & rtt var.
2263 			 */
2264 			if (tp->t_rtttime)
2265 				tcp_xmit_timer(tp, tcp_now - tp->t_rtttime);
2266 		} else
2267 			tp->t_state = TCPS_SYN_RECEIVED;
2268 
2269 		/*
2270 		 * Advance th->th_seq to correspond to first data byte.
2271 		 * If data, trim to stay within window,
2272 		 * dropping FIN if necessary.
2273 		 */
2274 		th->th_seq++;
2275 		if (tlen > tp->rcv_wnd) {
2276 			todrop = tlen - tp->rcv_wnd;
2277 			m_adj(m, -todrop);
2278 			tlen = tp->rcv_wnd;
2279 			tiflags &= ~TH_FIN;
2280 			tcps = TCP_STAT_GETREF();
2281 			tcps[TCP_STAT_RCVPACKAFTERWIN]++;
2282 			tcps[TCP_STAT_RCVBYTEAFTERWIN] += todrop;
2283 			TCP_STAT_PUTREF();
2284 		}
2285 		tp->snd_wl1 = th->th_seq - 1;
2286 		tp->rcv_up = th->th_seq;
2287 		goto step6;
2288 
2289 	/*
2290 	 * If the state is SYN_RECEIVED:
2291 	 *	If seg contains an ACK, but not for our SYN, drop the input
2292 	 *	and generate an RST.  See page 36, rfc793
2293 	 */
2294 	case TCPS_SYN_RECEIVED:
2295 		if ((tiflags & TH_ACK) &&
2296 		    (SEQ_LEQ(th->th_ack, tp->iss) ||
2297 		     SEQ_GT(th->th_ack, tp->snd_max)))
2298 			goto dropwithreset;
2299 		break;
2300 	}
2301 
2302 	/*
2303 	 * States other than LISTEN or SYN_SENT.
2304 	 * First check timestamp, if present.
2305 	 * Then check that at least some bytes of segment are within
2306 	 * receive window.  If segment begins before rcv_nxt,
2307 	 * drop leading data (and SYN); if nothing left, just ack.
2308 	 *
2309 	 * RFC 1323 PAWS: If we have a timestamp reply on this segment
2310 	 * and it's less than ts_recent, drop it.
2311 	 */
2312 	if (opti.ts_present && (tiflags & TH_RST) == 0 && tp->ts_recent &&
2313 	    TSTMP_LT(opti.ts_val, tp->ts_recent)) {
2314 
2315 		/* Check to see if ts_recent is over 24 days old.  */
2316 		if (tcp_now - tp->ts_recent_age > TCP_PAWS_IDLE) {
2317 			/*
2318 			 * Invalidate ts_recent.  If this segment updates
2319 			 * ts_recent, the age will be reset later and ts_recent
2320 			 * will get a valid value.  If it does not, setting
2321 			 * ts_recent to zero will at least satisfy the
2322 			 * requirement that zero be placed in the timestamp
2323 			 * echo reply when ts_recent isn't valid.  The
2324 			 * age isn't reset until we get a valid ts_recent
2325 			 * because we don't want out-of-order segments to be
2326 			 * dropped when ts_recent is old.
2327 			 */
2328 			tp->ts_recent = 0;
2329 		} else {
2330 			tcps = TCP_STAT_GETREF();
2331 			tcps[TCP_STAT_RCVDUPPACK]++;
2332 			tcps[TCP_STAT_RCVDUPBYTE] += tlen;
2333 			tcps[TCP_STAT_PAWSDROP]++;
2334 			TCP_STAT_PUTREF();
2335 			tcp_new_dsack(tp, th->th_seq, tlen);
2336 			goto dropafterack;
2337 		}
2338 	}
2339 
2340 	todrop = tp->rcv_nxt - th->th_seq;
2341 	dupseg = false;
2342 	if (todrop > 0) {
2343 		if (tiflags & TH_SYN) {
2344 			tiflags &= ~TH_SYN;
2345 			th->th_seq++;
2346 			if (th->th_urp > 1)
2347 				th->th_urp--;
2348 			else {
2349 				tiflags &= ~TH_URG;
2350 				th->th_urp = 0;
2351 			}
2352 			todrop--;
2353 		}
2354 		if (todrop > tlen ||
2355 		    (todrop == tlen && (tiflags & TH_FIN) == 0)) {
2356 			/*
2357 			 * Any valid FIN or RST must be to the left of the
2358 			 * window.  At this point the FIN or RST must be a
2359 			 * duplicate or out of sequence; drop it.
2360 			 */
2361 			if (tiflags & TH_RST)
2362 				goto drop;
2363 			tiflags &= ~(TH_FIN|TH_RST);
2364 			/*
2365 			 * Send an ACK to resynchronize and drop any data.
2366 			 * But keep on processing for RST or ACK.
2367 			 */
2368 			tp->t_flags |= TF_ACKNOW;
2369 			todrop = tlen;
2370 			dupseg = true;
2371 			tcps = TCP_STAT_GETREF();
2372 			tcps[TCP_STAT_RCVDUPPACK]++;
2373 			tcps[TCP_STAT_RCVDUPBYTE] += todrop;
2374 			TCP_STAT_PUTREF();
2375 		} else if ((tiflags & TH_RST) &&
2376 			   th->th_seq != tp->rcv_nxt) {
2377 			/*
2378 			 * Test for reset before adjusting the sequence
2379 			 * number for overlapping data.
2380 			 */
2381 			goto dropafterack_ratelim;
2382 		} else {
2383 			tcps = TCP_STAT_GETREF();
2384 			tcps[TCP_STAT_RCVPARTDUPPACK]++;
2385 			tcps[TCP_STAT_RCVPARTDUPBYTE] += todrop;
2386 			TCP_STAT_PUTREF();
2387 		}
2388 		tcp_new_dsack(tp, th->th_seq, todrop);
2389 		hdroptlen += todrop;	/*drop from head afterwards*/
2390 		th->th_seq += todrop;
2391 		tlen -= todrop;
2392 		if (th->th_urp > todrop)
2393 			th->th_urp -= todrop;
2394 		else {
2395 			tiflags &= ~TH_URG;
2396 			th->th_urp = 0;
2397 		}
2398 	}
2399 
2400 	/*
2401 	 * If new data are received on a connection after the
2402 	 * user processes are gone, then RST the other end.
2403 	 */
2404 	if ((so->so_state & SS_NOFDREF) &&
2405 	    tp->t_state > TCPS_CLOSE_WAIT && tlen) {
2406 		tp = tcp_close(tp);
2407 		TCP_STATINC(TCP_STAT_RCVAFTERCLOSE);
2408 		goto dropwithreset;
2409 	}
2410 
2411 	/*
2412 	 * If segment ends after window, drop trailing data
2413 	 * (and PUSH and FIN); if nothing left, just ACK.
2414 	 */
2415 	todrop = (th->th_seq + tlen) - (tp->rcv_nxt+tp->rcv_wnd);
2416 	if (todrop > 0) {
2417 		TCP_STATINC(TCP_STAT_RCVPACKAFTERWIN);
2418 		if (todrop >= tlen) {
2419 			/*
2420 			 * The segment actually starts after the window.
2421 			 * th->th_seq + tlen - tp->rcv_nxt - tp->rcv_wnd >= tlen
2422 			 * th->th_seq - tp->rcv_nxt - tp->rcv_wnd >= 0
2423 			 * th->th_seq >= tp->rcv_nxt + tp->rcv_wnd
2424 			 */
2425 			TCP_STATADD(TCP_STAT_RCVBYTEAFTERWIN, tlen);
2426 			/*
2427 			 * If a new connection request is received
2428 			 * while in TIME_WAIT, drop the old connection
2429 			 * and start over if the sequence numbers
2430 			 * are above the previous ones.
2431 			 *
2432 			 * NOTE: We will checksum the packet again, and
2433 			 * so we need to put the header fields back into
2434 			 * network order!
2435 			 * XXX This kind of sucks, but we don't expect
2436 			 * XXX this to happen very often, so maybe it
2437 			 * XXX doesn't matter so much.
2438 			 */
2439 			if (tiflags & TH_SYN &&
2440 			    tp->t_state == TCPS_TIME_WAIT &&
2441 			    SEQ_GT(th->th_seq, tp->rcv_nxt)) {
2442 				tp = tcp_close(tp);
2443 				tcp_fields_to_net(th);
2444 				goto findpcb;
2445 			}
2446 			/*
2447 			 * If window is closed can only take segments at
2448 			 * window edge, and have to drop data and PUSH from
2449 			 * incoming segments.  Continue processing, but
2450 			 * remember to ack.  Otherwise, drop segment
2451 			 * and (if not RST) ack.
2452 			 */
2453 			if (tp->rcv_wnd == 0 && th->th_seq == tp->rcv_nxt) {
2454 				tp->t_flags |= TF_ACKNOW;
2455 				TCP_STATINC(TCP_STAT_RCVWINPROBE);
2456 			} else
2457 				goto dropafterack;
2458 		} else
2459 			TCP_STATADD(TCP_STAT_RCVBYTEAFTERWIN, todrop);
2460 		m_adj(m, -todrop);
2461 		tlen -= todrop;
2462 		tiflags &= ~(TH_PUSH|TH_FIN);
2463 	}
2464 
2465 	/*
2466 	 * If last ACK falls within this segment's sequence numbers,
2467 	 *  record the timestamp.
2468 	 * NOTE:
2469 	 * 1) That the test incorporates suggestions from the latest
2470 	 *    proposal of the tcplw@cray.com list (Braden 1993/04/26).
2471 	 * 2) That updating only on newer timestamps interferes with
2472 	 *    our earlier PAWS tests, so this check should be solely
2473 	 *    predicated on the sequence space of this segment.
2474 	 * 3) That we modify the segment boundary check to be
2475 	 *        Last.ACK.Sent <= SEG.SEQ + SEG.Len
2476 	 *    instead of RFC1323's
2477 	 *        Last.ACK.Sent < SEG.SEQ + SEG.Len,
2478 	 *    This modified check allows us to overcome RFC1323's
2479 	 *    limitations as described in Stevens TCP/IP Illustrated
2480 	 *    Vol. 2 p.869. In such cases, we can still calculate the
2481 	 *    RTT correctly when RCV.NXT == Last.ACK.Sent.
2482 	 */
2483 	if (opti.ts_present &&
2484 	    SEQ_LEQ(th->th_seq, tp->last_ack_sent) &&
2485 	    SEQ_LEQ(tp->last_ack_sent, th->th_seq + tlen +
2486 		    ((tiflags & (TH_SYN|TH_FIN)) != 0))) {
2487 		tp->ts_recent_age = tcp_now;
2488 		tp->ts_recent = opti.ts_val;
2489 	}
2490 
2491 	/*
2492 	 * If the RST bit is set examine the state:
2493 	 *    SYN_RECEIVED STATE:
2494 	 *	If passive open, return to LISTEN state.
2495 	 *	If active open, inform user that connection was refused.
2496 	 *    ESTABLISHED, FIN_WAIT_1, FIN_WAIT2, CLOSE_WAIT STATES:
2497 	 *	Inform user that connection was reset, and close tcb.
2498 	 *    CLOSING, LAST_ACK, TIME_WAIT STATES
2499 	 *	Close the tcb.
2500 	 */
2501 	if (tiflags & TH_RST) {
2502 		if (th->th_seq != tp->rcv_nxt)
2503 			goto dropafterack_ratelim;
2504 
2505 		switch (tp->t_state) {
2506 		case TCPS_SYN_RECEIVED:
2507 			so->so_error = ECONNREFUSED;
2508 			goto close;
2509 
2510 		case TCPS_ESTABLISHED:
2511 		case TCPS_FIN_WAIT_1:
2512 		case TCPS_FIN_WAIT_2:
2513 		case TCPS_CLOSE_WAIT:
2514 			so->so_error = ECONNRESET;
2515 		close:
2516 			tp->t_state = TCPS_CLOSED;
2517 			TCP_STATINC(TCP_STAT_DROPS);
2518 			tp = tcp_close(tp);
2519 			goto drop;
2520 
2521 		case TCPS_CLOSING:
2522 		case TCPS_LAST_ACK:
2523 		case TCPS_TIME_WAIT:
2524 			tp = tcp_close(tp);
2525 			goto drop;
2526 		}
2527 	}
2528 
2529 	/*
2530 	 * Since we've covered the SYN-SENT and SYN-RECEIVED states above
2531 	 * we must be in a synchronized state.  RFC791 states (under RST
2532 	 * generation) that any unacceptable segment (an out-of-order SYN
2533 	 * qualifies) received in a synchronized state must elicit only an
2534 	 * empty acknowledgment segment ... and the connection remains in
2535 	 * the same state.
2536 	 */
2537 	if (tiflags & TH_SYN) {
2538 		if (tp->rcv_nxt == th->th_seq) {
2539 			tcp_respond(tp, m, m, th, (tcp_seq)0, th->th_ack - 1,
2540 			    TH_ACK);
2541 			if (tcp_saveti)
2542 				m_freem(tcp_saveti);
2543 			return;
2544 		}
2545 
2546 		goto dropafterack_ratelim;
2547 	}
2548 
2549 	/*
2550 	 * If the ACK bit is off we drop the segment and return.
2551 	 */
2552 	if ((tiflags & TH_ACK) == 0) {
2553 		if (tp->t_flags & TF_ACKNOW)
2554 			goto dropafterack;
2555 		else
2556 			goto drop;
2557 	}
2558 
2559 	/*
2560 	 * Ack processing.
2561 	 */
2562 	switch (tp->t_state) {
2563 
2564 	/*
2565 	 * In SYN_RECEIVED state if the ack ACKs our SYN then enter
2566 	 * ESTABLISHED state and continue processing, otherwise
2567 	 * send an RST.
2568 	 */
2569 	case TCPS_SYN_RECEIVED:
2570 		if (SEQ_GT(tp->snd_una, th->th_ack) ||
2571 		    SEQ_GT(th->th_ack, tp->snd_max))
2572 			goto dropwithreset;
2573 		TCP_STATINC(TCP_STAT_CONNECTS);
2574 		soisconnected(so);
2575 		tcp_established(tp);
2576 		/* Do window scaling? */
2577 		if ((tp->t_flags & (TF_RCVD_SCALE|TF_REQ_SCALE)) ==
2578 		    (TF_RCVD_SCALE|TF_REQ_SCALE)) {
2579 			tp->snd_scale = tp->requested_s_scale;
2580 			tp->rcv_scale = tp->request_r_scale;
2581 		}
2582 		TCP_REASS_LOCK(tp);
2583 		(void) tcp_reass(tp, NULL, NULL, &tlen);
2584 		tp->snd_wl1 = th->th_seq - 1;
2585 		/* fall into ... */
2586 
2587 	/*
2588 	 * In ESTABLISHED state: drop duplicate ACKs; ACK out of range
2589 	 * ACKs.  If the ack is in the range
2590 	 *	tp->snd_una < th->th_ack <= tp->snd_max
2591 	 * then advance tp->snd_una to th->th_ack and drop
2592 	 * data from the retransmission queue.  If this ACK reflects
2593 	 * more up to date window information we update our window information.
2594 	 */
2595 	case TCPS_ESTABLISHED:
2596 	case TCPS_FIN_WAIT_1:
2597 	case TCPS_FIN_WAIT_2:
2598 	case TCPS_CLOSE_WAIT:
2599 	case TCPS_CLOSING:
2600 	case TCPS_LAST_ACK:
2601 	case TCPS_TIME_WAIT:
2602 
2603 		if (SEQ_LEQ(th->th_ack, tp->snd_una)) {
2604 			if (tlen == 0 && !dupseg && tiwin == tp->snd_wnd) {
2605 				TCP_STATINC(TCP_STAT_RCVDUPACK);
2606 				/*
2607 				 * If we have outstanding data (other than
2608 				 * a window probe), this is a completely
2609 				 * duplicate ack (ie, window info didn't
2610 				 * change), the ack is the biggest we've
2611 				 * seen and we've seen exactly our rexmt
2612 				 * threshhold of them, assume a packet
2613 				 * has been dropped and retransmit it.
2614 				 * Kludge snd_nxt & the congestion
2615 				 * window so we send only this one
2616 				 * packet.
2617 				 */
2618 				if (TCP_TIMER_ISARMED(tp, TCPT_REXMT) == 0 ||
2619 				    th->th_ack != tp->snd_una)
2620 					tp->t_dupacks = 0;
2621 				else if (tp->t_partialacks < 0 &&
2622 					 (++tp->t_dupacks == tcprexmtthresh ||
2623 					 TCP_FACK_FASTRECOV(tp))) {
2624 					/*
2625 					 * Do the fast retransmit, and adjust
2626 					 * congestion control paramenters.
2627 					 */
2628 					if (tp->t_congctl->fast_retransmit(tp, th)) {
2629 						/* False fast retransmit */
2630 						break;
2631 					} else
2632 						goto drop;
2633 				} else if (tp->t_dupacks > tcprexmtthresh) {
2634 					tp->snd_cwnd += tp->t_segsz;
2635 					KERNEL_LOCK(1, NULL);
2636 					(void) tcp_output(tp);
2637 					KERNEL_UNLOCK_ONE(NULL);
2638 					goto drop;
2639 				}
2640 			} else {
2641 				/*
2642 				 * If the ack appears to be very old, only
2643 				 * allow data that is in-sequence.  This
2644 				 * makes it somewhat more difficult to insert
2645 				 * forged data by guessing sequence numbers.
2646 				 * Sent an ack to try to update the send
2647 				 * sequence number on the other side.
2648 				 */
2649 				if (tlen && th->th_seq != tp->rcv_nxt &&
2650 				    SEQ_LT(th->th_ack,
2651 				    tp->snd_una - tp->max_sndwnd))
2652 					goto dropafterack;
2653 			}
2654 			break;
2655 		}
2656 		/*
2657 		 * If the congestion window was inflated to account
2658 		 * for the other side's cached packets, retract it.
2659 		 */
2660 		/* XXX: make SACK have his own congestion control
2661 		 * struct -- rpaulo */
2662 		if (TCP_SACK_ENABLED(tp))
2663 			tcp_sack_newack(tp, th);
2664 		else
2665 			tp->t_congctl->fast_retransmit_newack(tp, th);
2666 		if (SEQ_GT(th->th_ack, tp->snd_max)) {
2667 			TCP_STATINC(TCP_STAT_RCVACKTOOMUCH);
2668 			goto dropafterack;
2669 		}
2670 		acked = th->th_ack - tp->snd_una;
2671 		tcps = TCP_STAT_GETREF();
2672 		tcps[TCP_STAT_RCVACKPACK]++;
2673 		tcps[TCP_STAT_RCVACKBYTE] += acked;
2674 		TCP_STAT_PUTREF();
2675 
2676 		/*
2677 		 * If we have a timestamp reply, update smoothed
2678 		 * round trip time.  If no timestamp is present but
2679 		 * transmit timer is running and timed sequence
2680 		 * number was acked, update smoothed round trip time.
2681 		 * Since we now have an rtt measurement, cancel the
2682 		 * timer backoff (cf., Phil Karn's retransmit alg.).
2683 		 * Recompute the initial retransmit timer.
2684 		 */
2685 		if (ts_rtt)
2686 			tcp_xmit_timer(tp, ts_rtt - 1);
2687 		else if (tp->t_rtttime && SEQ_GT(th->th_ack, tp->t_rtseq))
2688 			tcp_xmit_timer(tp, tcp_now - tp->t_rtttime);
2689 
2690 		/*
2691 		 * If all outstanding data is acked, stop retransmit
2692 		 * timer and remember to restart (more output or persist).
2693 		 * If there is more data to be acked, restart retransmit
2694 		 * timer, using current (possibly backed-off) value.
2695 		 */
2696 		if (th->th_ack == tp->snd_max) {
2697 			TCP_TIMER_DISARM(tp, TCPT_REXMT);
2698 			needoutput = 1;
2699 		} else if (TCP_TIMER_ISARMED(tp, TCPT_PERSIST) == 0)
2700 			TCP_TIMER_ARM(tp, TCPT_REXMT, tp->t_rxtcur);
2701 
2702 		/*
2703 		 * New data has been acked, adjust the congestion window.
2704 		 */
2705 		tp->t_congctl->newack(tp, th);
2706 
2707 		nd6_hint(tp);
2708 		if (acked > so->so_snd.sb_cc) {
2709 			tp->snd_wnd -= so->so_snd.sb_cc;
2710 			sbdrop(&so->so_snd, (int)so->so_snd.sb_cc);
2711 			ourfinisacked = 1;
2712 		} else {
2713 			if (acked > (tp->t_lastoff - tp->t_inoff))
2714 				tp->t_lastm = NULL;
2715 			sbdrop(&so->so_snd, acked);
2716 			tp->t_lastoff -= acked;
2717 			tp->snd_wnd -= acked;
2718 			ourfinisacked = 0;
2719 		}
2720 		sowwakeup(so);
2721 
2722 		icmp_check(tp, th, acked);
2723 
2724 		tp->snd_una = th->th_ack;
2725 		if (SEQ_GT(tp->snd_una, tp->snd_fack))
2726 			tp->snd_fack = tp->snd_una;
2727 		if (SEQ_LT(tp->snd_nxt, tp->snd_una))
2728 			tp->snd_nxt = tp->snd_una;
2729 		if (SEQ_LT(tp->snd_high, tp->snd_una))
2730 			tp->snd_high = tp->snd_una;
2731 
2732 		switch (tp->t_state) {
2733 
2734 		/*
2735 		 * In FIN_WAIT_1 STATE in addition to the processing
2736 		 * for the ESTABLISHED state if our FIN is now acknowledged
2737 		 * then enter FIN_WAIT_2.
2738 		 */
2739 		case TCPS_FIN_WAIT_1:
2740 			if (ourfinisacked) {
2741 				/*
2742 				 * If we can't receive any more
2743 				 * data, then closing user can proceed.
2744 				 * Starting the timer is contrary to the
2745 				 * specification, but if we don't get a FIN
2746 				 * we'll hang forever.
2747 				 */
2748 				if (so->so_state & SS_CANTRCVMORE) {
2749 					soisdisconnected(so);
2750 					if (tp->t_maxidle > 0)
2751 						TCP_TIMER_ARM(tp, TCPT_2MSL,
2752 						    tp->t_maxidle);
2753 				}
2754 				tp->t_state = TCPS_FIN_WAIT_2;
2755 			}
2756 			break;
2757 
2758 	 	/*
2759 		 * In CLOSING STATE in addition to the processing for
2760 		 * the ESTABLISHED state if the ACK acknowledges our FIN
2761 		 * then enter the TIME-WAIT state, otherwise ignore
2762 		 * the segment.
2763 		 */
2764 		case TCPS_CLOSING:
2765 			if (ourfinisacked) {
2766 				tp->t_state = TCPS_TIME_WAIT;
2767 				tcp_canceltimers(tp);
2768 				TCP_TIMER_ARM(tp, TCPT_2MSL, 2 * tp->t_msl);
2769 				soisdisconnected(so);
2770 			}
2771 			break;
2772 
2773 		/*
2774 		 * In LAST_ACK, we may still be waiting for data to drain
2775 		 * and/or to be acked, as well as for the ack of our FIN.
2776 		 * If our FIN is now acknowledged, delete the TCB,
2777 		 * enter the closed state and return.
2778 		 */
2779 		case TCPS_LAST_ACK:
2780 			if (ourfinisacked) {
2781 				tp = tcp_close(tp);
2782 				goto drop;
2783 			}
2784 			break;
2785 
2786 		/*
2787 		 * In TIME_WAIT state the only thing that should arrive
2788 		 * is a retransmission of the remote FIN.  Acknowledge
2789 		 * it and restart the finack timer.
2790 		 */
2791 		case TCPS_TIME_WAIT:
2792 			TCP_TIMER_ARM(tp, TCPT_2MSL, 2 * tp->t_msl);
2793 			goto dropafterack;
2794 		}
2795 	}
2796 
2797 step6:
2798 	/*
2799 	 * Update window information.
2800 	 * Don't look at window if no ACK: TAC's send garbage on first SYN.
2801 	 */
2802 	if ((tiflags & TH_ACK) && (SEQ_LT(tp->snd_wl1, th->th_seq) ||
2803 	    (tp->snd_wl1 == th->th_seq && (SEQ_LT(tp->snd_wl2, th->th_ack) ||
2804 	    (tp->snd_wl2 == th->th_ack && tiwin > tp->snd_wnd))))) {
2805 		/* keep track of pure window updates */
2806 		if (tlen == 0 &&
2807 		    tp->snd_wl2 == th->th_ack && tiwin > tp->snd_wnd)
2808 			TCP_STATINC(TCP_STAT_RCVWINUPD);
2809 		tp->snd_wnd = tiwin;
2810 		tp->snd_wl1 = th->th_seq;
2811 		tp->snd_wl2 = th->th_ack;
2812 		if (tp->snd_wnd > tp->max_sndwnd)
2813 			tp->max_sndwnd = tp->snd_wnd;
2814 		needoutput = 1;
2815 	}
2816 
2817 	/*
2818 	 * Process segments with URG.
2819 	 */
2820 	if ((tiflags & TH_URG) && th->th_urp &&
2821 	    TCPS_HAVERCVDFIN(tp->t_state) == 0) {
2822 		/*
2823 		 * This is a kludge, but if we receive and accept
2824 		 * random urgent pointers, we'll crash in
2825 		 * soreceive.  It's hard to imagine someone
2826 		 * actually wanting to send this much urgent data.
2827 		 */
2828 		if (th->th_urp + so->so_rcv.sb_cc > sb_max) {
2829 			th->th_urp = 0;			/* XXX */
2830 			tiflags &= ~TH_URG;		/* XXX */
2831 			goto dodata;			/* XXX */
2832 		}
2833 		/*
2834 		 * If this segment advances the known urgent pointer,
2835 		 * then mark the data stream.  This should not happen
2836 		 * in CLOSE_WAIT, CLOSING, LAST_ACK or TIME_WAIT STATES since
2837 		 * a FIN has been received from the remote side.
2838 		 * In these states we ignore the URG.
2839 		 *
2840 		 * According to RFC961 (Assigned Protocols),
2841 		 * the urgent pointer points to the last octet
2842 		 * of urgent data.  We continue, however,
2843 		 * to consider it to indicate the first octet
2844 		 * of data past the urgent section as the original
2845 		 * spec states (in one of two places).
2846 		 */
2847 		if (SEQ_GT(th->th_seq+th->th_urp, tp->rcv_up)) {
2848 			tp->rcv_up = th->th_seq + th->th_urp;
2849 			so->so_oobmark = so->so_rcv.sb_cc +
2850 			    (tp->rcv_up - tp->rcv_nxt) - 1;
2851 			if (so->so_oobmark == 0)
2852 				so->so_state |= SS_RCVATMARK;
2853 			sohasoutofband(so);
2854 			tp->t_oobflags &= ~(TCPOOB_HAVEDATA | TCPOOB_HADDATA);
2855 		}
2856 		/*
2857 		 * Remove out of band data so doesn't get presented to user.
2858 		 * This can happen independent of advancing the URG pointer,
2859 		 * but if two URG's are pending at once, some out-of-band
2860 		 * data may creep in... ick.
2861 		 */
2862 		if (th->th_urp <= (u_int16_t) tlen
2863 #ifdef SO_OOBINLINE
2864 		     && (so->so_options & SO_OOBINLINE) == 0
2865 #endif
2866 		     )
2867 			tcp_pulloutofband(so, th, m, hdroptlen);
2868 	} else
2869 		/*
2870 		 * If no out of band data is expected,
2871 		 * pull receive urgent pointer along
2872 		 * with the receive window.
2873 		 */
2874 		if (SEQ_GT(tp->rcv_nxt, tp->rcv_up))
2875 			tp->rcv_up = tp->rcv_nxt;
2876 dodata:							/* XXX */
2877 
2878 	/*
2879 	 * Process the segment text, merging it into the TCP sequencing queue,
2880 	 * and arranging for acknowledgement of receipt if necessary.
2881 	 * This process logically involves adjusting tp->rcv_wnd as data
2882 	 * is presented to the user (this happens in tcp_usrreq.c,
2883 	 * case PRU_RCVD).  If a FIN has already been received on this
2884 	 * connection then we just ignore the text.
2885 	 */
2886 	if ((tlen || (tiflags & TH_FIN)) &&
2887 	    TCPS_HAVERCVDFIN(tp->t_state) == 0) {
2888 		/*
2889 		 * Insert segment ti into reassembly queue of tcp with
2890 		 * control block tp.  Return TH_FIN if reassembly now includes
2891 		 * a segment with FIN.  The macro form does the common case
2892 		 * inline (segment is the next to be received on an
2893 		 * established connection, and the queue is empty),
2894 		 * avoiding linkage into and removal from the queue and
2895 		 * repetition of various conversions.
2896 		 * Set DELACK for segments received in order, but ack
2897 		 * immediately when segments are out of order
2898 		 * (so fast retransmit can work).
2899 		 */
2900 		/* NOTE: this was TCP_REASS() macro, but used only once */
2901 		TCP_REASS_LOCK(tp);
2902 		if (th->th_seq == tp->rcv_nxt &&
2903 		    TAILQ_FIRST(&tp->segq) == NULL &&
2904 		    tp->t_state == TCPS_ESTABLISHED) {
2905 			tcp_setup_ack(tp, th);
2906 			tp->rcv_nxt += tlen;
2907 			tiflags = th->th_flags & TH_FIN;
2908 			tcps = TCP_STAT_GETREF();
2909 			tcps[TCP_STAT_RCVPACK]++;
2910 			tcps[TCP_STAT_RCVBYTE] += tlen;
2911 			TCP_STAT_PUTREF();
2912 			nd6_hint(tp);
2913 			if (so->so_state & SS_CANTRCVMORE)
2914 				m_freem(m);
2915 			else {
2916 				m_adj(m, hdroptlen);
2917 				sbappendstream(&(so)->so_rcv, m);
2918 			}
2919 			TCP_REASS_UNLOCK(tp);
2920 			sorwakeup(so);
2921 		} else {
2922 			m_adj(m, hdroptlen);
2923 			tiflags = tcp_reass(tp, th, m, &tlen);
2924 			tp->t_flags |= TF_ACKNOW;
2925 		}
2926 
2927 		/*
2928 		 * Note the amount of data that peer has sent into
2929 		 * our window, in order to estimate the sender's
2930 		 * buffer size.
2931 		 */
2932 		len = so->so_rcv.sb_hiwat - (tp->rcv_adv - tp->rcv_nxt);
2933 	} else {
2934 		m_freem(m);
2935 		m = NULL;
2936 		tiflags &= ~TH_FIN;
2937 	}
2938 
2939 	/*
2940 	 * If FIN is received ACK the FIN and let the user know
2941 	 * that the connection is closing.  Ignore a FIN received before
2942 	 * the connection is fully established.
2943 	 */
2944 	if ((tiflags & TH_FIN) && TCPS_HAVEESTABLISHED(tp->t_state)) {
2945 		if (TCPS_HAVERCVDFIN(tp->t_state) == 0) {
2946 			socantrcvmore(so);
2947 			tp->t_flags |= TF_ACKNOW;
2948 			tp->rcv_nxt++;
2949 		}
2950 		switch (tp->t_state) {
2951 
2952 	 	/*
2953 		 * In ESTABLISHED STATE enter the CLOSE_WAIT state.
2954 		 */
2955 		case TCPS_ESTABLISHED:
2956 			tp->t_state = TCPS_CLOSE_WAIT;
2957 			break;
2958 
2959 	 	/*
2960 		 * If still in FIN_WAIT_1 STATE FIN has not been acked so
2961 		 * enter the CLOSING state.
2962 		 */
2963 		case TCPS_FIN_WAIT_1:
2964 			tp->t_state = TCPS_CLOSING;
2965 			break;
2966 
2967 	 	/*
2968 		 * In FIN_WAIT_2 state enter the TIME_WAIT state,
2969 		 * starting the time-wait timer, turning off the other
2970 		 * standard timers.
2971 		 */
2972 		case TCPS_FIN_WAIT_2:
2973 			tp->t_state = TCPS_TIME_WAIT;
2974 			tcp_canceltimers(tp);
2975 			TCP_TIMER_ARM(tp, TCPT_2MSL, 2 * tp->t_msl);
2976 			soisdisconnected(so);
2977 			break;
2978 
2979 		/*
2980 		 * In TIME_WAIT state restart the 2 MSL time_wait timer.
2981 		 */
2982 		case TCPS_TIME_WAIT:
2983 			TCP_TIMER_ARM(tp, TCPT_2MSL, 2 * tp->t_msl);
2984 			break;
2985 		}
2986 	}
2987 #ifdef TCP_DEBUG
2988 	if (so->so_options & SO_DEBUG)
2989 		tcp_trace(TA_INPUT, ostate, tp, tcp_saveti, 0);
2990 #endif
2991 
2992 	/*
2993 	 * Return any desired output.
2994 	 */
2995 	if (needoutput || (tp->t_flags & TF_ACKNOW)) {
2996 		KERNEL_LOCK(1, NULL);
2997 		(void) tcp_output(tp);
2998 		KERNEL_UNLOCK_ONE(NULL);
2999 	}
3000 	if (tcp_saveti)
3001 		m_freem(tcp_saveti);
3002 
3003 	if (tp->t_state == TCPS_TIME_WAIT
3004 	    && (so->so_state & SS_NOFDREF)
3005 	    && (tp->t_inpcb || af != AF_INET)
3006 	    && (tp->t_in6pcb || af != AF_INET6)
3007 	    && ((af == AF_INET ? tcp4_vtw_enable : tcp6_vtw_enable) & 1) != 0
3008 	    && TAILQ_EMPTY(&tp->segq)
3009 	    && vtw_add(af, tp)) {
3010 		;
3011 	}
3012 	return;
3013 
3014 badsyn:
3015 	/*
3016 	 * Received a bad SYN.  Increment counters and dropwithreset.
3017 	 */
3018 	TCP_STATINC(TCP_STAT_BADSYN);
3019 	tp = NULL;
3020 	goto dropwithreset;
3021 
3022 dropafterack:
3023 	/*
3024 	 * Generate an ACK dropping incoming segment if it occupies
3025 	 * sequence space, where the ACK reflects our state.
3026 	 */
3027 	if (tiflags & TH_RST)
3028 		goto drop;
3029 	goto dropafterack2;
3030 
3031 dropafterack_ratelim:
3032 	/*
3033 	 * We may want to rate-limit ACKs against SYN/RST attack.
3034 	 */
3035 	if (ppsratecheck(&tcp_ackdrop_ppslim_last, &tcp_ackdrop_ppslim_count,
3036 	    tcp_ackdrop_ppslim) == 0) {
3037 		/* XXX stat */
3038 		goto drop;
3039 	}
3040 	/* ...fall into dropafterack2... */
3041 
3042 dropafterack2:
3043 	m_freem(m);
3044 	tp->t_flags |= TF_ACKNOW;
3045 	KERNEL_LOCK(1, NULL);
3046 	(void) tcp_output(tp);
3047 	KERNEL_UNLOCK_ONE(NULL);
3048 	if (tcp_saveti)
3049 		m_freem(tcp_saveti);
3050 	return;
3051 
3052 dropwithreset_ratelim:
3053 	/*
3054 	 * We may want to rate-limit RSTs in certain situations,
3055 	 * particularly if we are sending an RST in response to
3056 	 * an attempt to connect to or otherwise communicate with
3057 	 * a port for which we have no socket.
3058 	 */
3059 	if (ppsratecheck(&tcp_rst_ppslim_last, &tcp_rst_ppslim_count,
3060 	    tcp_rst_ppslim) == 0) {
3061 		/* XXX stat */
3062 		goto drop;
3063 	}
3064 	/* ...fall into dropwithreset... */
3065 
3066 dropwithreset:
3067 	/*
3068 	 * Generate a RST, dropping incoming segment.
3069 	 * Make ACK acceptable to originator of segment.
3070 	 */
3071 	if (tiflags & TH_RST)
3072 		goto drop;
3073 
3074 	switch (af) {
3075 #ifdef INET6
3076 	case AF_INET6:
3077 		/* For following calls to tcp_respond */
3078 		if (IN6_IS_ADDR_MULTICAST(&ip6->ip6_dst))
3079 			goto drop;
3080 		break;
3081 #endif /* INET6 */
3082 	case AF_INET:
3083 		if (IN_MULTICAST(ip->ip_dst.s_addr) ||
3084 		    in_broadcast(ip->ip_dst, m->m_pkthdr.rcvif))
3085 			goto drop;
3086 	}
3087 
3088 	if (tiflags & TH_ACK)
3089 		(void)tcp_respond(tp, m, m, th, (tcp_seq)0, th->th_ack, TH_RST);
3090 	else {
3091 		if (tiflags & TH_SYN)
3092 			tlen++;
3093 		(void)tcp_respond(tp, m, m, th, th->th_seq + tlen, (tcp_seq)0,
3094 		    TH_RST|TH_ACK);
3095 	}
3096 	if (tcp_saveti)
3097 		m_freem(tcp_saveti);
3098 	return;
3099 
3100 badcsum:
3101 drop:
3102 	/*
3103 	 * Drop space held by incoming segment and return.
3104 	 */
3105 	if (tp) {
3106 		if (tp->t_inpcb)
3107 			so = tp->t_inpcb->inp_socket;
3108 #ifdef INET6
3109 		else if (tp->t_in6pcb)
3110 			so = tp->t_in6pcb->in6p_socket;
3111 #endif
3112 		else
3113 			so = NULL;
3114 #ifdef TCP_DEBUG
3115 		if (so && (so->so_options & SO_DEBUG) != 0)
3116 			tcp_trace(TA_DROP, ostate, tp, tcp_saveti, 0);
3117 #endif
3118 	}
3119 	if (tcp_saveti)
3120 		m_freem(tcp_saveti);
3121 	m_freem(m);
3122 	return;
3123 }
3124 
3125 #ifdef TCP_SIGNATURE
3126 int
3127 tcp_signature_apply(void *fstate, void *data, u_int len)
3128 {
3129 
3130 	MD5Update(fstate, (u_char *)data, len);
3131 	return (0);
3132 }
3133 
3134 struct secasvar *
3135 tcp_signature_getsav(struct mbuf *m, struct tcphdr *th)
3136 {
3137 	struct secasvar *sav;
3138 #ifdef IPSEC
3139 	union sockaddr_union dst;
3140 #endif
3141 	struct ip *ip;
3142 	struct ip6_hdr *ip6;
3143 
3144 	ip = mtod(m, struct ip *);
3145 	switch (ip->ip_v) {
3146 	case 4:
3147 		ip = mtod(m, struct ip *);
3148 		ip6 = NULL;
3149 		break;
3150 	case 6:
3151 		ip = NULL;
3152 		ip6 = mtod(m, struct ip6_hdr *);
3153 		break;
3154 	default:
3155 		return (NULL);
3156 	}
3157 
3158 #ifdef IPSEC
3159 	/* Extract the destination from the IP header in the mbuf. */
3160 	memset(&dst, 0, sizeof(union sockaddr_union));
3161 	if (ip !=NULL) {
3162 		dst.sa.sa_len = sizeof(struct sockaddr_in);
3163 		dst.sa.sa_family = AF_INET;
3164 		dst.sin.sin_addr = ip->ip_dst;
3165 	} else {
3166 		dst.sa.sa_len = sizeof(struct sockaddr_in6);
3167 		dst.sa.sa_family = AF_INET6;
3168 		dst.sin6.sin6_addr = ip6->ip6_dst;
3169 	}
3170 
3171 	/*
3172 	 * Look up an SADB entry which matches the address of the peer.
3173 	 */
3174 	sav = KEY_ALLOCSA(&dst, IPPROTO_TCP, htonl(TCP_SIG_SPI), 0, 0);
3175 #else
3176 	if (ip)
3177 		sav = key_allocsa(AF_INET, (void *)&ip->ip_src,
3178 		    (void *)&ip->ip_dst, IPPROTO_TCP,
3179 		    htonl(TCP_SIG_SPI), 0, 0);
3180 	else
3181 		sav = key_allocsa(AF_INET6, (void *)&ip6->ip6_src,
3182 		    (void *)&ip6->ip6_dst, IPPROTO_TCP,
3183 		    htonl(TCP_SIG_SPI), 0, 0);
3184 #endif
3185 
3186 	return (sav);	/* freesav must be performed by caller */
3187 }
3188 
3189 int
3190 tcp_signature(struct mbuf *m, struct tcphdr *th, int thoff,
3191     struct secasvar *sav, char *sig)
3192 {
3193 	MD5_CTX ctx;
3194 	struct ip *ip;
3195 	struct ipovly *ipovly;
3196 	struct ip6_hdr *ip6;
3197 	struct ippseudo ippseudo;
3198 	struct ip6_hdr_pseudo ip6pseudo;
3199 	struct tcphdr th0;
3200 	int l, tcphdrlen;
3201 
3202 	if (sav == NULL)
3203 		return (-1);
3204 
3205 	tcphdrlen = th->th_off * 4;
3206 
3207 	switch (mtod(m, struct ip *)->ip_v) {
3208 	case 4:
3209 		ip = mtod(m, struct ip *);
3210 		ip6 = NULL;
3211 		break;
3212 	case 6:
3213 		ip = NULL;
3214 		ip6 = mtod(m, struct ip6_hdr *);
3215 		break;
3216 	default:
3217 		return (-1);
3218 	}
3219 
3220 	MD5Init(&ctx);
3221 
3222 	if (ip) {
3223 		memset(&ippseudo, 0, sizeof(ippseudo));
3224 		ipovly = (struct ipovly *)ip;
3225 		ippseudo.ippseudo_src = ipovly->ih_src;
3226 		ippseudo.ippseudo_dst = ipovly->ih_dst;
3227 		ippseudo.ippseudo_pad = 0;
3228 		ippseudo.ippseudo_p = IPPROTO_TCP;
3229 		ippseudo.ippseudo_len = htons(m->m_pkthdr.len - thoff);
3230 		MD5Update(&ctx, (char *)&ippseudo, sizeof(ippseudo));
3231 	} else {
3232 		memset(&ip6pseudo, 0, sizeof(ip6pseudo));
3233 		ip6pseudo.ip6ph_src = ip6->ip6_src;
3234 		in6_clearscope(&ip6pseudo.ip6ph_src);
3235 		ip6pseudo.ip6ph_dst = ip6->ip6_dst;
3236 		in6_clearscope(&ip6pseudo.ip6ph_dst);
3237 		ip6pseudo.ip6ph_len = htons(m->m_pkthdr.len - thoff);
3238 		ip6pseudo.ip6ph_nxt = IPPROTO_TCP;
3239 		MD5Update(&ctx, (char *)&ip6pseudo, sizeof(ip6pseudo));
3240 	}
3241 
3242 	th0 = *th;
3243 	th0.th_sum = 0;
3244 	MD5Update(&ctx, (char *)&th0, sizeof(th0));
3245 
3246 	l = m->m_pkthdr.len - thoff - tcphdrlen;
3247 	if (l > 0)
3248 		m_apply(m, thoff + tcphdrlen,
3249 		    m->m_pkthdr.len - thoff - tcphdrlen,
3250 		    tcp_signature_apply, &ctx);
3251 
3252 	MD5Update(&ctx, _KEYBUF(sav->key_auth), _KEYLEN(sav->key_auth));
3253 	MD5Final(sig, &ctx);
3254 
3255 	return (0);
3256 }
3257 #endif
3258 
3259 /*
3260  * tcp_dooptions: parse and process tcp options.
3261  *
3262  * returns -1 if this segment should be dropped.  (eg. wrong signature)
3263  * otherwise returns 0.
3264  */
3265 
3266 static int
3267 tcp_dooptions(struct tcpcb *tp, const u_char *cp, int cnt,
3268     struct tcphdr *th,
3269     struct mbuf *m, int toff, struct tcp_opt_info *oi)
3270 {
3271 	u_int16_t mss;
3272 	int opt, optlen = 0;
3273 #ifdef TCP_SIGNATURE
3274 	void *sigp = NULL;
3275 	char sigbuf[TCP_SIGLEN];
3276 	struct secasvar *sav = NULL;
3277 #endif
3278 
3279 	for (; cp && cnt > 0; cnt -= optlen, cp += optlen) {
3280 		opt = cp[0];
3281 		if (opt == TCPOPT_EOL)
3282 			break;
3283 		if (opt == TCPOPT_NOP)
3284 			optlen = 1;
3285 		else {
3286 			if (cnt < 2)
3287 				break;
3288 			optlen = cp[1];
3289 			if (optlen < 2 || optlen > cnt)
3290 				break;
3291 		}
3292 		switch (opt) {
3293 
3294 		default:
3295 			continue;
3296 
3297 		case TCPOPT_MAXSEG:
3298 			if (optlen != TCPOLEN_MAXSEG)
3299 				continue;
3300 			if (!(th->th_flags & TH_SYN))
3301 				continue;
3302 			if (TCPS_HAVERCVDSYN(tp->t_state))
3303 				continue;
3304 			bcopy(cp + 2, &mss, sizeof(mss));
3305 			oi->maxseg = ntohs(mss);
3306 			break;
3307 
3308 		case TCPOPT_WINDOW:
3309 			if (optlen != TCPOLEN_WINDOW)
3310 				continue;
3311 			if (!(th->th_flags & TH_SYN))
3312 				continue;
3313 			if (TCPS_HAVERCVDSYN(tp->t_state))
3314 				continue;
3315 			tp->t_flags |= TF_RCVD_SCALE;
3316 			tp->requested_s_scale = cp[2];
3317 			if (tp->requested_s_scale > TCP_MAX_WINSHIFT) {
3318 #if 0	/*XXX*/
3319 				char *p;
3320 
3321 				if (ip)
3322 					p = ntohl(ip->ip_src);
3323 #ifdef INET6
3324 				else if (ip6)
3325 					p = ip6_sprintf(&ip6->ip6_src);
3326 #endif
3327 				else
3328 					p = "(unknown)";
3329 				log(LOG_ERR, "TCP: invalid wscale %d from %s, "
3330 				    "assuming %d\n",
3331 				    tp->requested_s_scale, p,
3332 				    TCP_MAX_WINSHIFT);
3333 #else
3334 				log(LOG_ERR, "TCP: invalid wscale %d, "
3335 				    "assuming %d\n",
3336 				    tp->requested_s_scale,
3337 				    TCP_MAX_WINSHIFT);
3338 #endif
3339 				tp->requested_s_scale = TCP_MAX_WINSHIFT;
3340 			}
3341 			break;
3342 
3343 		case TCPOPT_TIMESTAMP:
3344 			if (optlen != TCPOLEN_TIMESTAMP)
3345 				continue;
3346 			oi->ts_present = 1;
3347 			bcopy(cp + 2, &oi->ts_val, sizeof(oi->ts_val));
3348 			NTOHL(oi->ts_val);
3349 			bcopy(cp + 6, &oi->ts_ecr, sizeof(oi->ts_ecr));
3350 			NTOHL(oi->ts_ecr);
3351 
3352 			if (!(th->th_flags & TH_SYN))
3353 				continue;
3354 			if (TCPS_HAVERCVDSYN(tp->t_state))
3355 				continue;
3356 			/*
3357 			 * A timestamp received in a SYN makes
3358 			 * it ok to send timestamp requests and replies.
3359 			 */
3360 			tp->t_flags |= TF_RCVD_TSTMP;
3361 			tp->ts_recent = oi->ts_val;
3362 			tp->ts_recent_age = tcp_now;
3363                         break;
3364 
3365 		case TCPOPT_SACK_PERMITTED:
3366 			if (optlen != TCPOLEN_SACK_PERMITTED)
3367 				continue;
3368 			if (!(th->th_flags & TH_SYN))
3369 				continue;
3370 			if (TCPS_HAVERCVDSYN(tp->t_state))
3371 				continue;
3372 			if (tcp_do_sack) {
3373 				tp->t_flags |= TF_SACK_PERMIT;
3374 				tp->t_flags |= TF_WILL_SACK;
3375 			}
3376 			break;
3377 
3378 		case TCPOPT_SACK:
3379 			tcp_sack_option(tp, th, cp, optlen);
3380 			break;
3381 #ifdef TCP_SIGNATURE
3382 		case TCPOPT_SIGNATURE:
3383 			if (optlen != TCPOLEN_SIGNATURE)
3384 				continue;
3385 			if (sigp && memcmp(sigp, cp + 2, TCP_SIGLEN))
3386 				return (-1);
3387 
3388 			sigp = sigbuf;
3389 			memcpy(sigbuf, cp + 2, TCP_SIGLEN);
3390 			tp->t_flags |= TF_SIGNATURE;
3391 			break;
3392 #endif
3393 		}
3394 	}
3395 
3396 #ifndef TCP_SIGNATURE
3397 	return 0;
3398 #else
3399 	if (tp->t_flags & TF_SIGNATURE) {
3400 
3401 		sav = tcp_signature_getsav(m, th);
3402 
3403 		if (sav == NULL && tp->t_state == TCPS_LISTEN)
3404 			return (-1);
3405 	}
3406 
3407 	if ((sigp ? TF_SIGNATURE : 0) ^ (tp->t_flags & TF_SIGNATURE))
3408 		goto out;
3409 
3410 	if (sigp) {
3411 		char sig[TCP_SIGLEN];
3412 
3413 		tcp_fields_to_net(th);
3414 		if (tcp_signature(m, th, toff, sav, sig) < 0) {
3415 			tcp_fields_to_host(th);
3416 			goto out;
3417 		}
3418 		tcp_fields_to_host(th);
3419 
3420 		if (memcmp(sig, sigp, TCP_SIGLEN)) {
3421 			TCP_STATINC(TCP_STAT_BADSIG);
3422 			goto out;
3423 		} else
3424 			TCP_STATINC(TCP_STAT_GOODSIG);
3425 
3426 		key_sa_recordxfer(sav, m);
3427 		KEY_FREESAV(&sav);
3428 	}
3429 	return 0;
3430 out:
3431 	if (sav != NULL)
3432 		KEY_FREESAV(&sav);
3433 	return -1;
3434 #endif
3435 }
3436 
3437 /*
3438  * Pull out of band byte out of a segment so
3439  * it doesn't appear in the user's data queue.
3440  * It is still reflected in the segment length for
3441  * sequencing purposes.
3442  */
3443 void
3444 tcp_pulloutofband(struct socket *so, struct tcphdr *th,
3445     struct mbuf *m, int off)
3446 {
3447 	int cnt = off + th->th_urp - 1;
3448 
3449 	while (cnt >= 0) {
3450 		if (m->m_len > cnt) {
3451 			char *cp = mtod(m, char *) + cnt;
3452 			struct tcpcb *tp = sototcpcb(so);
3453 
3454 			tp->t_iobc = *cp;
3455 			tp->t_oobflags |= TCPOOB_HAVEDATA;
3456 			bcopy(cp+1, cp, (unsigned)(m->m_len - cnt - 1));
3457 			m->m_len--;
3458 			return;
3459 		}
3460 		cnt -= m->m_len;
3461 		m = m->m_next;
3462 		if (m == 0)
3463 			break;
3464 	}
3465 	panic("tcp_pulloutofband");
3466 }
3467 
3468 /*
3469  * Collect new round-trip time estimate
3470  * and update averages and current timeout.
3471  *
3472  * rtt is in units of slow ticks (typically 500 ms) -- essentially the
3473  * difference of two timestamps.
3474  */
3475 void
3476 tcp_xmit_timer(struct tcpcb *tp, uint32_t rtt)
3477 {
3478 	int32_t delta;
3479 
3480 	TCP_STATINC(TCP_STAT_RTTUPDATED);
3481 	if (tp->t_srtt != 0) {
3482 		/*
3483 		 * Compute the amount to add to srtt for smoothing,
3484 		 * *alpha, or 2^(-TCP_RTT_SHIFT).  Because
3485 		 * srtt is stored in 1/32 slow ticks, we conceptually
3486 		 * shift left 5 bits, subtract srtt to get the
3487 		 * diference, and then shift right by TCP_RTT_SHIFT
3488 		 * (3) to obtain 1/8 of the difference.
3489 		 */
3490 		delta = (rtt << 2) - (tp->t_srtt >> TCP_RTT_SHIFT);
3491 		/*
3492 		 * This can never happen, because delta's lowest
3493 		 * possible value is 1/8 of t_srtt.  But if it does,
3494 		 * set srtt to some reasonable value, here chosen
3495 		 * as 1/8 tick.
3496 		 */
3497 		if ((tp->t_srtt += delta) <= 0)
3498 			tp->t_srtt = 1 << 2;
3499 		/*
3500 		 * RFC2988 requires that rttvar be updated first.
3501 		 * This code is compliant because "delta" is the old
3502 		 * srtt minus the new observation (scaled).
3503 		 *
3504 		 * RFC2988 says:
3505 		 *   rttvar = (1-beta) * rttvar + beta * |srtt-observed|
3506 		 *
3507 		 * delta is in units of 1/32 ticks, and has then been
3508 		 * divided by 8.  This is equivalent to being in 1/16s
3509 		 * units and divided by 4.  Subtract from it 1/4 of
3510 		 * the existing rttvar to form the (signed) amount to
3511 		 * adjust.
3512 		 */
3513 		if (delta < 0)
3514 			delta = -delta;
3515 		delta -= (tp->t_rttvar >> TCP_RTTVAR_SHIFT);
3516 		/*
3517 		 * As with srtt, this should never happen.  There is
3518 		 * no support in RFC2988 for this operation.  But 1/4s
3519 		 * as rttvar when faced with something arguably wrong
3520 		 * is ok.
3521 		 */
3522 		if ((tp->t_rttvar += delta) <= 0)
3523 			tp->t_rttvar = 1 << 2;
3524 
3525 		/*
3526 		 * If srtt exceeds .01 second, ensure we use the 'remote' MSL
3527 		 * Problem is: it doesn't work.  Disabled by defaulting
3528 		 * tcp_rttlocal to 0; see corresponding code in
3529 		 * tcp_subr that selects local vs remote in a different way.
3530 		 *
3531 		 * The static branch prediction hint here should be removed
3532 		 * when the rtt estimator is fixed and the rtt_enable code
3533 		 * is turned back on.
3534 		 */
3535 		if (__predict_false(tcp_rttlocal) && tcp_msl_enable
3536 		    && tp->t_srtt > tcp_msl_remote_threshold
3537 		    && tp->t_msl  < tcp_msl_remote) {
3538 			tp->t_msl = tcp_msl_remote;
3539 		}
3540 	} else {
3541 		/*
3542 		 * This is the first measurement.  Per RFC2988, 2.2,
3543 		 * set rtt=R and srtt=R/2.
3544 		 * For srtt, storage representation is 1/32 ticks,
3545 		 * so shift left by 5.
3546 		 * For rttvar, storage representation is 1/16 ticks,
3547 		 * So shift left by 4, but then right by 1 to halve.
3548 		 */
3549 		tp->t_srtt = rtt << (TCP_RTT_SHIFT + 2);
3550 		tp->t_rttvar = rtt << (TCP_RTTVAR_SHIFT + 2 - 1);
3551 	}
3552 	tp->t_rtttime = 0;
3553 	tp->t_rxtshift = 0;
3554 
3555 	/*
3556 	 * the retransmit should happen at rtt + 4 * rttvar.
3557 	 * Because of the way we do the smoothing, srtt and rttvar
3558 	 * will each average +1/2 tick of bias.  When we compute
3559 	 * the retransmit timer, we want 1/2 tick of rounding and
3560 	 * 1 extra tick because of +-1/2 tick uncertainty in the
3561 	 * firing of the timer.  The bias will give us exactly the
3562 	 * 1.5 tick we need.  But, because the bias is
3563 	 * statistical, we have to test that we don't drop below
3564 	 * the minimum feasible timer (which is 2 ticks).
3565 	 */
3566 	TCPT_RANGESET(tp->t_rxtcur, TCP_REXMTVAL(tp),
3567 	    max(tp->t_rttmin, rtt + 2), TCPTV_REXMTMAX);
3568 
3569 	/*
3570 	 * We received an ack for a packet that wasn't retransmitted;
3571 	 * it is probably safe to discard any error indications we've
3572 	 * received recently.  This isn't quite right, but close enough
3573 	 * for now (a route might have failed after we sent a segment,
3574 	 * and the return path might not be symmetrical).
3575 	 */
3576 	tp->t_softerror = 0;
3577 }
3578 
3579 
3580 /*
3581  * TCP compressed state engine.  Currently used to hold compressed
3582  * state for SYN_RECEIVED.
3583  */
3584 
3585 u_long	syn_cache_count;
3586 u_int32_t syn_hash1, syn_hash2;
3587 
3588 #define SYN_HASH(sa, sp, dp) \
3589 	((((sa)->s_addr^syn_hash1)*(((((u_int32_t)(dp))<<16) + \
3590 				     ((u_int32_t)(sp)))^syn_hash2)))
3591 #ifndef INET6
3592 #define	SYN_HASHALL(hash, src, dst) \
3593 do {									\
3594 	hash = SYN_HASH(&((const struct sockaddr_in *)(src))->sin_addr,	\
3595 		((const struct sockaddr_in *)(src))->sin_port,		\
3596 		((const struct sockaddr_in *)(dst))->sin_port);		\
3597 } while (/*CONSTCOND*/ 0)
3598 #else
3599 #define SYN_HASH6(sa, sp, dp) \
3600 	((((sa)->s6_addr32[0] ^ (sa)->s6_addr32[3] ^ syn_hash1) * \
3601 	  (((((u_int32_t)(dp))<<16) + ((u_int32_t)(sp)))^syn_hash2)) \
3602 	 & 0x7fffffff)
3603 
3604 #define SYN_HASHALL(hash, src, dst) \
3605 do {									\
3606 	switch ((src)->sa_family) {					\
3607 	case AF_INET:							\
3608 		hash = SYN_HASH(&((const struct sockaddr_in *)(src))->sin_addr, \
3609 			((const struct sockaddr_in *)(src))->sin_port,	\
3610 			((const struct sockaddr_in *)(dst))->sin_port);	\
3611 		break;							\
3612 	case AF_INET6:							\
3613 		hash = SYN_HASH6(&((const struct sockaddr_in6 *)(src))->sin6_addr, \
3614 			((const struct sockaddr_in6 *)(src))->sin6_port,	\
3615 			((const struct sockaddr_in6 *)(dst))->sin6_port);	\
3616 		break;							\
3617 	default:							\
3618 		hash = 0;						\
3619 	}								\
3620 } while (/*CONSTCOND*/0)
3621 #endif /* INET6 */
3622 
3623 static struct pool syn_cache_pool;
3624 
3625 /*
3626  * We don't estimate RTT with SYNs, so each packet starts with the default
3627  * RTT and each timer step has a fixed timeout value.
3628  */
3629 #define	SYN_CACHE_TIMER_ARM(sc)						\
3630 do {									\
3631 	TCPT_RANGESET((sc)->sc_rxtcur,					\
3632 	    TCPTV_SRTTDFLT * tcp_backoff[(sc)->sc_rxtshift], TCPTV_MIN,	\
3633 	    TCPTV_REXMTMAX);						\
3634 	callout_reset(&(sc)->sc_timer,					\
3635 	    (sc)->sc_rxtcur * (hz / PR_SLOWHZ), syn_cache_timer, (sc));	\
3636 } while (/*CONSTCOND*/0)
3637 
3638 #define	SYN_CACHE_TIMESTAMP(sc)	(tcp_now - (sc)->sc_timebase)
3639 
3640 static inline void
3641 syn_cache_rm(struct syn_cache *sc)
3642 {
3643 	TAILQ_REMOVE(&tcp_syn_cache[sc->sc_bucketidx].sch_bucket,
3644 	    sc, sc_bucketq);
3645 	sc->sc_tp = NULL;
3646 	LIST_REMOVE(sc, sc_tpq);
3647 	tcp_syn_cache[sc->sc_bucketidx].sch_length--;
3648 	callout_stop(&sc->sc_timer);
3649 	syn_cache_count--;
3650 }
3651 
3652 static inline void
3653 syn_cache_put(struct syn_cache *sc)
3654 {
3655 	if (sc->sc_ipopts)
3656 		(void) m_free(sc->sc_ipopts);
3657 	rtcache_free(&sc->sc_route);
3658 	sc->sc_flags |= SCF_DEAD;
3659 	if (!callout_invoking(&sc->sc_timer))
3660 		callout_schedule(&(sc)->sc_timer, 1);
3661 }
3662 
3663 void
3664 syn_cache_init(void)
3665 {
3666 	int i;
3667 
3668 	pool_init(&syn_cache_pool, sizeof(struct syn_cache), 0, 0, 0,
3669 	    "synpl", NULL, IPL_SOFTNET);
3670 
3671 	/* Initialize the hash buckets. */
3672 	for (i = 0; i < tcp_syn_cache_size; i++)
3673 		TAILQ_INIT(&tcp_syn_cache[i].sch_bucket);
3674 }
3675 
3676 void
3677 syn_cache_insert(struct syn_cache *sc, struct tcpcb *tp)
3678 {
3679 	struct syn_cache_head *scp;
3680 	struct syn_cache *sc2;
3681 	int s;
3682 
3683 	/*
3684 	 * If there are no entries in the hash table, reinitialize
3685 	 * the hash secrets.
3686 	 */
3687 	if (syn_cache_count == 0) {
3688 		syn_hash1 = cprng_fast32();
3689 		syn_hash2 = cprng_fast32();
3690 	}
3691 
3692 	SYN_HASHALL(sc->sc_hash, &sc->sc_src.sa, &sc->sc_dst.sa);
3693 	sc->sc_bucketidx = sc->sc_hash % tcp_syn_cache_size;
3694 	scp = &tcp_syn_cache[sc->sc_bucketidx];
3695 
3696 	/*
3697 	 * Make sure that we don't overflow the per-bucket
3698 	 * limit or the total cache size limit.
3699 	 */
3700 	s = splsoftnet();
3701 	if (scp->sch_length >= tcp_syn_bucket_limit) {
3702 		TCP_STATINC(TCP_STAT_SC_BUCKETOVERFLOW);
3703 		/*
3704 		 * The bucket is full.  Toss the oldest element in the
3705 		 * bucket.  This will be the first entry in the bucket.
3706 		 */
3707 		sc2 = TAILQ_FIRST(&scp->sch_bucket);
3708 #ifdef DIAGNOSTIC
3709 		/*
3710 		 * This should never happen; we should always find an
3711 		 * entry in our bucket.
3712 		 */
3713 		if (sc2 == NULL)
3714 			panic("syn_cache_insert: bucketoverflow: impossible");
3715 #endif
3716 		syn_cache_rm(sc2);
3717 		syn_cache_put(sc2);	/* calls pool_put but see spl above */
3718 	} else if (syn_cache_count >= tcp_syn_cache_limit) {
3719 		struct syn_cache_head *scp2, *sce;
3720 
3721 		TCP_STATINC(TCP_STAT_SC_OVERFLOWED);
3722 		/*
3723 		 * The cache is full.  Toss the oldest entry in the
3724 		 * first non-empty bucket we can find.
3725 		 *
3726 		 * XXX We would really like to toss the oldest
3727 		 * entry in the cache, but we hope that this
3728 		 * condition doesn't happen very often.
3729 		 */
3730 		scp2 = scp;
3731 		if (TAILQ_EMPTY(&scp2->sch_bucket)) {
3732 			sce = &tcp_syn_cache[tcp_syn_cache_size];
3733 			for (++scp2; scp2 != scp; scp2++) {
3734 				if (scp2 >= sce)
3735 					scp2 = &tcp_syn_cache[0];
3736 				if (! TAILQ_EMPTY(&scp2->sch_bucket))
3737 					break;
3738 			}
3739 #ifdef DIAGNOSTIC
3740 			/*
3741 			 * This should never happen; we should always find a
3742 			 * non-empty bucket.
3743 			 */
3744 			if (scp2 == scp)
3745 				panic("syn_cache_insert: cacheoverflow: "
3746 				    "impossible");
3747 #endif
3748 		}
3749 		sc2 = TAILQ_FIRST(&scp2->sch_bucket);
3750 		syn_cache_rm(sc2);
3751 		syn_cache_put(sc2);	/* calls pool_put but see spl above */
3752 	}
3753 
3754 	/*
3755 	 * Initialize the entry's timer.
3756 	 */
3757 	sc->sc_rxttot = 0;
3758 	sc->sc_rxtshift = 0;
3759 	SYN_CACHE_TIMER_ARM(sc);
3760 
3761 	/* Link it from tcpcb entry */
3762 	LIST_INSERT_HEAD(&tp->t_sc, sc, sc_tpq);
3763 
3764 	/* Put it into the bucket. */
3765 	TAILQ_INSERT_TAIL(&scp->sch_bucket, sc, sc_bucketq);
3766 	scp->sch_length++;
3767 	syn_cache_count++;
3768 
3769 	TCP_STATINC(TCP_STAT_SC_ADDED);
3770 	splx(s);
3771 }
3772 
3773 /*
3774  * Walk the timer queues, looking for SYN,ACKs that need to be retransmitted.
3775  * If we have retransmitted an entry the maximum number of times, expire
3776  * that entry.
3777  */
3778 void
3779 syn_cache_timer(void *arg)
3780 {
3781 	struct syn_cache *sc = arg;
3782 
3783 	mutex_enter(softnet_lock);
3784 	KERNEL_LOCK(1, NULL);
3785 	callout_ack(&sc->sc_timer);
3786 
3787 	if (__predict_false(sc->sc_flags & SCF_DEAD)) {
3788 		TCP_STATINC(TCP_STAT_SC_DELAYED_FREE);
3789 		callout_destroy(&sc->sc_timer);
3790 		pool_put(&syn_cache_pool, sc);
3791 		KERNEL_UNLOCK_ONE(NULL);
3792 		mutex_exit(softnet_lock);
3793 		return;
3794 	}
3795 
3796 	if (__predict_false(sc->sc_rxtshift == TCP_MAXRXTSHIFT)) {
3797 		/* Drop it -- too many retransmissions. */
3798 		goto dropit;
3799 	}
3800 
3801 	/*
3802 	 * Compute the total amount of time this entry has
3803 	 * been on a queue.  If this entry has been on longer
3804 	 * than the keep alive timer would allow, expire it.
3805 	 */
3806 	sc->sc_rxttot += sc->sc_rxtcur;
3807 	if (sc->sc_rxttot >= tcp_keepinit)
3808 		goto dropit;
3809 
3810 	TCP_STATINC(TCP_STAT_SC_RETRANSMITTED);
3811 	(void) syn_cache_respond(sc, NULL);
3812 
3813 	/* Advance the timer back-off. */
3814 	sc->sc_rxtshift++;
3815 	SYN_CACHE_TIMER_ARM(sc);
3816 
3817 	KERNEL_UNLOCK_ONE(NULL);
3818 	mutex_exit(softnet_lock);
3819 	return;
3820 
3821  dropit:
3822 	TCP_STATINC(TCP_STAT_SC_TIMED_OUT);
3823 	syn_cache_rm(sc);
3824 	if (sc->sc_ipopts)
3825 		(void) m_free(sc->sc_ipopts);
3826 	rtcache_free(&sc->sc_route);
3827 	callout_destroy(&sc->sc_timer);
3828 	pool_put(&syn_cache_pool, sc);
3829 	KERNEL_UNLOCK_ONE(NULL);
3830 	mutex_exit(softnet_lock);
3831 }
3832 
3833 /*
3834  * Remove syn cache created by the specified tcb entry,
3835  * because this does not make sense to keep them
3836  * (if there's no tcb entry, syn cache entry will never be used)
3837  */
3838 void
3839 syn_cache_cleanup(struct tcpcb *tp)
3840 {
3841 	struct syn_cache *sc, *nsc;
3842 	int s;
3843 
3844 	s = splsoftnet();
3845 
3846 	for (sc = LIST_FIRST(&tp->t_sc); sc != NULL; sc = nsc) {
3847 		nsc = LIST_NEXT(sc, sc_tpq);
3848 
3849 #ifdef DIAGNOSTIC
3850 		if (sc->sc_tp != tp)
3851 			panic("invalid sc_tp in syn_cache_cleanup");
3852 #endif
3853 		syn_cache_rm(sc);
3854 		syn_cache_put(sc);	/* calls pool_put but see spl above */
3855 	}
3856 	/* just for safety */
3857 	LIST_INIT(&tp->t_sc);
3858 
3859 	splx(s);
3860 }
3861 
3862 /*
3863  * Find an entry in the syn cache.
3864  */
3865 struct syn_cache *
3866 syn_cache_lookup(const struct sockaddr *src, const struct sockaddr *dst,
3867     struct syn_cache_head **headp)
3868 {
3869 	struct syn_cache *sc;
3870 	struct syn_cache_head *scp;
3871 	u_int32_t hash;
3872 	int s;
3873 
3874 	SYN_HASHALL(hash, src, dst);
3875 
3876 	scp = &tcp_syn_cache[hash % tcp_syn_cache_size];
3877 	*headp = scp;
3878 	s = splsoftnet();
3879 	for (sc = TAILQ_FIRST(&scp->sch_bucket); sc != NULL;
3880 	     sc = TAILQ_NEXT(sc, sc_bucketq)) {
3881 		if (sc->sc_hash != hash)
3882 			continue;
3883 		if (!memcmp(&sc->sc_src, src, src->sa_len) &&
3884 		    !memcmp(&sc->sc_dst, dst, dst->sa_len)) {
3885 			splx(s);
3886 			return (sc);
3887 		}
3888 	}
3889 	splx(s);
3890 	return (NULL);
3891 }
3892 
3893 /*
3894  * This function gets called when we receive an ACK for a
3895  * socket in the LISTEN state.  We look up the connection
3896  * in the syn cache, and if its there, we pull it out of
3897  * the cache and turn it into a full-blown connection in
3898  * the SYN-RECEIVED state.
3899  *
3900  * The return values may not be immediately obvious, and their effects
3901  * can be subtle, so here they are:
3902  *
3903  *	NULL	SYN was not found in cache; caller should drop the
3904  *		packet and send an RST.
3905  *
3906  *	-1	We were unable to create the new connection, and are
3907  *		aborting it.  An ACK,RST is being sent to the peer
3908  *		(unless we got screwey sequence numbners; see below),
3909  *		because the 3-way handshake has been completed.  Caller
3910  *		should not free the mbuf, since we may be using it.  If
3911  *		we are not, we will free it.
3912  *
3913  *	Otherwise, the return value is a pointer to the new socket
3914  *	associated with the connection.
3915  */
3916 struct socket *
3917 syn_cache_get(struct sockaddr *src, struct sockaddr *dst,
3918     struct tcphdr *th, unsigned int hlen, unsigned int tlen,
3919     struct socket *so, struct mbuf *m)
3920 {
3921 	struct syn_cache *sc;
3922 	struct syn_cache_head *scp;
3923 	struct inpcb *inp = NULL;
3924 #ifdef INET6
3925 	struct in6pcb *in6p = NULL;
3926 #endif
3927 	struct tcpcb *tp = 0;
3928 	struct mbuf *am;
3929 	int s;
3930 	struct socket *oso;
3931 
3932 	s = splsoftnet();
3933 	if ((sc = syn_cache_lookup(src, dst, &scp)) == NULL) {
3934 		splx(s);
3935 		return (NULL);
3936 	}
3937 
3938 	/*
3939 	 * Verify the sequence and ack numbers.  Try getting the correct
3940 	 * response again.
3941 	 */
3942 	if ((th->th_ack != sc->sc_iss + 1) ||
3943 	    SEQ_LEQ(th->th_seq, sc->sc_irs) ||
3944 	    SEQ_GT(th->th_seq, sc->sc_irs + 1 + sc->sc_win)) {
3945 		(void) syn_cache_respond(sc, m);
3946 		splx(s);
3947 		return ((struct socket *)(-1));
3948 	}
3949 
3950 	/* Remove this cache entry */
3951 	syn_cache_rm(sc);
3952 	splx(s);
3953 
3954 	/*
3955 	 * Ok, create the full blown connection, and set things up
3956 	 * as they would have been set up if we had created the
3957 	 * connection when the SYN arrived.  If we can't create
3958 	 * the connection, abort it.
3959 	 */
3960 	/*
3961 	 * inp still has the OLD in_pcb stuff, set the
3962 	 * v6-related flags on the new guy, too.   This is
3963 	 * done particularly for the case where an AF_INET6
3964 	 * socket is bound only to a port, and a v4 connection
3965 	 * comes in on that port.
3966 	 * we also copy the flowinfo from the original pcb
3967 	 * to the new one.
3968 	 */
3969 	oso = so;
3970 	so = sonewconn(so, true);
3971 	if (so == NULL)
3972 		goto resetandabort;
3973 
3974 	switch (so->so_proto->pr_domain->dom_family) {
3975 #ifdef INET
3976 	case AF_INET:
3977 		inp = sotoinpcb(so);
3978 		break;
3979 #endif
3980 #ifdef INET6
3981 	case AF_INET6:
3982 		in6p = sotoin6pcb(so);
3983 		break;
3984 #endif
3985 	}
3986 	switch (src->sa_family) {
3987 #ifdef INET
3988 	case AF_INET:
3989 		if (inp) {
3990 			inp->inp_laddr = ((struct sockaddr_in *)dst)->sin_addr;
3991 			inp->inp_lport = ((struct sockaddr_in *)dst)->sin_port;
3992 			inp->inp_options = ip_srcroute();
3993 			in_pcbstate(inp, INP_BOUND);
3994 			if (inp->inp_options == NULL) {
3995 				inp->inp_options = sc->sc_ipopts;
3996 				sc->sc_ipopts = NULL;
3997 			}
3998 		}
3999 #ifdef INET6
4000 		else if (in6p) {
4001 			/* IPv4 packet to AF_INET6 socket */
4002 			memset(&in6p->in6p_laddr, 0, sizeof(in6p->in6p_laddr));
4003 			in6p->in6p_laddr.s6_addr16[5] = htons(0xffff);
4004 			bcopy(&((struct sockaddr_in *)dst)->sin_addr,
4005 				&in6p->in6p_laddr.s6_addr32[3],
4006 				sizeof(((struct sockaddr_in *)dst)->sin_addr));
4007 			in6p->in6p_lport = ((struct sockaddr_in *)dst)->sin_port;
4008 			in6totcpcb(in6p)->t_family = AF_INET;
4009 			if (sotoin6pcb(oso)->in6p_flags & IN6P_IPV6_V6ONLY)
4010 				in6p->in6p_flags |= IN6P_IPV6_V6ONLY;
4011 			else
4012 				in6p->in6p_flags &= ~IN6P_IPV6_V6ONLY;
4013 			in6_pcbstate(in6p, IN6P_BOUND);
4014 		}
4015 #endif
4016 		break;
4017 #endif
4018 #ifdef INET6
4019 	case AF_INET6:
4020 		if (in6p) {
4021 			in6p->in6p_laddr = ((struct sockaddr_in6 *)dst)->sin6_addr;
4022 			in6p->in6p_lport = ((struct sockaddr_in6 *)dst)->sin6_port;
4023 			in6_pcbstate(in6p, IN6P_BOUND);
4024 		}
4025 		break;
4026 #endif
4027 	}
4028 #ifdef INET6
4029 	if (in6p && in6totcpcb(in6p)->t_family == AF_INET6 && sotoinpcb(oso)) {
4030 		struct in6pcb *oin6p = sotoin6pcb(oso);
4031 		/* inherit socket options from the listening socket */
4032 		in6p->in6p_flags |= (oin6p->in6p_flags & IN6P_CONTROLOPTS);
4033 		if (in6p->in6p_flags & IN6P_CONTROLOPTS) {
4034 			m_freem(in6p->in6p_options);
4035 			in6p->in6p_options = 0;
4036 		}
4037 		ip6_savecontrol(in6p, &in6p->in6p_options,
4038 			mtod(m, struct ip6_hdr *), m);
4039 	}
4040 #endif
4041 
4042 #if defined(IPSEC)
4043 	/*
4044 	 * we make a copy of policy, instead of sharing the policy,
4045 	 * for better behavior in terms of SA lookup and dead SA removal.
4046 	 */
4047 	if (inp) {
4048 		/* copy old policy into new socket's */
4049 		if (ipsec_copy_pcbpolicy(sotoinpcb(oso)->inp_sp, inp->inp_sp))
4050 			printf("tcp_input: could not copy policy\n");
4051 	}
4052 #ifdef INET6
4053 	else if (in6p) {
4054 		/* copy old policy into new socket's */
4055 		if (ipsec_copy_pcbpolicy(sotoin6pcb(oso)->in6p_sp,
4056 		    in6p->in6p_sp))
4057 			printf("tcp_input: could not copy policy\n");
4058 	}
4059 #endif
4060 #endif
4061 
4062 	/*
4063 	 * Give the new socket our cached route reference.
4064 	 */
4065 	if (inp) {
4066 		rtcache_copy(&inp->inp_route, &sc->sc_route);
4067 		rtcache_free(&sc->sc_route);
4068 	}
4069 #ifdef INET6
4070 	else {
4071 		rtcache_copy(&in6p->in6p_route, &sc->sc_route);
4072 		rtcache_free(&sc->sc_route);
4073 	}
4074 #endif
4075 
4076 	am = m_get(M_DONTWAIT, MT_SONAME);	/* XXX */
4077 	if (am == NULL)
4078 		goto resetandabort;
4079 	MCLAIM(am, &tcp_mowner);
4080 	am->m_len = src->sa_len;
4081 	bcopy(src, mtod(am, void *), src->sa_len);
4082 	if (inp) {
4083 		if (in_pcbconnect(inp, am, &lwp0)) {
4084 			(void) m_free(am);
4085 			goto resetandabort;
4086 		}
4087 	}
4088 #ifdef INET6
4089 	else if (in6p) {
4090 		if (src->sa_family == AF_INET) {
4091 			/* IPv4 packet to AF_INET6 socket */
4092 			struct sockaddr_in6 *sin6;
4093 			sin6 = mtod(am, struct sockaddr_in6 *);
4094 			am->m_len = sizeof(*sin6);
4095 			memset(sin6, 0, sizeof(*sin6));
4096 			sin6->sin6_family = AF_INET6;
4097 			sin6->sin6_len = sizeof(*sin6);
4098 			sin6->sin6_port = ((struct sockaddr_in *)src)->sin_port;
4099 			sin6->sin6_addr.s6_addr16[5] = htons(0xffff);
4100 			bcopy(&((struct sockaddr_in *)src)->sin_addr,
4101 				&sin6->sin6_addr.s6_addr32[3],
4102 				sizeof(sin6->sin6_addr.s6_addr32[3]));
4103 		}
4104 		if (in6_pcbconnect(in6p, am, NULL)) {
4105 			(void) m_free(am);
4106 			goto resetandabort;
4107 		}
4108 	}
4109 #endif
4110 	else {
4111 		(void) m_free(am);
4112 		goto resetandabort;
4113 	}
4114 	(void) m_free(am);
4115 
4116 	if (inp)
4117 		tp = intotcpcb(inp);
4118 #ifdef INET6
4119 	else if (in6p)
4120 		tp = in6totcpcb(in6p);
4121 #endif
4122 	else
4123 		tp = NULL;
4124 	tp->t_flags = sototcpcb(oso)->t_flags & TF_NODELAY;
4125 	if (sc->sc_request_r_scale != 15) {
4126 		tp->requested_s_scale = sc->sc_requested_s_scale;
4127 		tp->request_r_scale = sc->sc_request_r_scale;
4128 		tp->snd_scale = sc->sc_requested_s_scale;
4129 		tp->rcv_scale = sc->sc_request_r_scale;
4130 		tp->t_flags |= TF_REQ_SCALE|TF_RCVD_SCALE;
4131 	}
4132 	if (sc->sc_flags & SCF_TIMESTAMP)
4133 		tp->t_flags |= TF_REQ_TSTMP|TF_RCVD_TSTMP;
4134 	tp->ts_timebase = sc->sc_timebase;
4135 
4136 	tp->t_template = tcp_template(tp);
4137 	if (tp->t_template == 0) {
4138 		tp = tcp_drop(tp, ENOBUFS);	/* destroys socket */
4139 		so = NULL;
4140 		m_freem(m);
4141 		goto abort;
4142 	}
4143 
4144 	tp->iss = sc->sc_iss;
4145 	tp->irs = sc->sc_irs;
4146 	tcp_sendseqinit(tp);
4147 	tcp_rcvseqinit(tp);
4148 	tp->t_state = TCPS_SYN_RECEIVED;
4149 	TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepinit);
4150 	TCP_STATINC(TCP_STAT_ACCEPTS);
4151 
4152 	if ((sc->sc_flags & SCF_SACK_PERMIT) && tcp_do_sack)
4153 		tp->t_flags |= TF_WILL_SACK;
4154 
4155 	if ((sc->sc_flags & SCF_ECN_PERMIT) && tcp_do_ecn)
4156 		tp->t_flags |= TF_ECN_PERMIT;
4157 
4158 #ifdef TCP_SIGNATURE
4159 	if (sc->sc_flags & SCF_SIGNATURE)
4160 		tp->t_flags |= TF_SIGNATURE;
4161 #endif
4162 
4163 	/* Initialize tp->t_ourmss before we deal with the peer's! */
4164 	tp->t_ourmss = sc->sc_ourmaxseg;
4165 	tcp_mss_from_peer(tp, sc->sc_peermaxseg);
4166 
4167 	/*
4168 	 * Initialize the initial congestion window.  If we
4169 	 * had to retransmit the SYN,ACK, we must initialize cwnd
4170 	 * to 1 segment (i.e. the Loss Window).
4171 	 */
4172 	if (sc->sc_rxtshift)
4173 		tp->snd_cwnd = tp->t_peermss;
4174 	else {
4175 		int ss = tcp_init_win;
4176 #ifdef INET
4177 		if (inp != NULL && in_localaddr(inp->inp_faddr))
4178 			ss = tcp_init_win_local;
4179 #endif
4180 #ifdef INET6
4181 		if (in6p != NULL && in6_localaddr(&in6p->in6p_faddr))
4182 			ss = tcp_init_win_local;
4183 #endif
4184 		tp->snd_cwnd = TCP_INITIAL_WINDOW(ss, tp->t_peermss);
4185 	}
4186 
4187 	tcp_rmx_rtt(tp);
4188 	tp->snd_wl1 = sc->sc_irs;
4189 	tp->rcv_up = sc->sc_irs + 1;
4190 
4191 	/*
4192 	 * This is what whould have happened in tcp_output() when
4193 	 * the SYN,ACK was sent.
4194 	 */
4195 	tp->snd_up = tp->snd_una;
4196 	tp->snd_max = tp->snd_nxt = tp->iss+1;
4197 	TCP_TIMER_ARM(tp, TCPT_REXMT, tp->t_rxtcur);
4198 	if (sc->sc_win > 0 && SEQ_GT(tp->rcv_nxt + sc->sc_win, tp->rcv_adv))
4199 		tp->rcv_adv = tp->rcv_nxt + sc->sc_win;
4200 	tp->last_ack_sent = tp->rcv_nxt;
4201 	tp->t_partialacks = -1;
4202 	tp->t_dupacks = 0;
4203 
4204 	TCP_STATINC(TCP_STAT_SC_COMPLETED);
4205 	s = splsoftnet();
4206 	syn_cache_put(sc);
4207 	splx(s);
4208 	return (so);
4209 
4210 resetandabort:
4211 	(void)tcp_respond(NULL, m, m, th, (tcp_seq)0, th->th_ack, TH_RST);
4212 abort:
4213 	if (so != NULL) {
4214 		(void) soqremque(so, 1);
4215 		(void) soabort(so);
4216 		mutex_enter(softnet_lock);
4217 	}
4218 	s = splsoftnet();
4219 	syn_cache_put(sc);
4220 	splx(s);
4221 	TCP_STATINC(TCP_STAT_SC_ABORTED);
4222 	return ((struct socket *)(-1));
4223 }
4224 
4225 /*
4226  * This function is called when we get a RST for a
4227  * non-existent connection, so that we can see if the
4228  * connection is in the syn cache.  If it is, zap it.
4229  */
4230 
4231 void
4232 syn_cache_reset(struct sockaddr *src, struct sockaddr *dst, struct tcphdr *th)
4233 {
4234 	struct syn_cache *sc;
4235 	struct syn_cache_head *scp;
4236 	int s = splsoftnet();
4237 
4238 	if ((sc = syn_cache_lookup(src, dst, &scp)) == NULL) {
4239 		splx(s);
4240 		return;
4241 	}
4242 	if (SEQ_LT(th->th_seq, sc->sc_irs) ||
4243 	    SEQ_GT(th->th_seq, sc->sc_irs+1)) {
4244 		splx(s);
4245 		return;
4246 	}
4247 	syn_cache_rm(sc);
4248 	TCP_STATINC(TCP_STAT_SC_RESET);
4249 	syn_cache_put(sc);	/* calls pool_put but see spl above */
4250 	splx(s);
4251 }
4252 
4253 void
4254 syn_cache_unreach(const struct sockaddr *src, const struct sockaddr *dst,
4255     struct tcphdr *th)
4256 {
4257 	struct syn_cache *sc;
4258 	struct syn_cache_head *scp;
4259 	int s;
4260 
4261 	s = splsoftnet();
4262 	if ((sc = syn_cache_lookup(src, dst, &scp)) == NULL) {
4263 		splx(s);
4264 		return;
4265 	}
4266 	/* If the sequence number != sc_iss, then it's a bogus ICMP msg */
4267 	if (ntohl (th->th_seq) != sc->sc_iss) {
4268 		splx(s);
4269 		return;
4270 	}
4271 
4272 	/*
4273 	 * If we've retransmitted 3 times and this is our second error,
4274 	 * we remove the entry.  Otherwise, we allow it to continue on.
4275 	 * This prevents us from incorrectly nuking an entry during a
4276 	 * spurious network outage.
4277 	 *
4278 	 * See tcp_notify().
4279 	 */
4280 	if ((sc->sc_flags & SCF_UNREACH) == 0 || sc->sc_rxtshift < 3) {
4281 		sc->sc_flags |= SCF_UNREACH;
4282 		splx(s);
4283 		return;
4284 	}
4285 
4286 	syn_cache_rm(sc);
4287 	TCP_STATINC(TCP_STAT_SC_UNREACH);
4288 	syn_cache_put(sc);	/* calls pool_put but see spl above */
4289 	splx(s);
4290 }
4291 
4292 /*
4293  * Given a LISTEN socket and an inbound SYN request, add
4294  * this to the syn cache, and send back a segment:
4295  *	<SEQ=ISS><ACK=RCV_NXT><CTL=SYN,ACK>
4296  * to the source.
4297  *
4298  * IMPORTANT NOTE: We do _NOT_ ACK data that might accompany the SYN.
4299  * Doing so would require that we hold onto the data and deliver it
4300  * to the application.  However, if we are the target of a SYN-flood
4301  * DoS attack, an attacker could send data which would eventually
4302  * consume all available buffer space if it were ACKed.  By not ACKing
4303  * the data, we avoid this DoS scenario.
4304  */
4305 
4306 int
4307 syn_cache_add(struct sockaddr *src, struct sockaddr *dst, struct tcphdr *th,
4308     unsigned int hlen, struct socket *so, struct mbuf *m, u_char *optp,
4309     int optlen, struct tcp_opt_info *oi)
4310 {
4311 	struct tcpcb tb, *tp;
4312 	long win;
4313 	struct syn_cache *sc;
4314 	struct syn_cache_head *scp;
4315 	struct mbuf *ipopts;
4316 	struct tcp_opt_info opti;
4317 	int s;
4318 
4319 	tp = sototcpcb(so);
4320 
4321 	memset(&opti, 0, sizeof(opti));
4322 
4323 	/*
4324 	 * RFC1122 4.2.3.10, p. 104: discard bcast/mcast SYN
4325 	 *
4326 	 * Note this check is performed in tcp_input() very early on.
4327 	 */
4328 
4329 	/*
4330 	 * Initialize some local state.
4331 	 */
4332 	win = sbspace(&so->so_rcv);
4333 	if (win > TCP_MAXWIN)
4334 		win = TCP_MAXWIN;
4335 
4336 	switch (src->sa_family) {
4337 #ifdef INET
4338 	case AF_INET:
4339 		/*
4340 		 * Remember the IP options, if any.
4341 		 */
4342 		ipopts = ip_srcroute();
4343 		break;
4344 #endif
4345 	default:
4346 		ipopts = NULL;
4347 	}
4348 
4349 #ifdef TCP_SIGNATURE
4350 	if (optp || (tp->t_flags & TF_SIGNATURE))
4351 #else
4352 	if (optp)
4353 #endif
4354 	{
4355 		tb.t_flags = tcp_do_rfc1323 ? (TF_REQ_SCALE|TF_REQ_TSTMP) : 0;
4356 #ifdef TCP_SIGNATURE
4357 		tb.t_flags |= (tp->t_flags & TF_SIGNATURE);
4358 #endif
4359 		tb.t_state = TCPS_LISTEN;
4360 		if (tcp_dooptions(&tb, optp, optlen, th, m, m->m_pkthdr.len -
4361 		    sizeof(struct tcphdr) - optlen - hlen, oi) < 0)
4362 			return (0);
4363 	} else
4364 		tb.t_flags = 0;
4365 
4366 	/*
4367 	 * See if we already have an entry for this connection.
4368 	 * If we do, resend the SYN,ACK.  We do not count this
4369 	 * as a retransmission (XXX though maybe we should).
4370 	 */
4371 	if ((sc = syn_cache_lookup(src, dst, &scp)) != NULL) {
4372 		TCP_STATINC(TCP_STAT_SC_DUPESYN);
4373 		if (ipopts) {
4374 			/*
4375 			 * If we were remembering a previous source route,
4376 			 * forget it and use the new one we've been given.
4377 			 */
4378 			if (sc->sc_ipopts)
4379 				(void) m_free(sc->sc_ipopts);
4380 			sc->sc_ipopts = ipopts;
4381 		}
4382 		sc->sc_timestamp = tb.ts_recent;
4383 		if (syn_cache_respond(sc, m) == 0) {
4384 			uint64_t *tcps = TCP_STAT_GETREF();
4385 			tcps[TCP_STAT_SNDACKS]++;
4386 			tcps[TCP_STAT_SNDTOTAL]++;
4387 			TCP_STAT_PUTREF();
4388 		}
4389 		return (1);
4390 	}
4391 
4392 	s = splsoftnet();
4393 	sc = pool_get(&syn_cache_pool, PR_NOWAIT);
4394 	splx(s);
4395 	if (sc == NULL) {
4396 		if (ipopts)
4397 			(void) m_free(ipopts);
4398 		return (0);
4399 	}
4400 
4401 	/*
4402 	 * Fill in the cache, and put the necessary IP and TCP
4403 	 * options into the reply.
4404 	 */
4405 	memset(sc, 0, sizeof(struct syn_cache));
4406 	callout_init(&sc->sc_timer, CALLOUT_MPSAFE);
4407 	bcopy(src, &sc->sc_src, src->sa_len);
4408 	bcopy(dst, &sc->sc_dst, dst->sa_len);
4409 	sc->sc_flags = 0;
4410 	sc->sc_ipopts = ipopts;
4411 	sc->sc_irs = th->th_seq;
4412 	switch (src->sa_family) {
4413 #ifdef INET
4414 	case AF_INET:
4415 	    {
4416 		struct sockaddr_in *srcin = (void *) src;
4417 		struct sockaddr_in *dstin = (void *) dst;
4418 
4419 		sc->sc_iss = tcp_new_iss1(&dstin->sin_addr,
4420 		    &srcin->sin_addr, dstin->sin_port,
4421 		    srcin->sin_port, sizeof(dstin->sin_addr), 0);
4422 		break;
4423 	    }
4424 #endif /* INET */
4425 #ifdef INET6
4426 	case AF_INET6:
4427 	    {
4428 		struct sockaddr_in6 *srcin6 = (void *) src;
4429 		struct sockaddr_in6 *dstin6 = (void *) dst;
4430 
4431 		sc->sc_iss = tcp_new_iss1(&dstin6->sin6_addr,
4432 		    &srcin6->sin6_addr, dstin6->sin6_port,
4433 		    srcin6->sin6_port, sizeof(dstin6->sin6_addr), 0);
4434 		break;
4435 	    }
4436 #endif /* INET6 */
4437 	}
4438 	sc->sc_peermaxseg = oi->maxseg;
4439 	sc->sc_ourmaxseg = tcp_mss_to_advertise(m->m_flags & M_PKTHDR ?
4440 						m->m_pkthdr.rcvif : NULL,
4441 						sc->sc_src.sa.sa_family);
4442 	sc->sc_win = win;
4443 	sc->sc_timebase = tcp_now - 1;	/* see tcp_newtcpcb() */
4444 	sc->sc_timestamp = tb.ts_recent;
4445 	if ((tb.t_flags & (TF_REQ_TSTMP|TF_RCVD_TSTMP)) ==
4446 	    (TF_REQ_TSTMP|TF_RCVD_TSTMP))
4447 		sc->sc_flags |= SCF_TIMESTAMP;
4448 	if ((tb.t_flags & (TF_RCVD_SCALE|TF_REQ_SCALE)) ==
4449 	    (TF_RCVD_SCALE|TF_REQ_SCALE)) {
4450 		sc->sc_requested_s_scale = tb.requested_s_scale;
4451 		sc->sc_request_r_scale = 0;
4452 		/*
4453 		 * Pick the smallest possible scaling factor that
4454 		 * will still allow us to scale up to sb_max.
4455 		 *
4456 		 * We do this because there are broken firewalls that
4457 		 * will corrupt the window scale option, leading to
4458 		 * the other endpoint believing that our advertised
4459 		 * window is unscaled.  At scale factors larger than
4460 		 * 5 the unscaled window will drop below 1500 bytes,
4461 		 * leading to serious problems when traversing these
4462 		 * broken firewalls.
4463 		 *
4464 		 * With the default sbmax of 256K, a scale factor
4465 		 * of 3 will be chosen by this algorithm.  Those who
4466 		 * choose a larger sbmax should watch out
4467 		 * for the compatiblity problems mentioned above.
4468 		 *
4469 		 * RFC1323: The Window field in a SYN (i.e., a <SYN>
4470 		 * or <SYN,ACK>) segment itself is never scaled.
4471 		 */
4472 		while (sc->sc_request_r_scale < TCP_MAX_WINSHIFT &&
4473 		    (TCP_MAXWIN << sc->sc_request_r_scale) < sb_max)
4474 			sc->sc_request_r_scale++;
4475 	} else {
4476 		sc->sc_requested_s_scale = 15;
4477 		sc->sc_request_r_scale = 15;
4478 	}
4479 	if ((tb.t_flags & TF_SACK_PERMIT) && tcp_do_sack)
4480 		sc->sc_flags |= SCF_SACK_PERMIT;
4481 
4482 	/*
4483 	 * ECN setup packet recieved.
4484 	 */
4485 	if ((th->th_flags & (TH_ECE|TH_CWR)) && tcp_do_ecn)
4486 		sc->sc_flags |= SCF_ECN_PERMIT;
4487 
4488 #ifdef TCP_SIGNATURE
4489 	if (tb.t_flags & TF_SIGNATURE)
4490 		sc->sc_flags |= SCF_SIGNATURE;
4491 #endif
4492 	sc->sc_tp = tp;
4493 	if (syn_cache_respond(sc, m) == 0) {
4494 		uint64_t *tcps = TCP_STAT_GETREF();
4495 		tcps[TCP_STAT_SNDACKS]++;
4496 		tcps[TCP_STAT_SNDTOTAL]++;
4497 		TCP_STAT_PUTREF();
4498 		syn_cache_insert(sc, tp);
4499 	} else {
4500 		s = splsoftnet();
4501 		/*
4502 		 * syn_cache_put() will try to schedule the timer, so
4503 		 * we need to initialize it
4504 		 */
4505 		SYN_CACHE_TIMER_ARM(sc);
4506 		syn_cache_put(sc);
4507 		splx(s);
4508 		TCP_STATINC(TCP_STAT_SC_DROPPED);
4509 	}
4510 	return (1);
4511 }
4512 
4513 /*
4514  * syn_cache_respond: (re)send SYN+ACK.
4515  *
4516  * returns 0 on success.  otherwise returns an errno, typically ENOBUFS.
4517  */
4518 
4519 int
4520 syn_cache_respond(struct syn_cache *sc, struct mbuf *m)
4521 {
4522 #ifdef INET6
4523 	struct rtentry *rt;
4524 #endif
4525 	struct route *ro;
4526 	u_int8_t *optp;
4527 	int optlen, error;
4528 	u_int16_t tlen;
4529 	struct ip *ip = NULL;
4530 #ifdef INET6
4531 	struct ip6_hdr *ip6 = NULL;
4532 #endif
4533 	struct tcpcb *tp = NULL;
4534 	struct tcphdr *th;
4535 	u_int hlen;
4536 	struct socket *so;
4537 
4538 	ro = &sc->sc_route;
4539 	switch (sc->sc_src.sa.sa_family) {
4540 	case AF_INET:
4541 		hlen = sizeof(struct ip);
4542 		break;
4543 #ifdef INET6
4544 	case AF_INET6:
4545 		hlen = sizeof(struct ip6_hdr);
4546 		break;
4547 #endif
4548 	default:
4549 		if (m)
4550 			m_freem(m);
4551 		return (EAFNOSUPPORT);
4552 	}
4553 
4554 	/* Compute the size of the TCP options. */
4555 	optlen = 4 + (sc->sc_request_r_scale != 15 ? 4 : 0) +
4556 	    ((sc->sc_flags & SCF_SACK_PERMIT) ? (TCPOLEN_SACK_PERMITTED + 2) : 0) +
4557 #ifdef TCP_SIGNATURE
4558 	    ((sc->sc_flags & SCF_SIGNATURE) ? (TCPOLEN_SIGNATURE + 2) : 0) +
4559 #endif
4560 	    ((sc->sc_flags & SCF_TIMESTAMP) ? TCPOLEN_TSTAMP_APPA : 0);
4561 
4562 	tlen = hlen + sizeof(struct tcphdr) + optlen;
4563 
4564 	/*
4565 	 * Create the IP+TCP header from scratch.
4566 	 */
4567 	if (m)
4568 		m_freem(m);
4569 #ifdef DIAGNOSTIC
4570 	if (max_linkhdr + tlen > MCLBYTES)
4571 		return (ENOBUFS);
4572 #endif
4573 	MGETHDR(m, M_DONTWAIT, MT_DATA);
4574 	if (m && (max_linkhdr + tlen) > MHLEN) {
4575 		MCLGET(m, M_DONTWAIT);
4576 		if ((m->m_flags & M_EXT) == 0) {
4577 			m_freem(m);
4578 			m = NULL;
4579 		}
4580 	}
4581 	if (m == NULL)
4582 		return (ENOBUFS);
4583 	MCLAIM(m, &tcp_tx_mowner);
4584 
4585 	/* Fixup the mbuf. */
4586 	m->m_data += max_linkhdr;
4587 	m->m_len = m->m_pkthdr.len = tlen;
4588 	if (sc->sc_tp) {
4589 		tp = sc->sc_tp;
4590 		if (tp->t_inpcb)
4591 			so = tp->t_inpcb->inp_socket;
4592 #ifdef INET6
4593 		else if (tp->t_in6pcb)
4594 			so = tp->t_in6pcb->in6p_socket;
4595 #endif
4596 		else
4597 			so = NULL;
4598 	} else
4599 		so = NULL;
4600 	m->m_pkthdr.rcvif = NULL;
4601 	memset(mtod(m, u_char *), 0, tlen);
4602 
4603 	switch (sc->sc_src.sa.sa_family) {
4604 	case AF_INET:
4605 		ip = mtod(m, struct ip *);
4606 		ip->ip_v = 4;
4607 		ip->ip_dst = sc->sc_src.sin.sin_addr;
4608 		ip->ip_src = sc->sc_dst.sin.sin_addr;
4609 		ip->ip_p = IPPROTO_TCP;
4610 		th = (struct tcphdr *)(ip + 1);
4611 		th->th_dport = sc->sc_src.sin.sin_port;
4612 		th->th_sport = sc->sc_dst.sin.sin_port;
4613 		break;
4614 #ifdef INET6
4615 	case AF_INET6:
4616 		ip6 = mtod(m, struct ip6_hdr *);
4617 		ip6->ip6_vfc = IPV6_VERSION;
4618 		ip6->ip6_dst = sc->sc_src.sin6.sin6_addr;
4619 		ip6->ip6_src = sc->sc_dst.sin6.sin6_addr;
4620 		ip6->ip6_nxt = IPPROTO_TCP;
4621 		/* ip6_plen will be updated in ip6_output() */
4622 		th = (struct tcphdr *)(ip6 + 1);
4623 		th->th_dport = sc->sc_src.sin6.sin6_port;
4624 		th->th_sport = sc->sc_dst.sin6.sin6_port;
4625 		break;
4626 #endif
4627 	default:
4628 		th = NULL;
4629 	}
4630 
4631 	th->th_seq = htonl(sc->sc_iss);
4632 	th->th_ack = htonl(sc->sc_irs + 1);
4633 	th->th_off = (sizeof(struct tcphdr) + optlen) >> 2;
4634 	th->th_flags = TH_SYN|TH_ACK;
4635 	th->th_win = htons(sc->sc_win);
4636 	/* th_sum already 0 */
4637 	/* th_urp already 0 */
4638 
4639 	/* Tack on the TCP options. */
4640 	optp = (u_int8_t *)(th + 1);
4641 	*optp++ = TCPOPT_MAXSEG;
4642 	*optp++ = 4;
4643 	*optp++ = (sc->sc_ourmaxseg >> 8) & 0xff;
4644 	*optp++ = sc->sc_ourmaxseg & 0xff;
4645 
4646 	if (sc->sc_request_r_scale != 15) {
4647 		*((u_int32_t *)optp) = htonl(TCPOPT_NOP << 24 |
4648 		    TCPOPT_WINDOW << 16 | TCPOLEN_WINDOW << 8 |
4649 		    sc->sc_request_r_scale);
4650 		optp += 4;
4651 	}
4652 
4653 	if (sc->sc_flags & SCF_TIMESTAMP) {
4654 		u_int32_t *lp = (u_int32_t *)(optp);
4655 		/* Form timestamp option as shown in appendix A of RFC 1323. */
4656 		*lp++ = htonl(TCPOPT_TSTAMP_HDR);
4657 		*lp++ = htonl(SYN_CACHE_TIMESTAMP(sc));
4658 		*lp   = htonl(sc->sc_timestamp);
4659 		optp += TCPOLEN_TSTAMP_APPA;
4660 	}
4661 
4662 	if (sc->sc_flags & SCF_SACK_PERMIT) {
4663 		u_int8_t *p = optp;
4664 
4665 		/* Let the peer know that we will SACK. */
4666 		p[0] = TCPOPT_SACK_PERMITTED;
4667 		p[1] = 2;
4668 		p[2] = TCPOPT_NOP;
4669 		p[3] = TCPOPT_NOP;
4670 		optp += 4;
4671 	}
4672 
4673 	/*
4674 	 * Send ECN SYN-ACK setup packet.
4675 	 * Routes can be asymetric, so, even if we receive a packet
4676 	 * with ECE and CWR set, we must not assume no one will block
4677 	 * the ECE packet we are about to send.
4678 	 */
4679 	if ((sc->sc_flags & SCF_ECN_PERMIT) && tp &&
4680 	    SEQ_GEQ(tp->snd_nxt, tp->snd_max)) {
4681 		th->th_flags |= TH_ECE;
4682 		TCP_STATINC(TCP_STAT_ECN_SHS);
4683 
4684 		/*
4685 		 * draft-ietf-tcpm-ecnsyn-00.txt
4686 		 *
4687 		 * "[...] a TCP node MAY respond to an ECN-setup
4688 		 * SYN packet by setting ECT in the responding
4689 		 * ECN-setup SYN/ACK packet, indicating to routers
4690 		 * that the SYN/ACK packet is ECN-Capable.
4691 		 * This allows a congested router along the path
4692 		 * to mark the packet instead of dropping the
4693 		 * packet as an indication of congestion."
4694 		 *
4695 		 * "[...] There can be a great benefit in setting
4696 		 * an ECN-capable codepoint in SYN/ACK packets [...]
4697 		 * Congestion is  most likely to occur in
4698 		 * the server-to-client direction.  As a result,
4699 		 * setting an ECN-capable codepoint in SYN/ACK
4700 		 * packets can reduce the occurence of three-second
4701 		 * retransmit timeouts resulting from the drop
4702 		 * of SYN/ACK packets."
4703 		 *
4704 		 * Page 4 and 6, January 2006.
4705 		 */
4706 
4707 		switch (sc->sc_src.sa.sa_family) {
4708 #ifdef INET
4709 		case AF_INET:
4710 			ip->ip_tos |= IPTOS_ECN_ECT0;
4711 			break;
4712 #endif
4713 #ifdef INET6
4714 		case AF_INET6:
4715 			ip6->ip6_flow |= htonl(IPTOS_ECN_ECT0 << 20);
4716 			break;
4717 #endif
4718 		}
4719 		TCP_STATINC(TCP_STAT_ECN_ECT);
4720 	}
4721 
4722 #ifdef TCP_SIGNATURE
4723 	if (sc->sc_flags & SCF_SIGNATURE) {
4724 		struct secasvar *sav;
4725 		u_int8_t *sigp;
4726 
4727 		sav = tcp_signature_getsav(m, th);
4728 
4729 		if (sav == NULL) {
4730 			if (m)
4731 				m_freem(m);
4732 			return (EPERM);
4733 		}
4734 
4735 		*optp++ = TCPOPT_SIGNATURE;
4736 		*optp++ = TCPOLEN_SIGNATURE;
4737 		sigp = optp;
4738 		memset(optp, 0, TCP_SIGLEN);
4739 		optp += TCP_SIGLEN;
4740 		*optp++ = TCPOPT_NOP;
4741 		*optp++ = TCPOPT_EOL;
4742 
4743 		(void)tcp_signature(m, th, hlen, sav, sigp);
4744 
4745 		key_sa_recordxfer(sav, m);
4746 		KEY_FREESAV(&sav);
4747 	}
4748 #endif
4749 
4750 	/* Compute the packet's checksum. */
4751 	switch (sc->sc_src.sa.sa_family) {
4752 	case AF_INET:
4753 		ip->ip_len = htons(tlen - hlen);
4754 		th->th_sum = 0;
4755 		th->th_sum = in4_cksum(m, IPPROTO_TCP, hlen, tlen - hlen);
4756 		break;
4757 #ifdef INET6
4758 	case AF_INET6:
4759 		ip6->ip6_plen = htons(tlen - hlen);
4760 		th->th_sum = 0;
4761 		th->th_sum = in6_cksum(m, IPPROTO_TCP, hlen, tlen - hlen);
4762 		break;
4763 #endif
4764 	}
4765 
4766 	/*
4767 	 * Fill in some straggling IP bits.  Note the stack expects
4768 	 * ip_len to be in host order, for convenience.
4769 	 */
4770 	switch (sc->sc_src.sa.sa_family) {
4771 #ifdef INET
4772 	case AF_INET:
4773 		ip->ip_len = htons(tlen);
4774 		ip->ip_ttl = ip_defttl;
4775 		/* XXX tos? */
4776 		break;
4777 #endif
4778 #ifdef INET6
4779 	case AF_INET6:
4780 		ip6->ip6_vfc &= ~IPV6_VERSION_MASK;
4781 		ip6->ip6_vfc |= IPV6_VERSION;
4782 		ip6->ip6_plen = htons(tlen - hlen);
4783 		/* ip6_hlim will be initialized afterwards */
4784 		/* XXX flowlabel? */
4785 		break;
4786 #endif
4787 	}
4788 
4789 	/* XXX use IPsec policy on listening socket, on SYN ACK */
4790 	tp = sc->sc_tp;
4791 
4792 	switch (sc->sc_src.sa.sa_family) {
4793 #ifdef INET
4794 	case AF_INET:
4795 		error = ip_output(m, sc->sc_ipopts, ro,
4796 		    (ip_mtudisc ? IP_MTUDISC : 0),
4797 		    NULL, so);
4798 		break;
4799 #endif
4800 #ifdef INET6
4801 	case AF_INET6:
4802 		ip6->ip6_hlim = in6_selecthlim(NULL,
4803 				(rt = rtcache_validate(ro)) != NULL ? rt->rt_ifp
4804 				                                    : NULL);
4805 
4806 		error = ip6_output(m, NULL /*XXX*/, ro, 0, NULL, so, NULL);
4807 		break;
4808 #endif
4809 	default:
4810 		error = EAFNOSUPPORT;
4811 		break;
4812 	}
4813 	return (error);
4814 }
4815