xref: /netbsd-src/sys/netinet/tcp_usrreq.c (revision aad9773e38ed2370a628a6416e098f9008fc10a7)
1 /*	$NetBSD: tcp_usrreq.c,v 1.202 2014/11/10 18:52:51 maxv Exp $	*/
2 
3 /*
4  * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
5  * All rights reserved.
6  *
7  * Redistribution and use in source and binary forms, with or without
8  * modification, are permitted provided that the following conditions
9  * are met:
10  * 1. Redistributions of source code must retain the above copyright
11  *    notice, this list of conditions and the following disclaimer.
12  * 2. Redistributions in binary form must reproduce the above copyright
13  *    notice, this list of conditions and the following disclaimer in the
14  *    documentation and/or other materials provided with the distribution.
15  * 3. Neither the name of the project nor the names of its contributors
16  *    may be used to endorse or promote products derived from this software
17  *    without specific prior written permission.
18  *
19  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
20  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
23  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29  * SUCH DAMAGE.
30  */
31 
32 /*-
33  * Copyright (c) 1997, 1998, 2005, 2006 The NetBSD Foundation, Inc.
34  * All rights reserved.
35  *
36  * This code is derived from software contributed to The NetBSD Foundation
37  * by Jason R. Thorpe and Kevin M. Lahey of the Numerical Aerospace Simulation
38  * Facility, NASA Ames Research Center.
39  * This code is derived from software contributed to The NetBSD Foundation
40  * by Charles M. Hannum.
41  * This code is derived from software contributed to The NetBSD Foundation
42  * by Rui Paulo.
43  *
44  * Redistribution and use in source and binary forms, with or without
45  * modification, are permitted provided that the following conditions
46  * are met:
47  * 1. Redistributions of source code must retain the above copyright
48  *    notice, this list of conditions and the following disclaimer.
49  * 2. Redistributions in binary form must reproduce the above copyright
50  *    notice, this list of conditions and the following disclaimer in the
51  *    documentation and/or other materials provided with the distribution.
52  *
53  * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
54  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
55  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
56  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
57  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
58  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
59  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
60  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
61  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
62  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
63  * POSSIBILITY OF SUCH DAMAGE.
64  */
65 
66 /*
67  * Copyright (c) 1982, 1986, 1988, 1993, 1995
68  *	The Regents of the University of California.  All rights reserved.
69  *
70  * Redistribution and use in source and binary forms, with or without
71  * modification, are permitted provided that the following conditions
72  * are met:
73  * 1. Redistributions of source code must retain the above copyright
74  *    notice, this list of conditions and the following disclaimer.
75  * 2. Redistributions in binary form must reproduce the above copyright
76  *    notice, this list of conditions and the following disclaimer in the
77  *    documentation and/or other materials provided with the distribution.
78  * 3. Neither the name of the University nor the names of its contributors
79  *    may be used to endorse or promote products derived from this software
80  *    without specific prior written permission.
81  *
82  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
83  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
84  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
85  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
86  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
87  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
88  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
89  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
90  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
91  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
92  * SUCH DAMAGE.
93  *
94  *	@(#)tcp_usrreq.c	8.5 (Berkeley) 6/21/95
95  */
96 
97 /*
98  * TCP protocol interface to socket abstraction.
99  */
100 
101 #include <sys/cdefs.h>
102 __KERNEL_RCSID(0, "$NetBSD: tcp_usrreq.c,v 1.202 2014/11/10 18:52:51 maxv Exp $");
103 
104 #include "opt_inet.h"
105 #include "opt_ipsec.h"
106 #include "opt_tcp_debug.h"
107 #include "opt_mbuftrace.h"
108 
109 #include <sys/param.h>
110 #include <sys/systm.h>
111 #include <sys/kernel.h>
112 #include <sys/mbuf.h>
113 #include <sys/socket.h>
114 #include <sys/socketvar.h>
115 #include <sys/protosw.h>
116 #include <sys/errno.h>
117 #include <sys/stat.h>
118 #include <sys/proc.h>
119 #include <sys/domain.h>
120 #include <sys/sysctl.h>
121 #include <sys/kauth.h>
122 #include <sys/uidinfo.h>
123 
124 #include <net/if.h>
125 #include <net/route.h>
126 
127 #include <netinet/in.h>
128 #include <netinet/in_systm.h>
129 #include <netinet/in_var.h>
130 #include <netinet/ip.h>
131 #include <netinet/in_pcb.h>
132 #include <netinet/ip_var.h>
133 #include <netinet/in_offload.h>
134 
135 #ifdef INET6
136 #ifndef INET
137 #include <netinet/in.h>
138 #endif
139 #include <netinet/ip6.h>
140 #include <netinet6/in6_pcb.h>
141 #include <netinet6/ip6_var.h>
142 #include <netinet6/scope6_var.h>
143 #endif
144 
145 #include <netinet/tcp.h>
146 #include <netinet/tcp_fsm.h>
147 #include <netinet/tcp_seq.h>
148 #include <netinet/tcp_timer.h>
149 #include <netinet/tcp_var.h>
150 #include <netinet/tcp_private.h>
151 #include <netinet/tcp_congctl.h>
152 #include <netinet/tcpip.h>
153 #include <netinet/tcp_debug.h>
154 #include <netinet/tcp_vtw.h>
155 
156 #include "opt_tcp_space.h"
157 
158 static int
159 tcp_debug_capture(struct tcpcb *tp, int req)
160 {
161 #ifdef KPROF
162 	tcp_acounts[tp->t_state][req]++;
163 #endif
164 #ifdef TCP_DEBUG
165 	return tp->t_state;
166 #endif
167 	return 0;
168 }
169 
170 static inline void
171 tcp_debug_trace(struct socket *so, struct tcpcb *tp, int ostate, int req)
172 {
173 #ifdef TCP_DEBUG
174 	if (tp && (so->so_options & SO_DEBUG))
175 		tcp_trace(TA_USER, ostate, tp, NULL, req);
176 #endif
177 }
178 
179 static int
180 tcp_getpcb(struct socket *so, struct inpcb **inp,
181     struct in6pcb **in6p, struct tcpcb **tp)
182 {
183 
184 	KASSERT(solocked(so));
185 
186 	/*
187 	 * When a TCP is attached to a socket, then there will be
188 	 * a (struct inpcb) pointed at by the socket, and this
189 	 * structure will point at a subsidary (struct tcpcb).
190 	 */
191 	switch (so->so_proto->pr_domain->dom_family) {
192 #ifdef INET
193 	case PF_INET:
194 		*inp = sotoinpcb(so);
195 		if (*inp == NULL)
196 			return EINVAL;
197 		*tp = intotcpcb(*inp);
198 		break;
199 #endif
200 #ifdef INET6
201 	case PF_INET6:
202 		*in6p = sotoin6pcb(so);
203 		if (*in6p == NULL)
204 			return EINVAL;
205 		*tp = in6totcpcb(*in6p);
206 		break;
207 #endif
208 	default:
209 		return EAFNOSUPPORT;
210 	}
211 
212 	KASSERT(tp != NULL);
213 
214 	return 0;
215 }
216 
217 /*
218  * Process a TCP user request for TCP tb.  If this is a send request
219  * then m is the mbuf chain of send data.  If this is a timer expiration
220  * (called from the software clock routine), then timertype tells which timer.
221  */
222 static int
223 tcp_usrreq(struct socket *so, int req, struct mbuf *m, struct mbuf *nam,
224     struct mbuf *control, struct lwp *l)
225 {
226 	KASSERT(req != PRU_ATTACH);
227 	KASSERT(req != PRU_DETACH);
228 	KASSERT(req != PRU_ACCEPT);
229 	KASSERT(req != PRU_BIND);
230 	KASSERT(req != PRU_LISTEN);
231 	KASSERT(req != PRU_CONNECT);
232 	KASSERT(req != PRU_CONNECT2);
233 	KASSERT(req != PRU_DISCONNECT);
234 	KASSERT(req != PRU_SHUTDOWN);
235 	KASSERT(req != PRU_ABORT);
236 	KASSERT(req != PRU_CONTROL);
237 	KASSERT(req != PRU_SENSE);
238 	KASSERT(req != PRU_PEERADDR);
239 	KASSERT(req != PRU_SOCKADDR);
240 	KASSERT(req != PRU_RCVD);
241 	KASSERT(req != PRU_RCVOOB);
242 	KASSERT(req != PRU_SEND);
243 	KASSERT(req != PRU_SENDOOB);
244 	KASSERT(req != PRU_PURGEIF);
245 
246 	KASSERT(solocked(so));
247 
248 	panic("tcp_usrreq");
249 
250 	return 0;
251 }
252 
253 static void
254 change_keepalive(struct socket *so, struct tcpcb *tp)
255 {
256 	tp->t_maxidle = tp->t_keepcnt * tp->t_keepintvl;
257 	TCP_TIMER_DISARM(tp, TCPT_KEEP);
258 	TCP_TIMER_DISARM(tp, TCPT_2MSL);
259 
260 	if (tp->t_state == TCPS_SYN_RECEIVED ||
261 	    tp->t_state == TCPS_SYN_SENT) {
262 		TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepinit);
263 	} else if (so->so_options & SO_KEEPALIVE &&
264 	    tp->t_state <= TCPS_CLOSE_WAIT) {
265 		TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepintvl);
266 	} else {
267 		TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepidle);
268 	}
269 
270 	if ((tp->t_state == TCPS_FIN_WAIT_2) && (tp->t_maxidle > 0))
271 		TCP_TIMER_ARM(tp, TCPT_2MSL, tp->t_maxidle);
272 }
273 
274 int
275 tcp_ctloutput(int op, struct socket *so, struct sockopt *sopt)
276 {
277 	int error = 0, s;
278 	struct inpcb *inp;
279 #ifdef INET6
280 	struct in6pcb *in6p;
281 #endif
282 	struct tcpcb *tp;
283 	u_int ui;
284 	int family;	/* family of the socket */
285 	int level, optname, optval;
286 
287 	level = sopt->sopt_level;
288 	optname = sopt->sopt_name;
289 
290 	family = so->so_proto->pr_domain->dom_family;
291 
292 	s = splsoftnet();
293 	switch (family) {
294 #ifdef INET
295 	case PF_INET:
296 		inp = sotoinpcb(so);
297 #ifdef INET6
298 		in6p = NULL;
299 #endif
300 		break;
301 #endif
302 #ifdef INET6
303 	case PF_INET6:
304 		inp = NULL;
305 		in6p = sotoin6pcb(so);
306 		break;
307 #endif
308 	default:
309 		splx(s);
310 		panic("%s: af %d", __func__, family);
311 	}
312 #ifndef INET6
313 	if (inp == NULL)
314 #else
315 	if (inp == NULL && in6p == NULL)
316 #endif
317 	{
318 		splx(s);
319 		return (ECONNRESET);
320 	}
321 	if (level != IPPROTO_TCP) {
322 		switch (family) {
323 #ifdef INET
324 		case PF_INET:
325 			error = ip_ctloutput(op, so, sopt);
326 			break;
327 #endif
328 #ifdef INET6
329 		case PF_INET6:
330 			error = ip6_ctloutput(op, so, sopt);
331 			break;
332 #endif
333 		}
334 		splx(s);
335 		return (error);
336 	}
337 	if (inp)
338 		tp = intotcpcb(inp);
339 #ifdef INET6
340 	else if (in6p)
341 		tp = in6totcpcb(in6p);
342 #endif
343 	else
344 		tp = NULL;
345 
346 	switch (op) {
347 	case PRCO_SETOPT:
348 		switch (optname) {
349 #ifdef TCP_SIGNATURE
350 		case TCP_MD5SIG:
351 			error = sockopt_getint(sopt, &optval);
352 			if (error)
353 				break;
354 			if (optval > 0)
355 				tp->t_flags |= TF_SIGNATURE;
356 			else
357 				tp->t_flags &= ~TF_SIGNATURE;
358 			break;
359 #endif /* TCP_SIGNATURE */
360 
361 		case TCP_NODELAY:
362 			error = sockopt_getint(sopt, &optval);
363 			if (error)
364 				break;
365 			if (optval)
366 				tp->t_flags |= TF_NODELAY;
367 			else
368 				tp->t_flags &= ~TF_NODELAY;
369 			break;
370 
371 		case TCP_MAXSEG:
372 			error = sockopt_getint(sopt, &optval);
373 			if (error)
374 				break;
375 			if (optval > 0 && optval <= tp->t_peermss)
376 				tp->t_peermss = optval; /* limit on send size */
377 			else
378 				error = EINVAL;
379 			break;
380 #ifdef notyet
381 		case TCP_CONGCTL:
382 			/* XXX string overflow XXX */
383 			error = tcp_congctl_select(tp, sopt->sopt_data);
384 			break;
385 #endif
386 
387 		case TCP_KEEPIDLE:
388 			error = sockopt_get(sopt, &ui, sizeof(ui));
389 			if (error)
390 				break;
391 			if (ui > 0) {
392 				tp->t_keepidle = ui;
393 				change_keepalive(so, tp);
394 			} else
395 				error = EINVAL;
396 			break;
397 
398 		case TCP_KEEPINTVL:
399 			error = sockopt_get(sopt, &ui, sizeof(ui));
400 			if (error)
401 				break;
402 			if (ui > 0) {
403 				tp->t_keepintvl = ui;
404 				change_keepalive(so, tp);
405 			} else
406 				error = EINVAL;
407 			break;
408 
409 		case TCP_KEEPCNT:
410 			error = sockopt_get(sopt, &ui, sizeof(ui));
411 			if (error)
412 				break;
413 			if (ui > 0) {
414 				tp->t_keepcnt = ui;
415 				change_keepalive(so, tp);
416 			} else
417 				error = EINVAL;
418 			break;
419 
420 		case TCP_KEEPINIT:
421 			error = sockopt_get(sopt, &ui, sizeof(ui));
422 			if (error)
423 				break;
424 			if (ui > 0) {
425 				tp->t_keepinit = ui;
426 				change_keepalive(so, tp);
427 			} else
428 				error = EINVAL;
429 			break;
430 
431 		default:
432 			error = ENOPROTOOPT;
433 			break;
434 		}
435 		break;
436 
437 	case PRCO_GETOPT:
438 		switch (optname) {
439 #ifdef TCP_SIGNATURE
440 		case TCP_MD5SIG:
441 			optval = (tp->t_flags & TF_SIGNATURE) ? 1 : 0;
442 			error = sockopt_set(sopt, &optval, sizeof(optval));
443 			break;
444 #endif
445 		case TCP_NODELAY:
446 			optval = tp->t_flags & TF_NODELAY;
447 			error = sockopt_set(sopt, &optval, sizeof(optval));
448 			break;
449 		case TCP_MAXSEG:
450 			optval = tp->t_peermss;
451 			error = sockopt_set(sopt, &optval, sizeof(optval));
452 			break;
453 #ifdef notyet
454 		case TCP_CONGCTL:
455 			break;
456 #endif
457 		default:
458 			error = ENOPROTOOPT;
459 			break;
460 		}
461 		break;
462 	}
463 	splx(s);
464 	return (error);
465 }
466 
467 #ifndef TCP_SENDSPACE
468 #define	TCP_SENDSPACE	1024*32
469 #endif
470 int	tcp_sendspace = TCP_SENDSPACE;
471 #ifndef TCP_RECVSPACE
472 #define	TCP_RECVSPACE	1024*32
473 #endif
474 int	tcp_recvspace = TCP_RECVSPACE;
475 
476 /*
477  * tcp_attach: attach TCP protocol to socket, allocating internet protocol
478  * control block, TCP control block, buffer space and entering LISTEN state
479  * if to accept connections.
480  */
481 static int
482 tcp_attach(struct socket *so, int proto)
483 {
484 	struct tcpcb *tp;
485 	struct inpcb *inp;
486 #ifdef INET6
487 	struct in6pcb *in6p;
488 #endif
489 	int s, error, family;
490 
491 	/* Assign the lock (must happen even if we will error out). */
492 	s = splsoftnet();
493 	sosetlock(so);
494 	KASSERT(solocked(so));
495 
496 	family = so->so_proto->pr_domain->dom_family;
497 	switch (family) {
498 #ifdef INET
499 	case PF_INET:
500 		inp = sotoinpcb(so);
501 #ifdef INET6
502 		in6p = NULL;
503 #endif
504 		break;
505 #endif
506 #ifdef INET6
507 	case PF_INET6:
508 		inp = NULL;
509 		in6p = sotoin6pcb(so);
510 		break;
511 #endif
512 	default:
513 		error = EAFNOSUPPORT;
514 		goto out;
515 	}
516 
517 	KASSERT(inp == NULL);
518 #ifdef INET6
519 	KASSERT(in6p == NULL);
520 #endif
521 
522 #ifdef MBUFTRACE
523 	so->so_mowner = &tcp_sock_mowner;
524 	so->so_rcv.sb_mowner = &tcp_sock_rx_mowner;
525 	so->so_snd.sb_mowner = &tcp_sock_tx_mowner;
526 #endif
527 	if (so->so_snd.sb_hiwat == 0 || so->so_rcv.sb_hiwat == 0) {
528 		error = soreserve(so, tcp_sendspace, tcp_recvspace);
529 		if (error)
530 			goto out;
531 	}
532 
533 	so->so_rcv.sb_flags |= SB_AUTOSIZE;
534 	so->so_snd.sb_flags |= SB_AUTOSIZE;
535 
536 	switch (family) {
537 #ifdef INET
538 	case PF_INET:
539 		error = in_pcballoc(so, &tcbtable);
540 		if (error)
541 			goto out;
542 		inp = sotoinpcb(so);
543 #ifdef INET6
544 		in6p = NULL;
545 #endif
546 		break;
547 #endif
548 #ifdef INET6
549 	case PF_INET6:
550 		error = in6_pcballoc(so, &tcbtable);
551 		if (error)
552 			goto out;
553 		inp = NULL;
554 		in6p = sotoin6pcb(so);
555 		break;
556 #endif
557 	default:
558 		error = EAFNOSUPPORT;
559 		goto out;
560 	}
561 	if (inp)
562 		tp = tcp_newtcpcb(family, (void *)inp);
563 #ifdef INET6
564 	else if (in6p)
565 		tp = tcp_newtcpcb(family, (void *)in6p);
566 #endif
567 	else
568 		tp = NULL;
569 
570 	if (tp == NULL) {
571 		int nofd = so->so_state & SS_NOFDREF;	/* XXX */
572 
573 		so->so_state &= ~SS_NOFDREF;	/* don't free the socket yet */
574 #ifdef INET
575 		if (inp)
576 			in_pcbdetach(inp);
577 #endif
578 #ifdef INET6
579 		if (in6p)
580 			in6_pcbdetach(in6p);
581 #endif
582 		so->so_state |= nofd;
583 		error = ENOBUFS;
584 		goto out;
585 	}
586 	tp->t_state = TCPS_CLOSED;
587 	if ((so->so_options & SO_LINGER) && so->so_linger == 0) {
588 		so->so_linger = TCP_LINGERTIME;
589 	}
590 out:
591 	KASSERT(solocked(so));
592 	splx(s);
593 	return error;
594 }
595 
596 static void
597 tcp_detach(struct socket *so)
598 {
599 	struct inpcb *inp = NULL;
600 	struct in6pcb *in6p = NULL;
601 	struct tcpcb *tp = NULL;
602 	int s;
603 
604 	if (tcp_getpcb(so, &inp, &in6p, &tp) != 0)
605 		return;
606 
607 	s = splsoftnet();
608 	(void)tcp_disconnect1(tp);
609 	splx(s);
610 }
611 
612 static int
613 tcp_accept(struct socket *so, struct mbuf *nam)
614 {
615 	struct inpcb *inp = NULL;
616 	struct in6pcb *in6p = NULL;
617 	struct tcpcb *tp = NULL;
618 	int ostate = 0;
619 	int error = 0;
620 	int s;
621 
622 	if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0)
623 		return error;
624 
625 	ostate = tcp_debug_capture(tp, PRU_ACCEPT);
626 
627 	/*
628 	 * Accept a connection.  Essentially all the work is
629 	 * done at higher levels; just return the address
630 	 * of the peer, storing through addr.
631 	 */
632 	s = splsoftnet();
633 #ifdef INET
634 	if (inp) {
635 		in_setpeeraddr(inp, nam);
636 	}
637 #endif
638 #ifdef INET6
639 	if (in6p) {
640 		in6_setpeeraddr(in6p, nam);
641 	}
642 #endif
643 	tcp_debug_trace(so, tp, ostate, PRU_ACCEPT);
644 	splx(s);
645 
646 	return 0;
647 }
648 
649 static int
650 tcp_bind(struct socket *so, struct mbuf *nam, struct lwp *l)
651 {
652 	struct inpcb *inp = NULL;
653 	struct in6pcb *in6p = NULL;
654 	struct tcpcb *tp = NULL;
655 	int s;
656 	int error = 0;
657 	int ostate = 0;
658 
659 	if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0)
660 		return error;
661 
662 	ostate = tcp_debug_capture(tp, PRU_BIND);
663 
664 	/*
665 	 * Give the socket an address.
666 	 */
667 	s = splsoftnet();
668 	switch (so->so_proto->pr_domain->dom_family) {
669 #ifdef INET
670 	case PF_INET:
671 		error = in_pcbbind(inp, nam, l);
672 		break;
673 #endif
674 #ifdef INET6
675 	case PF_INET6:
676 		error = in6_pcbbind(in6p, nam, l);
677 		if (!error) {
678 			/* mapped addr case */
679 			if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_laddr))
680 				tp->t_family = AF_INET;
681 			else
682 				tp->t_family = AF_INET6;
683 		}
684 		break;
685 #endif
686 	}
687 	tcp_debug_trace(so, tp, ostate, PRU_BIND);
688 	splx(s);
689 
690 	return error;
691 }
692 
693 static int
694 tcp_listen(struct socket *so, struct lwp *l)
695 {
696 	struct inpcb *inp = NULL;
697 	struct in6pcb *in6p = NULL;
698 	struct tcpcb *tp = NULL;
699 	int error = 0;
700 	int ostate = 0;
701 	int s;
702 
703 	if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0)
704 		return error;
705 
706 	ostate = tcp_debug_capture(tp, PRU_LISTEN);
707 
708 	/*
709 	 * Prepare to accept connections.
710 	 */
711 	s = splsoftnet();
712 #ifdef INET
713 	if (inp && inp->inp_lport == 0) {
714 		error = in_pcbbind(inp, NULL, l);
715 		if (error)
716 			goto release;
717 	}
718 #endif
719 #ifdef INET6
720 	if (in6p && in6p->in6p_lport == 0) {
721 		error = in6_pcbbind(in6p, NULL, l);
722 		if (error)
723 			goto release;
724 	}
725 #endif
726 	tp->t_state = TCPS_LISTEN;
727 
728 release:
729 	tcp_debug_trace(so, tp, ostate, PRU_LISTEN);
730 	splx(s);
731 
732 	return error;
733 }
734 
735 static int
736 tcp_connect(struct socket *so, struct mbuf *nam, struct lwp *l)
737 {
738 	struct inpcb *inp = NULL;
739 	struct in6pcb *in6p = NULL;
740 	struct tcpcb *tp = NULL;
741 	int s;
742 	int error = 0;
743 	int ostate = 0;
744 
745 	if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0)
746 		return error;
747 
748 	ostate = tcp_debug_capture(tp, PRU_CONNECT);
749 
750 	/*
751 	 * Initiate connection to peer.
752 	 * Create a template for use in transmissions on this connection.
753 	 * Enter SYN_SENT state, and mark socket as connecting.
754 	 * Start keep-alive timer, and seed output sequence space.
755 	 * Send initial segment on connection.
756 	 */
757 	s = splsoftnet();
758 #ifdef INET
759 	if (inp) {
760 		if (inp->inp_lport == 0) {
761 			error = in_pcbbind(inp, NULL, l);
762 			if (error)
763 				goto release;
764 		}
765 		error = in_pcbconnect(inp, nam, l);
766 	}
767 #endif
768 #ifdef INET6
769 	if (in6p) {
770 		if (in6p->in6p_lport == 0) {
771 			error = in6_pcbbind(in6p, NULL, l);
772 			if (error)
773 				goto release;
774 		}
775 		error = in6_pcbconnect(in6p, nam, l);
776 		if (!error) {
777 			/* mapped addr case */
778 			if (IN6_IS_ADDR_V4MAPPED(&in6p->in6p_faddr))
779 				tp->t_family = AF_INET;
780 			else
781 				tp->t_family = AF_INET6;
782 		}
783 	}
784 #endif
785 	if (error)
786 		goto release;
787 	tp->t_template = tcp_template(tp);
788 	if (tp->t_template == 0) {
789 #ifdef INET
790 		if (inp)
791 			in_pcbdisconnect(inp);
792 #endif
793 #ifdef INET6
794 		if (in6p)
795 			in6_pcbdisconnect(in6p);
796 #endif
797 		error = ENOBUFS;
798 		goto release;
799 	}
800 	/*
801 	 * Compute window scaling to request.
802 	 * XXX: This should be moved to tcp_output().
803 	 */
804 	while (tp->request_r_scale < TCP_MAX_WINSHIFT &&
805 	    (TCP_MAXWIN << tp->request_r_scale) < sb_max)
806 		tp->request_r_scale++;
807 	soisconnecting(so);
808 	TCP_STATINC(TCP_STAT_CONNATTEMPT);
809 	tp->t_state = TCPS_SYN_SENT;
810 	TCP_TIMER_ARM(tp, TCPT_KEEP, tp->t_keepinit);
811 	tp->iss = tcp_new_iss(tp, 0);
812 	tcp_sendseqinit(tp);
813 	error = tcp_output(tp);
814 
815 release:
816 	tcp_debug_trace(so, tp, ostate, PRU_CONNECT);
817 	splx(s);
818 
819 	return error;
820 }
821 
822 static int
823 tcp_connect2(struct socket *so, struct socket *so2)
824 {
825 	struct inpcb *inp = NULL;
826 	struct in6pcb *in6p = NULL;
827 	struct tcpcb *tp = NULL;
828 	int error = 0;
829 	int ostate = 0;
830 
831 	KASSERT(solocked(so));
832 
833 	if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0)
834 		return error;
835 
836 	ostate = tcp_debug_capture(tp, PRU_CONNECT2);
837 
838 	tcp_debug_trace(so, tp, ostate, PRU_CONNECT2);
839 
840 	return EOPNOTSUPP;
841 }
842 
843 static int
844 tcp_disconnect(struct socket *so)
845 {
846 	struct inpcb *inp = NULL;
847 	struct in6pcb *in6p = NULL;
848 	struct tcpcb *tp = NULL;
849 	int error = 0;
850 	int ostate = 0;
851 	int s;
852 
853 	if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0)
854 		return error;
855 
856 	ostate = tcp_debug_capture(tp, PRU_DISCONNECT);
857 
858 	/*
859 	 * Initiate disconnect from peer.
860 	 * If connection never passed embryonic stage, just drop;
861 	 * else if don't need to let data drain, then can just drop anyways,
862 	 * else have to begin TCP shutdown process: mark socket disconnecting,
863 	 * drain unread data, state switch to reflect user close, and
864 	 * send segment (e.g. FIN) to peer.  Socket will be really disconnected
865 	 * when peer sends FIN and acks ours.
866 	 *
867 	 * SHOULD IMPLEMENT LATER PRU_CONNECT VIA REALLOC TCPCB.
868 	 */
869 	s = splsoftnet();
870 	tp = tcp_disconnect1(tp);
871 	tcp_debug_trace(so, tp, ostate, PRU_DISCONNECT);
872 	splx(s);
873 
874 	return error;
875 }
876 
877 static int
878 tcp_shutdown(struct socket *so)
879 {
880 	struct inpcb *inp = NULL;
881 	struct in6pcb *in6p = NULL;
882 	struct tcpcb *tp = NULL;
883 	int error = 0;
884 	int ostate = 0;
885 	int s;
886 
887 	if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0)
888 		return error;
889 
890 	ostate = tcp_debug_capture(tp, PRU_SHUTDOWN);
891 	/*
892 	 * Mark the connection as being incapable of further output.
893 	 */
894 	s = splsoftnet();
895 	socantsendmore(so);
896 	tp = tcp_usrclosed(tp);
897 	if (tp)
898 		error = tcp_output(tp);
899 	tcp_debug_trace(so, tp, ostate, PRU_SHUTDOWN);
900 	splx(s);
901 
902 	return error;
903 }
904 
905 static int
906 tcp_abort(struct socket *so)
907 {
908 	struct inpcb *inp = NULL;
909 	struct in6pcb *in6p = NULL;
910 	struct tcpcb *tp = NULL;
911 	int error = 0;
912 	int ostate = 0;
913 	int s;
914 
915 	if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0)
916 		return error;
917 
918 	ostate = tcp_debug_capture(tp, PRU_ABORT);
919 
920 	/*
921 	 * Abort the TCP.
922 	 */
923 	s = splsoftnet();
924 	tp = tcp_drop(tp, ECONNABORTED);
925 	tcp_debug_trace(so, tp, ostate, PRU_ABORT);
926 	splx(s);
927 
928 	return error;
929 }
930 
931 static int
932 tcp_ioctl(struct socket *so, u_long cmd, void *nam, struct ifnet *ifp)
933 {
934 	switch (so->so_proto->pr_domain->dom_family) {
935 #ifdef INET
936 	case PF_INET:
937 		return in_control(so, cmd, nam, ifp);
938 #endif
939 #ifdef INET6
940 	case PF_INET6:
941 		return in6_control(so, cmd, nam, ifp);
942 #endif
943 	default:
944 		return EAFNOSUPPORT;
945 	}
946 }
947 
948 static int
949 tcp_stat(struct socket *so, struct stat *ub)
950 {
951 	KASSERT(solocked(so));
952 
953 	/* stat: don't bother with a blocksize.  */
954 	return 0;
955 }
956 
957 static int
958 tcp_peeraddr(struct socket *so, struct mbuf *nam)
959 {
960 	struct inpcb *inp = NULL;
961 	struct in6pcb *in6p = NULL;
962 	struct tcpcb *tp = NULL;
963 	int ostate = 0;
964 	int error = 0;
965 	int s;
966 
967 	if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0)
968 		return error;
969 
970 	ostate = tcp_debug_capture(tp, PRU_PEERADDR);
971 
972 	s = splsoftnet();
973 #ifdef INET
974 	if (inp)
975 		in_setpeeraddr(inp, nam);
976 #endif
977 #ifdef INET6
978 	if (in6p)
979 		in6_setpeeraddr(in6p, nam);
980 #endif
981 	tcp_debug_trace(so, tp, ostate, PRU_PEERADDR);
982 	splx(s);
983 
984 	return 0;
985 }
986 
987 static int
988 tcp_sockaddr(struct socket *so, struct mbuf *nam)
989 {
990 	struct inpcb *inp = NULL;
991 	struct in6pcb *in6p = NULL;
992 	struct tcpcb *tp = NULL;
993 	int ostate = 0;
994 	int error = 0;
995 	int s;
996 
997 	if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0)
998 		return error;
999 
1000 	ostate = tcp_debug_capture(tp, PRU_SOCKADDR);
1001 
1002 	s = splsoftnet();
1003 #ifdef INET
1004 	if (inp)
1005 		in_setsockaddr(inp, nam);
1006 #endif
1007 #ifdef INET6
1008 	if (in6p)
1009 		in6_setsockaddr(in6p, nam);
1010 #endif
1011 	tcp_debug_trace(so, tp, ostate, PRU_SOCKADDR);
1012 	splx(s);
1013 
1014 	return 0;
1015 }
1016 
1017 static int
1018 tcp_rcvd(struct socket *so, int flags, struct lwp *l)
1019 {
1020 	struct inpcb *inp = NULL;
1021 	struct in6pcb *in6p = NULL;
1022 	struct tcpcb *tp = NULL;
1023 	int ostate = 0;
1024 	int error = 0;
1025 	int s;
1026 
1027 	if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0)
1028 		return error;
1029 
1030 	ostate = tcp_debug_capture(tp, PRU_RCVD);
1031 
1032 	/*
1033 	 * After a receive, possibly send window update to peer.
1034 	 *
1035 	 * soreceive() calls this function when a user receives
1036 	 * ancillary data on a listening socket. We don't call
1037 	 * tcp_output in such a case, since there is no header
1038 	 * template for a listening socket and hence the kernel
1039 	 * will panic.
1040 	 */
1041 	s = splsoftnet();
1042 	if ((so->so_state & (SS_ISCONNECTED|SS_ISCONNECTING)) != 0)
1043 		(void) tcp_output(tp);
1044 	splx(s);
1045 
1046 	tcp_debug_trace(so, tp, ostate, PRU_RCVD);
1047 
1048 	return 0;
1049 }
1050 
1051 static int
1052 tcp_recvoob(struct socket *so, struct mbuf *m, int flags)
1053 {
1054 	struct inpcb *inp = NULL;
1055 	struct in6pcb *in6p = NULL;
1056 	struct tcpcb *tp = NULL;
1057 	int ostate = 0;
1058 	int error = 0;
1059 	int s;
1060 
1061 	if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0)
1062 		return error;
1063 
1064 	ostate = tcp_debug_capture(tp, PRU_RCVOOB);
1065 
1066 	s = splsoftnet();
1067 	if ((so->so_oobmark == 0 &&
1068 	    (so->so_state & SS_RCVATMARK) == 0) ||
1069 	    so->so_options & SO_OOBINLINE ||
1070 	    tp->t_oobflags & TCPOOB_HADDATA) {
1071 		splx(s);
1072 		return EINVAL;
1073 	}
1074 
1075 	if ((tp->t_oobflags & TCPOOB_HAVEDATA) == 0) {
1076 		splx(s);
1077 		return EWOULDBLOCK;
1078 	}
1079 
1080 	m->m_len = 1;
1081 	*mtod(m, char *) = tp->t_iobc;
1082 	if ((flags & MSG_PEEK) == 0)
1083 		tp->t_oobflags ^= (TCPOOB_HAVEDATA | TCPOOB_HADDATA);
1084 
1085 	tcp_debug_trace(so, tp, ostate, PRU_RCVOOB);
1086 	splx(s);
1087 
1088 	return 0;
1089 }
1090 
1091 static int
1092 tcp_send(struct socket *so, struct mbuf *m, struct mbuf *nam,
1093     struct mbuf *control, struct lwp *l)
1094 {
1095 	struct inpcb *inp = NULL;
1096 	struct in6pcb *in6p = NULL;
1097 	struct tcpcb *tp = NULL;
1098 	int ostate = 0;
1099 	int error = 0;
1100 	int s;
1101 
1102 	if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0)
1103 		return error;
1104 
1105 	ostate = tcp_debug_capture(tp, PRU_SEND);
1106 
1107 	/*
1108 	 * Do a send by putting data in output queue and updating urgent
1109 	 * marker if URG set.  Possibly send more data.
1110 	 */
1111 	s = splsoftnet();
1112 	if (control && control->m_len) {
1113 		m_freem(control);
1114 		m_freem(m);
1115 		tcp_debug_trace(so, tp, ostate, PRU_SEND);
1116 		splx(s);
1117 		return EINVAL;
1118 	}
1119 
1120 	sbappendstream(&so->so_snd, m);
1121 	error = tcp_output(tp);
1122 	tcp_debug_trace(so, tp, ostate, PRU_SEND);
1123 	splx(s);
1124 
1125 	return error;
1126 }
1127 
1128 static int
1129 tcp_sendoob(struct socket *so, struct mbuf *m, struct mbuf *control)
1130 {
1131 	struct inpcb *inp = NULL;
1132 	struct in6pcb *in6p = NULL;
1133 	struct tcpcb *tp = NULL;
1134 	int ostate = 0;
1135 	int error = 0;
1136 	int s;
1137 
1138 	if ((error = tcp_getpcb(so, &inp, &in6p, &tp)) != 0)
1139 		return error;
1140 
1141 	ostate = tcp_debug_capture(tp, PRU_SENDOOB);
1142 
1143 	s = splsoftnet();
1144 	if (sbspace(&so->so_snd) < -512) {
1145 		m_freem(m);
1146 		splx(s);
1147 		return ENOBUFS;
1148 	}
1149 	/*
1150 	 * According to RFC961 (Assigned Protocols),
1151 	 * the urgent pointer points to the last octet
1152 	 * of urgent data.  We continue, however,
1153 	 * to consider it to indicate the first octet
1154 	 * of data past the urgent section.
1155 	 * Otherwise, snd_up should be one lower.
1156 	 */
1157 	sbappendstream(&so->so_snd, m);
1158 	tp->snd_up = tp->snd_una + so->so_snd.sb_cc;
1159 	tp->t_force = 1;
1160 	error = tcp_output(tp);
1161 	tp->t_force = 0;
1162 	tcp_debug_trace(so, tp, ostate, PRU_SENDOOB);
1163 	splx(s);
1164 
1165 	return error;
1166 }
1167 
1168 static int
1169 tcp_purgeif(struct socket *so, struct ifnet *ifp)
1170 {
1171 	int s;
1172 
1173 	s = splsoftnet();
1174 	mutex_enter(softnet_lock);
1175 	switch (so->so_proto->pr_domain->dom_family) {
1176 #ifdef INET
1177 	case PF_INET:
1178 		in_pcbpurgeif0(&tcbtable, ifp);
1179 		in_purgeif(ifp);
1180 		in_pcbpurgeif(&tcbtable, ifp);
1181 		break;
1182 #endif
1183 #ifdef INET6
1184 	case PF_INET6:
1185 		in6_pcbpurgeif0(&tcbtable, ifp);
1186 		in6_purgeif(ifp);
1187 		in6_pcbpurgeif(&tcbtable, ifp);
1188 		break;
1189 #endif
1190 	default:
1191 		mutex_exit(softnet_lock);
1192 		splx(s);
1193 		return EAFNOSUPPORT;
1194 	}
1195 	mutex_exit(softnet_lock);
1196 	splx(s);
1197 
1198 	return 0;
1199 }
1200 
1201 /*
1202  * Initiate (or continue) disconnect.
1203  * If embryonic state, just send reset (once).
1204  * If in ``let data drain'' option and linger null, just drop.
1205  * Otherwise (hard), mark socket disconnecting and drop
1206  * current input data; switch states based on user close, and
1207  * send segment to peer (with FIN).
1208  */
1209 struct tcpcb *
1210 tcp_disconnect1(struct tcpcb *tp)
1211 {
1212 	struct socket *so;
1213 
1214 	if (tp->t_inpcb)
1215 		so = tp->t_inpcb->inp_socket;
1216 #ifdef INET6
1217 	else if (tp->t_in6pcb)
1218 		so = tp->t_in6pcb->in6p_socket;
1219 #endif
1220 	else
1221 		so = NULL;
1222 
1223 	if (TCPS_HAVEESTABLISHED(tp->t_state) == 0)
1224 		tp = tcp_close(tp);
1225 	else if ((so->so_options & SO_LINGER) && so->so_linger == 0)
1226 		tp = tcp_drop(tp, 0);
1227 	else {
1228 		soisdisconnecting(so);
1229 		sbflush(&so->so_rcv);
1230 		tp = tcp_usrclosed(tp);
1231 		if (tp)
1232 			(void) tcp_output(tp);
1233 	}
1234 	return (tp);
1235 }
1236 
1237 /*
1238  * User issued close, and wish to trail through shutdown states:
1239  * if never received SYN, just forget it.  If got a SYN from peer,
1240  * but haven't sent FIN, then go to FIN_WAIT_1 state to send peer a FIN.
1241  * If already got a FIN from peer, then almost done; go to LAST_ACK
1242  * state.  In all other cases, have already sent FIN to peer (e.g.
1243  * after PRU_SHUTDOWN), and just have to play tedious game waiting
1244  * for peer to send FIN or not respond to keep-alives, etc.
1245  * We can let the user exit from the close as soon as the FIN is acked.
1246  */
1247 struct tcpcb *
1248 tcp_usrclosed(struct tcpcb *tp)
1249 {
1250 
1251 	switch (tp->t_state) {
1252 
1253 	case TCPS_CLOSED:
1254 	case TCPS_LISTEN:
1255 	case TCPS_SYN_SENT:
1256 		tp->t_state = TCPS_CLOSED;
1257 		tp = tcp_close(tp);
1258 		break;
1259 
1260 	case TCPS_SYN_RECEIVED:
1261 	case TCPS_ESTABLISHED:
1262 		tp->t_state = TCPS_FIN_WAIT_1;
1263 		break;
1264 
1265 	case TCPS_CLOSE_WAIT:
1266 		tp->t_state = TCPS_LAST_ACK;
1267 		break;
1268 	}
1269 	if (tp && tp->t_state >= TCPS_FIN_WAIT_2) {
1270 		struct socket *so;
1271 		if (tp->t_inpcb)
1272 			so = tp->t_inpcb->inp_socket;
1273 #ifdef INET6
1274 		else if (tp->t_in6pcb)
1275 			so = tp->t_in6pcb->in6p_socket;
1276 #endif
1277 		else
1278 			so = NULL;
1279 		if (so)
1280 			soisdisconnected(so);
1281 		/*
1282 		 * If we are in FIN_WAIT_2, we arrived here because the
1283 		 * application did a shutdown of the send side.  Like the
1284 		 * case of a transition from FIN_WAIT_1 to FIN_WAIT_2 after
1285 		 * a full close, we start a timer to make sure sockets are
1286 		 * not left in FIN_WAIT_2 forever.
1287 		 */
1288 		if ((tp->t_state == TCPS_FIN_WAIT_2) && (tp->t_maxidle > 0))
1289 			TCP_TIMER_ARM(tp, TCPT_2MSL, tp->t_maxidle);
1290 		else if (tp->t_state == TCPS_TIME_WAIT
1291 			 && ((tp->t_inpcb
1292 			      && (tcp4_vtw_enable & 1)
1293 			      && vtw_add(AF_INET, tp))
1294 			     ||
1295 			     (tp->t_in6pcb
1296 			      && (tcp6_vtw_enable & 1)
1297 			      && vtw_add(AF_INET6, tp)))) {
1298 			tp = 0;
1299 		}
1300 	}
1301 	return (tp);
1302 }
1303 
1304 /*
1305  * sysctl helper routine for net.inet.ip.mssdflt.  it can't be less
1306  * than 32.
1307  */
1308 static int
1309 sysctl_net_inet_tcp_mssdflt(SYSCTLFN_ARGS)
1310 {
1311 	int error, mssdflt;
1312 	struct sysctlnode node;
1313 
1314 	mssdflt = tcp_mssdflt;
1315 	node = *rnode;
1316 	node.sysctl_data = &mssdflt;
1317 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1318 	if (error || newp == NULL)
1319 		return (error);
1320 
1321 	if (mssdflt < 32)
1322 		return (EINVAL);
1323 	tcp_mssdflt = mssdflt;
1324 
1325 	mutex_enter(softnet_lock);
1326 	tcp_tcpcb_template();
1327 	mutex_exit(softnet_lock);
1328 
1329 	return (0);
1330 }
1331 
1332 /*
1333  * sysctl helper for TCP CB template update
1334  */
1335 static int
1336 sysctl_update_tcpcb_template(SYSCTLFN_ARGS)
1337 {
1338 	int t, error;
1339 	struct sysctlnode node;
1340 
1341 	/* follow procedures in sysctl(9) manpage */
1342 	t = *(int *)rnode->sysctl_data;
1343 	node = *rnode;
1344 	node.sysctl_data = &t;
1345 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1346 	if (error || newp == NULL)
1347 		return error;
1348 
1349 	if (t < 0)
1350 		return EINVAL;
1351 
1352 	*(int *)rnode->sysctl_data = t;
1353 
1354 	mutex_enter(softnet_lock);
1355 	tcp_tcpcb_template();
1356 	mutex_exit(softnet_lock);
1357 
1358 	return 0;
1359 }
1360 
1361 /*
1362  * sysctl helper routine for setting port related values under
1363  * net.inet.ip and net.inet6.ip6.  does basic range checking and does
1364  * additional checks for each type.  this code has placed in
1365  * tcp_input.c since INET and INET6 both use the same tcp code.
1366  *
1367  * this helper is not static so that both inet and inet6 can use it.
1368  */
1369 int
1370 sysctl_net_inet_ip_ports(SYSCTLFN_ARGS)
1371 {
1372 	int error, tmp;
1373 	int apmin, apmax;
1374 #ifndef IPNOPRIVPORTS
1375 	int lpmin, lpmax;
1376 #endif /* IPNOPRIVPORTS */
1377 	struct sysctlnode node;
1378 
1379 	if (namelen != 0)
1380 		return (EINVAL);
1381 
1382 	switch (name[-3]) {
1383 #ifdef INET
1384 	    case PF_INET:
1385 		apmin = anonportmin;
1386 		apmax = anonportmax;
1387 #ifndef IPNOPRIVPORTS
1388 		lpmin = lowportmin;
1389 		lpmax = lowportmax;
1390 #endif /* IPNOPRIVPORTS */
1391 		break;
1392 #endif /* INET */
1393 #ifdef INET6
1394 	    case PF_INET6:
1395 		apmin = ip6_anonportmin;
1396 		apmax = ip6_anonportmax;
1397 #ifndef IPNOPRIVPORTS
1398 		lpmin = ip6_lowportmin;
1399 		lpmax = ip6_lowportmax;
1400 #endif /* IPNOPRIVPORTS */
1401 		break;
1402 #endif /* INET6 */
1403 	    default:
1404 		return (EINVAL);
1405 	}
1406 
1407 	/*
1408 	 * insert temporary copy into node, perform lookup on
1409 	 * temporary, then restore pointer
1410 	 */
1411 	node = *rnode;
1412 	tmp = *(int*)rnode->sysctl_data;
1413 	node.sysctl_data = &tmp;
1414 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1415 	if (error || newp == NULL)
1416 		return (error);
1417 
1418 	/*
1419 	 * simple port range check
1420 	 */
1421 	if (tmp < 0 || tmp > 65535)
1422 		return (EINVAL);
1423 
1424 	/*
1425 	 * per-node range checks
1426 	 */
1427 	switch (rnode->sysctl_num) {
1428 	case IPCTL_ANONPORTMIN:
1429 	case IPV6CTL_ANONPORTMIN:
1430 		if (tmp >= apmax)
1431 			return (EINVAL);
1432 #ifndef IPNOPRIVPORTS
1433 		if (tmp < IPPORT_RESERVED)
1434                         return (EINVAL);
1435 #endif /* IPNOPRIVPORTS */
1436 		break;
1437 
1438 	case IPCTL_ANONPORTMAX:
1439 	case IPV6CTL_ANONPORTMAX:
1440                 if (apmin >= tmp)
1441 			return (EINVAL);
1442 #ifndef IPNOPRIVPORTS
1443 		if (tmp < IPPORT_RESERVED)
1444                         return (EINVAL);
1445 #endif /* IPNOPRIVPORTS */
1446 		break;
1447 
1448 #ifndef IPNOPRIVPORTS
1449 	case IPCTL_LOWPORTMIN:
1450 	case IPV6CTL_LOWPORTMIN:
1451 		if (tmp >= lpmax ||
1452 		    tmp > IPPORT_RESERVEDMAX ||
1453 		    tmp < IPPORT_RESERVEDMIN)
1454 			return (EINVAL);
1455 		break;
1456 
1457 	case IPCTL_LOWPORTMAX:
1458 	case IPV6CTL_LOWPORTMAX:
1459 		if (lpmin >= tmp ||
1460 		    tmp > IPPORT_RESERVEDMAX ||
1461 		    tmp < IPPORT_RESERVEDMIN)
1462 			return (EINVAL);
1463 		break;
1464 #endif /* IPNOPRIVPORTS */
1465 
1466 	default:
1467 		return (EINVAL);
1468 	}
1469 
1470 	*(int*)rnode->sysctl_data = tmp;
1471 
1472 	return (0);
1473 }
1474 
1475 static inline int
1476 copyout_uid(struct socket *sockp, void *oldp, size_t *oldlenp)
1477 {
1478 	if (oldp) {
1479 		size_t sz;
1480 		uid_t uid;
1481 		int error;
1482 
1483 		if (sockp->so_cred == NULL)
1484 			return EPERM;
1485 
1486 		uid = kauth_cred_geteuid(sockp->so_cred);
1487 		sz = MIN(sizeof(uid), *oldlenp);
1488 		if ((error = copyout(&uid, oldp, sz)) != 0)
1489 			return error;
1490 	}
1491 	*oldlenp = sizeof(uid_t);
1492 	return 0;
1493 }
1494 
1495 static inline int
1496 inet4_ident_core(struct in_addr raddr, u_int rport,
1497     struct in_addr laddr, u_int lport,
1498     void *oldp, size_t *oldlenp,
1499     struct lwp *l, int dodrop)
1500 {
1501 	struct inpcb *inp;
1502 	struct socket *sockp;
1503 
1504 	inp = in_pcblookup_connect(&tcbtable, raddr, rport, laddr, lport, 0);
1505 
1506 	if (inp == NULL || (sockp = inp->inp_socket) == NULL)
1507 		return ESRCH;
1508 
1509 	if (dodrop) {
1510 		struct tcpcb *tp;
1511 		int error;
1512 
1513 		if (inp == NULL || (tp = intotcpcb(inp)) == NULL ||
1514 		    (inp->inp_socket->so_options & SO_ACCEPTCONN) != 0)
1515 			return ESRCH;
1516 
1517 		error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_SOCKET,
1518 		    KAUTH_REQ_NETWORK_SOCKET_DROP, inp->inp_socket, tp, NULL);
1519 		if (error)
1520 			return (error);
1521 
1522 		(void)tcp_drop(tp, ECONNABORTED);
1523 		return 0;
1524 	}
1525 	else
1526 		return copyout_uid(sockp, oldp, oldlenp);
1527 }
1528 
1529 #ifdef INET6
1530 static inline int
1531 inet6_ident_core(struct in6_addr *raddr, u_int rport,
1532     struct in6_addr *laddr, u_int lport,
1533     void *oldp, size_t *oldlenp,
1534     struct lwp *l, int dodrop)
1535 {
1536 	struct in6pcb *in6p;
1537 	struct socket *sockp;
1538 
1539 	in6p = in6_pcblookup_connect(&tcbtable, raddr, rport, laddr, lport, 0, 0);
1540 
1541 	if (in6p == NULL || (sockp = in6p->in6p_socket) == NULL)
1542 		return ESRCH;
1543 
1544 	if (dodrop) {
1545 		struct tcpcb *tp;
1546 		int error;
1547 
1548 		if (in6p == NULL || (tp = in6totcpcb(in6p)) == NULL ||
1549 		    (in6p->in6p_socket->so_options & SO_ACCEPTCONN) != 0)
1550 			return ESRCH;
1551 
1552 		error = kauth_authorize_network(l->l_cred, KAUTH_NETWORK_SOCKET,
1553 		    KAUTH_REQ_NETWORK_SOCKET_DROP, in6p->in6p_socket, tp, NULL);
1554 		if (error)
1555 			return (error);
1556 
1557 		(void)tcp_drop(tp, ECONNABORTED);
1558 		return 0;
1559 	}
1560 	else
1561 		return copyout_uid(sockp, oldp, oldlenp);
1562 }
1563 #endif
1564 
1565 /*
1566  * sysctl helper routine for the net.inet.tcp.drop and
1567  * net.inet6.tcp6.drop nodes.
1568  */
1569 #define sysctl_net_inet_tcp_drop sysctl_net_inet_tcp_ident
1570 
1571 /*
1572  * sysctl helper routine for the net.inet.tcp.ident and
1573  * net.inet6.tcp6.ident nodes.  contains backwards compat code for the
1574  * old way of looking up the ident information for ipv4 which involves
1575  * stuffing the port/addr pairs into the mib lookup.
1576  */
1577 static int
1578 sysctl_net_inet_tcp_ident(SYSCTLFN_ARGS)
1579 {
1580 #ifdef INET
1581 	struct sockaddr_in *si4[2];
1582 #endif /* INET */
1583 #ifdef INET6
1584 	struct sockaddr_in6 *si6[2];
1585 #endif /* INET6 */
1586 	struct sockaddr_storage sa[2];
1587 	int error, pf, dodrop;
1588 
1589 	dodrop = name[-1] == TCPCTL_DROP;
1590 	if (dodrop) {
1591 		if (oldp != NULL || *oldlenp != 0)
1592 			return EINVAL;
1593 		if (newp == NULL)
1594 			return EPERM;
1595 		if (newlen < sizeof(sa))
1596 			return ENOMEM;
1597 	}
1598 	if (namelen != 4 && namelen != 0)
1599 		return EINVAL;
1600 	if (name[-2] != IPPROTO_TCP)
1601 		return EINVAL;
1602 	pf = name[-3];
1603 
1604 	/* old style lookup, ipv4 only */
1605 	if (namelen == 4) {
1606 #ifdef INET
1607 		struct in_addr laddr, raddr;
1608 		u_int lport, rport;
1609 
1610 		if (pf != PF_INET)
1611 			return EPROTONOSUPPORT;
1612 		raddr.s_addr = (uint32_t)name[0];
1613 		rport = (u_int)name[1];
1614 		laddr.s_addr = (uint32_t)name[2];
1615 		lport = (u_int)name[3];
1616 
1617 		mutex_enter(softnet_lock);
1618 		error = inet4_ident_core(raddr, rport, laddr, lport,
1619 		    oldp, oldlenp, l, dodrop);
1620 		mutex_exit(softnet_lock);
1621 		return error;
1622 #else /* INET */
1623 		return EINVAL;
1624 #endif /* INET */
1625 	}
1626 
1627 	if (newp == NULL || newlen != sizeof(sa))
1628 		return EINVAL;
1629 	error = copyin(newp, &sa, newlen);
1630 	if (error)
1631 		return error;
1632 
1633 	/*
1634 	 * requested families must match
1635 	 */
1636 	if (pf != sa[0].ss_family || sa[0].ss_family != sa[1].ss_family)
1637 		return EINVAL;
1638 
1639 	switch (pf) {
1640 #ifdef INET6
1641 	case PF_INET6:
1642 		si6[0] = (struct sockaddr_in6*)&sa[0];
1643 		si6[1] = (struct sockaddr_in6*)&sa[1];
1644 		if (si6[0]->sin6_len != sizeof(*si6[0]) ||
1645 		    si6[1]->sin6_len != sizeof(*si6[1]))
1646 			return EINVAL;
1647 
1648 		if (!IN6_IS_ADDR_V4MAPPED(&si6[0]->sin6_addr) &&
1649 		    !IN6_IS_ADDR_V4MAPPED(&si6[1]->sin6_addr)) {
1650 			error = sa6_embedscope(si6[0], ip6_use_defzone);
1651 			if (error)
1652 				return error;
1653 			error = sa6_embedscope(si6[1], ip6_use_defzone);
1654 			if (error)
1655 				return error;
1656 
1657 			mutex_enter(softnet_lock);
1658 			error = inet6_ident_core(&si6[0]->sin6_addr,
1659 			    si6[0]->sin6_port, &si6[1]->sin6_addr,
1660 			    si6[1]->sin6_port, oldp, oldlenp, l, dodrop);
1661 			mutex_exit(softnet_lock);
1662 			return error;
1663 		}
1664 
1665 		if (IN6_IS_ADDR_V4MAPPED(&si6[0]->sin6_addr) !=
1666 		    IN6_IS_ADDR_V4MAPPED(&si6[1]->sin6_addr))
1667 			return EINVAL;
1668 
1669 		in6_sin6_2_sin_in_sock((struct sockaddr *)&sa[0]);
1670 		in6_sin6_2_sin_in_sock((struct sockaddr *)&sa[1]);
1671 		/*FALLTHROUGH*/
1672 #endif /* INET6 */
1673 #ifdef INET
1674 	case PF_INET:
1675 		si4[0] = (struct sockaddr_in*)&sa[0];
1676 		si4[1] = (struct sockaddr_in*)&sa[1];
1677 		if (si4[0]->sin_len != sizeof(*si4[0]) ||
1678 		    si4[0]->sin_len != sizeof(*si4[1]))
1679 			return EINVAL;
1680 
1681 		mutex_enter(softnet_lock);
1682 		error = inet4_ident_core(si4[0]->sin_addr, si4[0]->sin_port,
1683 		    si4[1]->sin_addr, si4[1]->sin_port,
1684 		    oldp, oldlenp, l, dodrop);
1685 		mutex_exit(softnet_lock);
1686 		return error;
1687 #endif /* INET */
1688 	default:
1689 		return EPROTONOSUPPORT;
1690 	}
1691 }
1692 
1693 /*
1694  * sysctl helper for the inet and inet6 pcblists.  handles tcp/udp and
1695  * inet/inet6, as well as raw pcbs for each.  specifically not
1696  * declared static so that raw sockets and udp/udp6 can use it as
1697  * well.
1698  */
1699 int
1700 sysctl_inpcblist(SYSCTLFN_ARGS)
1701 {
1702 #ifdef INET
1703 	struct sockaddr_in *in;
1704 	const struct inpcb *inp;
1705 #endif
1706 #ifdef INET6
1707 	struct sockaddr_in6 *in6;
1708 	const struct in6pcb *in6p;
1709 #endif
1710 	struct inpcbtable *pcbtbl = __UNCONST(rnode->sysctl_data);
1711 	const struct inpcb_hdr *inph;
1712 	struct tcpcb *tp;
1713 	struct kinfo_pcb pcb;
1714 	char *dp;
1715 	size_t len, needed, elem_size, out_size;
1716 	int error, elem_count, pf, proto, pf2;
1717 
1718 	if (namelen != 4)
1719 		return (EINVAL);
1720 
1721 	if (oldp != NULL) {
1722 		    len = *oldlenp;
1723 		    elem_size = name[2];
1724 		    elem_count = name[3];
1725 		    if (elem_size != sizeof(pcb))
1726 			    return EINVAL;
1727 	} else {
1728 		    len = 0;
1729 		    elem_count = INT_MAX;
1730 		    elem_size = sizeof(pcb);
1731 	}
1732 	error = 0;
1733 	dp = oldp;
1734 	out_size = elem_size;
1735 	needed = 0;
1736 
1737 	if (namelen == 1 && name[0] == CTL_QUERY)
1738 		return (sysctl_query(SYSCTLFN_CALL(rnode)));
1739 
1740 	if (name - oname != 4)
1741 		return (EINVAL);
1742 
1743 	pf = oname[1];
1744 	proto = oname[2];
1745 	pf2 = (oldp != NULL) ? pf : 0;
1746 
1747 	mutex_enter(softnet_lock);
1748 
1749 	TAILQ_FOREACH(inph, &pcbtbl->inpt_queue, inph_queue) {
1750 #ifdef INET
1751 		inp = (const struct inpcb *)inph;
1752 #endif
1753 #ifdef INET6
1754 		in6p = (const struct in6pcb *)inph;
1755 #endif
1756 
1757 		if (inph->inph_af != pf)
1758 			continue;
1759 
1760 		if (kauth_authorize_network(l->l_cred, KAUTH_NETWORK_SOCKET,
1761 		    KAUTH_REQ_NETWORK_SOCKET_CANSEE, inph->inph_socket, NULL,
1762 		    NULL) != 0)
1763 			continue;
1764 
1765 		memset(&pcb, 0, sizeof(pcb));
1766 
1767 		pcb.ki_family = pf;
1768 		pcb.ki_type = proto;
1769 
1770 		switch (pf2) {
1771 		case 0:
1772 			/* just probing for size */
1773 			break;
1774 #ifdef INET
1775 		case PF_INET:
1776 			pcb.ki_family = inp->inp_socket->so_proto->
1777 			    pr_domain->dom_family;
1778 			pcb.ki_type = inp->inp_socket->so_proto->
1779 			    pr_type;
1780 			pcb.ki_protocol = inp->inp_socket->so_proto->
1781 			    pr_protocol;
1782 			pcb.ki_pflags = inp->inp_flags;
1783 
1784 			pcb.ki_sostate = inp->inp_socket->so_state;
1785 			pcb.ki_prstate = inp->inp_state;
1786 			if (proto == IPPROTO_TCP) {
1787 				tp = intotcpcb(inp);
1788 				pcb.ki_tstate = tp->t_state;
1789 				pcb.ki_tflags = tp->t_flags;
1790 			}
1791 
1792 			pcb.ki_pcbaddr = PTRTOUINT64(inp);
1793 			pcb.ki_ppcbaddr = PTRTOUINT64(inp->inp_ppcb);
1794 			pcb.ki_sockaddr = PTRTOUINT64(inp->inp_socket);
1795 
1796 			pcb.ki_rcvq = inp->inp_socket->so_rcv.sb_cc;
1797 			pcb.ki_sndq = inp->inp_socket->so_snd.sb_cc;
1798 
1799 			in = satosin(&pcb.ki_src);
1800 			in->sin_len = sizeof(*in);
1801 			in->sin_family = pf;
1802 			in->sin_port = inp->inp_lport;
1803 			in->sin_addr = inp->inp_laddr;
1804 			if (pcb.ki_prstate >= INP_CONNECTED) {
1805 				in = satosin(&pcb.ki_dst);
1806 				in->sin_len = sizeof(*in);
1807 				in->sin_family = pf;
1808 				in->sin_port = inp->inp_fport;
1809 				in->sin_addr = inp->inp_faddr;
1810 			}
1811 			break;
1812 #endif
1813 #ifdef INET6
1814 		case PF_INET6:
1815 			pcb.ki_family = in6p->in6p_socket->so_proto->
1816 			    pr_domain->dom_family;
1817 			pcb.ki_type = in6p->in6p_socket->so_proto->pr_type;
1818 			pcb.ki_protocol = in6p->in6p_socket->so_proto->
1819 			    pr_protocol;
1820 			pcb.ki_pflags = in6p->in6p_flags;
1821 
1822 			pcb.ki_sostate = in6p->in6p_socket->so_state;
1823 			pcb.ki_prstate = in6p->in6p_state;
1824 			if (proto == IPPROTO_TCP) {
1825 				tp = in6totcpcb(in6p);
1826 				pcb.ki_tstate = tp->t_state;
1827 				pcb.ki_tflags = tp->t_flags;
1828 			}
1829 
1830 			pcb.ki_pcbaddr = PTRTOUINT64(in6p);
1831 			pcb.ki_ppcbaddr = PTRTOUINT64(in6p->in6p_ppcb);
1832 			pcb.ki_sockaddr = PTRTOUINT64(in6p->in6p_socket);
1833 
1834 			pcb.ki_rcvq = in6p->in6p_socket->so_rcv.sb_cc;
1835 			pcb.ki_sndq = in6p->in6p_socket->so_snd.sb_cc;
1836 
1837 			in6 = satosin6(&pcb.ki_src);
1838 			in6->sin6_len = sizeof(*in6);
1839 			in6->sin6_family = pf;
1840 			in6->sin6_port = in6p->in6p_lport;
1841 			in6->sin6_flowinfo = in6p->in6p_flowinfo;
1842 			in6->sin6_addr = in6p->in6p_laddr;
1843 			in6->sin6_scope_id = 0; /* XXX? */
1844 
1845 			if (pcb.ki_prstate >= IN6P_CONNECTED) {
1846 				in6 = satosin6(&pcb.ki_dst);
1847 				in6->sin6_len = sizeof(*in6);
1848 				in6->sin6_family = pf;
1849 				in6->sin6_port = in6p->in6p_fport;
1850 				in6->sin6_flowinfo = in6p->in6p_flowinfo;
1851 				in6->sin6_addr = in6p->in6p_faddr;
1852 				in6->sin6_scope_id = 0; /* XXX? */
1853 			}
1854 			break;
1855 #endif
1856 		}
1857 
1858 		if (len >= elem_size && elem_count > 0) {
1859 			error = copyout(&pcb, dp, out_size);
1860 			if (error) {
1861 				mutex_exit(softnet_lock);
1862 				return (error);
1863 			}
1864 			dp += elem_size;
1865 			len -= elem_size;
1866 		}
1867 		needed += elem_size;
1868 		if (elem_count > 0 && elem_count != INT_MAX)
1869 			elem_count--;
1870 	}
1871 
1872 	*oldlenp = needed;
1873 	if (oldp == NULL)
1874 		*oldlenp += PCB_SLOP * sizeof(struct kinfo_pcb);
1875 
1876 	mutex_exit(softnet_lock);
1877 
1878 	return (error);
1879 }
1880 
1881 static int
1882 sysctl_tcp_congctl(SYSCTLFN_ARGS)
1883 {
1884 	struct sysctlnode node;
1885 	int error;
1886 	char newname[TCPCC_MAXLEN];
1887 
1888 	strlcpy(newname, tcp_congctl_global_name, sizeof(newname) - 1);
1889 
1890 	node = *rnode;
1891 	node.sysctl_data = newname;
1892 	node.sysctl_size = sizeof(newname);
1893 
1894 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1895 
1896 	if (error ||
1897 	    newp == NULL ||
1898 	    strncmp(newname, tcp_congctl_global_name, sizeof(newname)) == 0)
1899 		return error;
1900 
1901 	mutex_enter(softnet_lock);
1902 	error = tcp_congctl_select(NULL, newname);
1903 	mutex_exit(softnet_lock);
1904 
1905 	return error;
1906 }
1907 
1908 static int
1909 sysctl_tcp_init_win(SYSCTLFN_ARGS)
1910 {
1911 	int error;
1912 	u_int iw;
1913 	struct sysctlnode node;
1914 
1915 	iw = *(u_int *)rnode->sysctl_data;
1916 	node = *rnode;
1917 	node.sysctl_data = &iw;
1918 	node.sysctl_size = sizeof(iw);
1919 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1920 	if (error || newp == NULL)
1921 		return error;
1922 
1923 	if (iw >= __arraycount(tcp_init_win_max))
1924 		return EINVAL;
1925 	*(u_int *)rnode->sysctl_data = iw;
1926 	return 0;
1927 }
1928 
1929 static int
1930 sysctl_tcp_keep(SYSCTLFN_ARGS)
1931 {
1932 	int error;
1933 	u_int tmp;
1934 	struct sysctlnode node;
1935 
1936 	node = *rnode;
1937 	tmp = *(u_int *)rnode->sysctl_data;
1938 	node.sysctl_data = &tmp;
1939 
1940 	error = sysctl_lookup(SYSCTLFN_CALL(&node));
1941 	if (error || newp == NULL)
1942 		return error;
1943 
1944 	mutex_enter(softnet_lock);
1945 
1946 	*(u_int *)rnode->sysctl_data = tmp;
1947 	tcp_tcpcb_template();	/* update the template */
1948 
1949 	mutex_exit(softnet_lock);
1950 	return 0;
1951 }
1952 
1953 static int
1954 sysctl_net_inet_tcp_stats(SYSCTLFN_ARGS)
1955 {
1956 
1957 	return (NETSTAT_SYSCTL(tcpstat_percpu, TCP_NSTATS));
1958 }
1959 
1960 /*
1961  * this (second stage) setup routine is a replacement for tcp_sysctl()
1962  * (which is currently used for ipv4 and ipv6)
1963  */
1964 static void
1965 sysctl_net_inet_tcp_setup2(struct sysctllog **clog, int pf, const char *pfname,
1966 			   const char *tcpname)
1967 {
1968 	const struct sysctlnode *sack_node;
1969 	const struct sysctlnode *abc_node;
1970 	const struct sysctlnode *ecn_node;
1971 	const struct sysctlnode *congctl_node;
1972 	const struct sysctlnode *mslt_node;
1973 	const struct sysctlnode *vtw_node;
1974 #ifdef TCP_DEBUG
1975 	extern struct tcp_debug tcp_debug[TCP_NDEBUG];
1976 	extern int tcp_debx;
1977 #endif
1978 
1979 	sysctl_createv(clog, 0, NULL, NULL,
1980 		       CTLFLAG_PERMANENT,
1981 		       CTLTYPE_NODE, pfname, NULL,
1982 		       NULL, 0, NULL, 0,
1983 		       CTL_NET, pf, CTL_EOL);
1984 	sysctl_createv(clog, 0, NULL, NULL,
1985 		       CTLFLAG_PERMANENT,
1986 		       CTLTYPE_NODE, tcpname,
1987 		       SYSCTL_DESCR("TCP related settings"),
1988 		       NULL, 0, NULL, 0,
1989 		       CTL_NET, pf, IPPROTO_TCP, CTL_EOL);
1990 
1991 	sysctl_createv(clog, 0, NULL, NULL,
1992 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1993 		       CTLTYPE_INT, "rfc1323",
1994 		       SYSCTL_DESCR("Enable RFC1323 TCP extensions"),
1995 		       sysctl_update_tcpcb_template, 0, &tcp_do_rfc1323, 0,
1996 		       CTL_NET, pf, IPPROTO_TCP, TCPCTL_RFC1323, CTL_EOL);
1997 	sysctl_createv(clog, 0, NULL, NULL,
1998 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1999 		       CTLTYPE_INT, "sendspace",
2000 		       SYSCTL_DESCR("Default TCP send buffer size"),
2001 		       NULL, 0, &tcp_sendspace, 0,
2002 		       CTL_NET, pf, IPPROTO_TCP, TCPCTL_SENDSPACE, CTL_EOL);
2003 	sysctl_createv(clog, 0, NULL, NULL,
2004 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2005 		       CTLTYPE_INT, "recvspace",
2006 		       SYSCTL_DESCR("Default TCP receive buffer size"),
2007 		       NULL, 0, &tcp_recvspace, 0,
2008 		       CTL_NET, pf, IPPROTO_TCP, TCPCTL_RECVSPACE, CTL_EOL);
2009 	sysctl_createv(clog, 0, NULL, NULL,
2010 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2011 		       CTLTYPE_INT, "mssdflt",
2012 		       SYSCTL_DESCR("Default maximum segment size"),
2013 		       sysctl_net_inet_tcp_mssdflt, 0, &tcp_mssdflt, 0,
2014 		       CTL_NET, pf, IPPROTO_TCP, TCPCTL_MSSDFLT, CTL_EOL);
2015 	sysctl_createv(clog, 0, NULL, NULL,
2016 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2017 		       CTLTYPE_INT, "minmss",
2018 		       SYSCTL_DESCR("Lower limit for TCP maximum segment size"),
2019 		       NULL, 0, &tcp_minmss, 0,
2020 		       CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL);
2021 	sysctl_createv(clog, 0, NULL, NULL,
2022 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2023 		       CTLTYPE_INT, "msl",
2024 		       SYSCTL_DESCR("Maximum Segment Life"),
2025 		       NULL, 0, &tcp_msl, 0,
2026 		       CTL_NET, pf, IPPROTO_TCP, TCPCTL_MSL, CTL_EOL);
2027 	sysctl_createv(clog, 0, NULL, NULL,
2028 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2029 		       CTLTYPE_INT, "syn_cache_limit",
2030 		       SYSCTL_DESCR("Maximum number of entries in the TCP "
2031 				    "compressed state engine"),
2032 		       NULL, 0, &tcp_syn_cache_limit, 0,
2033 		       CTL_NET, pf, IPPROTO_TCP, TCPCTL_SYN_CACHE_LIMIT,
2034 		       CTL_EOL);
2035 	sysctl_createv(clog, 0, NULL, NULL,
2036 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2037 		       CTLTYPE_INT, "syn_bucket_limit",
2038 		       SYSCTL_DESCR("Maximum number of entries per hash "
2039 				    "bucket in the TCP compressed state "
2040 				    "engine"),
2041 		       NULL, 0, &tcp_syn_bucket_limit, 0,
2042 		       CTL_NET, pf, IPPROTO_TCP, TCPCTL_SYN_BUCKET_LIMIT,
2043 		       CTL_EOL);
2044 #if 0 /* obsoleted */
2045 	sysctl_createv(clog, 0, NULL, NULL,
2046 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2047 		       CTLTYPE_INT, "syn_cache_interval",
2048 		       SYSCTL_DESCR("TCP compressed state engine's timer interval"),
2049 		       NULL, 0, &tcp_syn_cache_interval, 0,
2050 		       CTL_NET, pf, IPPROTO_TCP, TCPCTL_SYN_CACHE_INTER,
2051 		       CTL_EOL);
2052 #endif
2053 	sysctl_createv(clog, 0, NULL, NULL,
2054 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2055 		       CTLTYPE_INT, "init_win",
2056 		       SYSCTL_DESCR("Initial TCP congestion window"),
2057 		       sysctl_tcp_init_win, 0, &tcp_init_win, 0,
2058 		       CTL_NET, pf, IPPROTO_TCP, TCPCTL_INIT_WIN, CTL_EOL);
2059 	sysctl_createv(clog, 0, NULL, NULL,
2060 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2061 		       CTLTYPE_INT, "mss_ifmtu",
2062 		       SYSCTL_DESCR("Use interface MTU for calculating MSS"),
2063 		       NULL, 0, &tcp_mss_ifmtu, 0,
2064 		       CTL_NET, pf, IPPROTO_TCP, TCPCTL_MSS_IFMTU, CTL_EOL);
2065 	sysctl_createv(clog, 0, NULL, &sack_node,
2066 		       CTLFLAG_PERMANENT,
2067 		       CTLTYPE_NODE, "sack",
2068 		       SYSCTL_DESCR("RFC2018 Selective ACKnowledgement tunables"),
2069 		       NULL, 0, NULL, 0,
2070 		       CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_EOL);
2071 
2072 	/* Congctl subtree */
2073 	sysctl_createv(clog, 0, NULL, &congctl_node,
2074 		       CTLFLAG_PERMANENT,
2075 		       CTLTYPE_NODE, "congctl",
2076 		       SYSCTL_DESCR("TCP Congestion Control"),
2077 	    	       NULL, 0, NULL, 0,
2078 		       CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL);
2079 	sysctl_createv(clog, 0, &congctl_node, NULL,
2080 		       CTLFLAG_PERMANENT,
2081 		       CTLTYPE_STRING, "available",
2082 		       SYSCTL_DESCR("Available Congestion Control Mechanisms"),
2083 		       NULL, 0, tcp_congctl_avail, 0, CTL_CREATE, CTL_EOL);
2084 	sysctl_createv(clog, 0, &congctl_node, NULL,
2085 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2086 		       CTLTYPE_STRING, "selected",
2087 		       SYSCTL_DESCR("Selected Congestion Control Mechanism"),
2088 		       sysctl_tcp_congctl, 0, NULL, TCPCC_MAXLEN,
2089 		       CTL_CREATE, CTL_EOL);
2090 
2091 	sysctl_createv(clog, 0, NULL, NULL,
2092 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2093 		       CTLTYPE_INT, "win_scale",
2094 		       SYSCTL_DESCR("Use RFC1323 window scale options"),
2095 		       sysctl_update_tcpcb_template, 0, &tcp_do_win_scale, 0,
2096 		       CTL_NET, pf, IPPROTO_TCP, TCPCTL_WSCALE, CTL_EOL);
2097 	sysctl_createv(clog, 0, NULL, NULL,
2098 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2099 		       CTLTYPE_INT, "timestamps",
2100 		       SYSCTL_DESCR("Use RFC1323 time stamp options"),
2101 		       sysctl_update_tcpcb_template, 0, &tcp_do_timestamps, 0,
2102 		       CTL_NET, pf, IPPROTO_TCP, TCPCTL_TSTAMP, CTL_EOL);
2103 	sysctl_createv(clog, 0, NULL, NULL,
2104 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2105 		       CTLTYPE_INT, "compat_42",
2106 		       SYSCTL_DESCR("Enable workarounds for 4.2BSD TCP bugs"),
2107 		       NULL, 0, &tcp_compat_42, 0,
2108 		       CTL_NET, pf, IPPROTO_TCP, TCPCTL_COMPAT_42, CTL_EOL);
2109 	sysctl_createv(clog, 0, NULL, NULL,
2110 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2111 		       CTLTYPE_INT, "cwm",
2112 		       SYSCTL_DESCR("Hughes/Touch/Heidemann Congestion Window "
2113 				    "Monitoring"),
2114 		       NULL, 0, &tcp_cwm, 0,
2115 		       CTL_NET, pf, IPPROTO_TCP, TCPCTL_CWM, CTL_EOL);
2116 	sysctl_createv(clog, 0, NULL, NULL,
2117 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2118 		       CTLTYPE_INT, "cwm_burstsize",
2119 		       SYSCTL_DESCR("Congestion Window Monitoring allowed "
2120 				    "burst count in packets"),
2121 		       NULL, 0, &tcp_cwm_burstsize, 0,
2122 		       CTL_NET, pf, IPPROTO_TCP, TCPCTL_CWM_BURSTSIZE,
2123 		       CTL_EOL);
2124 	sysctl_createv(clog, 0, NULL, NULL,
2125 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2126 		       CTLTYPE_INT, "ack_on_push",
2127 		       SYSCTL_DESCR("Immediately return ACK when PSH is "
2128 				    "received"),
2129 		       NULL, 0, &tcp_ack_on_push, 0,
2130 		       CTL_NET, pf, IPPROTO_TCP, TCPCTL_ACK_ON_PUSH, CTL_EOL);
2131 	sysctl_createv(clog, 0, NULL, NULL,
2132 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2133 		       CTLTYPE_INT, "keepidle",
2134 		       SYSCTL_DESCR("Allowed connection idle ticks before a "
2135 				    "keepalive probe is sent"),
2136 		       sysctl_tcp_keep, 0, &tcp_keepidle, 0,
2137 		       CTL_NET, pf, IPPROTO_TCP, TCPCTL_KEEPIDLE, CTL_EOL);
2138 	sysctl_createv(clog, 0, NULL, NULL,
2139 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2140 		       CTLTYPE_INT, "keepintvl",
2141 		       SYSCTL_DESCR("Ticks before next keepalive probe is sent"),
2142 		       sysctl_tcp_keep, 0, &tcp_keepintvl, 0,
2143 		       CTL_NET, pf, IPPROTO_TCP, TCPCTL_KEEPINTVL, CTL_EOL);
2144 	sysctl_createv(clog, 0, NULL, NULL,
2145 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2146 		       CTLTYPE_INT, "keepcnt",
2147 		       SYSCTL_DESCR("Number of keepalive probes to send"),
2148 		       sysctl_tcp_keep, 0, &tcp_keepcnt, 0,
2149 		       CTL_NET, pf, IPPROTO_TCP, TCPCTL_KEEPCNT, CTL_EOL);
2150 	sysctl_createv(clog, 0, NULL, NULL,
2151 		       CTLFLAG_PERMANENT|CTLFLAG_IMMEDIATE,
2152 		       CTLTYPE_INT, "slowhz",
2153 		       SYSCTL_DESCR("Keepalive ticks per second"),
2154 		       NULL, PR_SLOWHZ, NULL, 0,
2155 		       CTL_NET, pf, IPPROTO_TCP, TCPCTL_SLOWHZ, CTL_EOL);
2156 	sysctl_createv(clog, 0, NULL, NULL,
2157 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2158 		       CTLTYPE_INT, "log_refused",
2159 		       SYSCTL_DESCR("Log refused TCP connections"),
2160 		       NULL, 0, &tcp_log_refused, 0,
2161 		       CTL_NET, pf, IPPROTO_TCP, TCPCTL_LOG_REFUSED, CTL_EOL);
2162 #if 0 /* obsoleted */
2163 	sysctl_createv(clog, 0, NULL, NULL,
2164 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2165 		       CTLTYPE_INT, "rstratelimit", NULL,
2166 		       NULL, 0, &tcp_rst_ratelim, 0,
2167 		       CTL_NET, pf, IPPROTO_TCP, TCPCTL_RSTRATELIMIT, CTL_EOL);
2168 #endif
2169 	sysctl_createv(clog, 0, NULL, NULL,
2170 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2171 		       CTLTYPE_INT, "rstppslimit",
2172 		       SYSCTL_DESCR("Maximum number of RST packets to send "
2173 				    "per second"),
2174 		       NULL, 0, &tcp_rst_ppslim, 0,
2175 		       CTL_NET, pf, IPPROTO_TCP, TCPCTL_RSTPPSLIMIT, CTL_EOL);
2176 	sysctl_createv(clog, 0, NULL, NULL,
2177 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2178 		       CTLTYPE_INT, "delack_ticks",
2179 		       SYSCTL_DESCR("Number of ticks to delay sending an ACK"),
2180 		       NULL, 0, &tcp_delack_ticks, 0,
2181 		       CTL_NET, pf, IPPROTO_TCP, TCPCTL_DELACK_TICKS, CTL_EOL);
2182 	sysctl_createv(clog, 0, NULL, NULL,
2183 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2184 		       CTLTYPE_INT, "init_win_local",
2185 		       SYSCTL_DESCR("Initial TCP window size (in segments)"),
2186 		       sysctl_tcp_init_win, 0, &tcp_init_win_local, 0,
2187 		       CTL_NET, pf, IPPROTO_TCP, TCPCTL_INIT_WIN_LOCAL,
2188 		       CTL_EOL);
2189 	sysctl_createv(clog, 0, NULL, NULL,
2190 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2191 		       CTLTYPE_STRUCT, "ident",
2192 		       SYSCTL_DESCR("RFC1413 Identification Protocol lookups"),
2193 		       sysctl_net_inet_tcp_ident, 0, NULL, sizeof(uid_t),
2194 		       CTL_NET, pf, IPPROTO_TCP, TCPCTL_IDENT, CTL_EOL);
2195 	sysctl_createv(clog, 0, NULL, NULL,
2196 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2197 		       CTLTYPE_INT, "do_loopback_cksum",
2198 		       SYSCTL_DESCR("Perform TCP checksum on loopback"),
2199 		       NULL, 0, &tcp_do_loopback_cksum, 0,
2200 		       CTL_NET, pf, IPPROTO_TCP, TCPCTL_LOOPBACKCKSUM,
2201 		       CTL_EOL);
2202 	sysctl_createv(clog, 0, NULL, NULL,
2203 		       CTLFLAG_PERMANENT,
2204 		       CTLTYPE_STRUCT, "pcblist",
2205 		       SYSCTL_DESCR("TCP protocol control block list"),
2206 		       sysctl_inpcblist, 0, &tcbtable, 0,
2207 		       CTL_NET, pf, IPPROTO_TCP, CTL_CREATE,
2208 		       CTL_EOL);
2209 	sysctl_createv(clog, 0, NULL, NULL,
2210 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2211 		       CTLTYPE_INT, "keepinit",
2212 		       SYSCTL_DESCR("Ticks before initial tcp connection times out"),
2213 		       sysctl_tcp_keep, 0, &tcp_keepinit, 0,
2214 		       CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL);
2215 
2216 	/* TCP socket buffers auto-sizing nodes */
2217 	sysctl_createv(clog, 0, NULL, NULL,
2218 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2219 		       CTLTYPE_INT, "recvbuf_auto",
2220 		       SYSCTL_DESCR("Enable automatic receive "
2221 		           "buffer sizing (experimental)"),
2222 		       NULL, 0, &tcp_do_autorcvbuf, 0,
2223 		       CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL);
2224 	sysctl_createv(clog, 0, NULL, NULL,
2225 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2226 		       CTLTYPE_INT, "recvbuf_inc",
2227 		       SYSCTL_DESCR("Incrementor step size of "
2228 		           "automatic receive buffer"),
2229 		       NULL, 0, &tcp_autorcvbuf_inc, 0,
2230 		       CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL);
2231 	sysctl_createv(clog, 0, NULL, NULL,
2232 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2233 		       CTLTYPE_INT, "recvbuf_max",
2234 		       SYSCTL_DESCR("Max size of automatic receive buffer"),
2235 		       NULL, 0, &tcp_autorcvbuf_max, 0,
2236 		       CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL);
2237 
2238 	sysctl_createv(clog, 0, NULL, NULL,
2239 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2240 		       CTLTYPE_INT, "sendbuf_auto",
2241 		       SYSCTL_DESCR("Enable automatic send "
2242 		           "buffer sizing (experimental)"),
2243 		       NULL, 0, &tcp_do_autosndbuf, 0,
2244 		       CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL);
2245 	sysctl_createv(clog, 0, NULL, NULL,
2246 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2247 		       CTLTYPE_INT, "sendbuf_inc",
2248 		       SYSCTL_DESCR("Incrementor step size of "
2249 		           "automatic send buffer"),
2250 		       NULL, 0, &tcp_autosndbuf_inc, 0,
2251 		       CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL);
2252 	sysctl_createv(clog, 0, NULL, NULL,
2253 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2254 		       CTLTYPE_INT, "sendbuf_max",
2255 		       SYSCTL_DESCR("Max size of automatic send buffer"),
2256 		       NULL, 0, &tcp_autosndbuf_max, 0,
2257 		       CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL);
2258 
2259 	/* ECN subtree */
2260 	sysctl_createv(clog, 0, NULL, &ecn_node,
2261 	    	       CTLFLAG_PERMANENT,
2262 		       CTLTYPE_NODE, "ecn",
2263 	    	       SYSCTL_DESCR("RFC3168 Explicit Congestion Notification"),
2264 	    	       NULL, 0, NULL, 0,
2265 		       CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL);
2266 	sysctl_createv(clog, 0, &ecn_node, NULL,
2267 	    	       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2268 		       CTLTYPE_INT, "enable",
2269 		       SYSCTL_DESCR("Enable TCP Explicit Congestion "
2270 			   "Notification"),
2271 	    	       NULL, 0, &tcp_do_ecn, 0, CTL_CREATE, CTL_EOL);
2272 	sysctl_createv(clog, 0, &ecn_node, NULL,
2273 	    	       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2274 		       CTLTYPE_INT, "maxretries",
2275 		       SYSCTL_DESCR("Number of times to retry ECN setup "
2276 			       "before disabling ECN on the connection"),
2277 	    	       NULL, 0, &tcp_ecn_maxretries, 0, CTL_CREATE, CTL_EOL);
2278 
2279 	/* SACK gets its own little subtree. */
2280 	sysctl_createv(clog, 0, NULL, &sack_node,
2281 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2282 		       CTLTYPE_INT, "enable",
2283 		       SYSCTL_DESCR("Enable RFC2018 Selective ACKnowledgement"),
2284 		       NULL, 0, &tcp_do_sack, 0,
2285 		       CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_CREATE, CTL_EOL);
2286 	sysctl_createv(clog, 0, NULL, &sack_node,
2287 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2288 		       CTLTYPE_INT, "maxholes",
2289 		       SYSCTL_DESCR("Maximum number of TCP SACK holes allowed per connection"),
2290 		       NULL, 0, &tcp_sack_tp_maxholes, 0,
2291 		       CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_CREATE, CTL_EOL);
2292 	sysctl_createv(clog, 0, NULL, &sack_node,
2293 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2294 		       CTLTYPE_INT, "globalmaxholes",
2295 		       SYSCTL_DESCR("Global maximum number of TCP SACK holes"),
2296 		       NULL, 0, &tcp_sack_globalmaxholes, 0,
2297 		       CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_CREATE, CTL_EOL);
2298 	sysctl_createv(clog, 0, NULL, &sack_node,
2299 		       CTLFLAG_PERMANENT,
2300 		       CTLTYPE_INT, "globalholes",
2301 		       SYSCTL_DESCR("Global number of TCP SACK holes"),
2302 		       NULL, 0, &tcp_sack_globalholes, 0,
2303 		       CTL_NET, pf, IPPROTO_TCP, TCPCTL_SACK, CTL_CREATE, CTL_EOL);
2304 
2305 	sysctl_createv(clog, 0, NULL, NULL,
2306 		       CTLFLAG_PERMANENT,
2307 		       CTLTYPE_STRUCT, "stats",
2308 		       SYSCTL_DESCR("TCP statistics"),
2309 		       sysctl_net_inet_tcp_stats, 0, NULL, 0,
2310 		       CTL_NET, pf, IPPROTO_TCP, TCPCTL_STATS,
2311 		       CTL_EOL);
2312         sysctl_createv(clog, 0, NULL, NULL,
2313                        CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2314                        CTLTYPE_INT, "local_by_rtt",
2315                        SYSCTL_DESCR("Use RTT estimator to decide which hosts "
2316 				    "are local"),
2317 		       NULL, 0, &tcp_rttlocal, 0,
2318 		       CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL);
2319 #ifdef TCP_DEBUG
2320 	sysctl_createv(clog, 0, NULL, NULL,
2321 		       CTLFLAG_PERMANENT,
2322 		       CTLTYPE_STRUCT, "debug",
2323 		       SYSCTL_DESCR("TCP sockets debug information"),
2324 		       NULL, 0, &tcp_debug, sizeof(tcp_debug),
2325 		       CTL_NET, pf, IPPROTO_TCP, TCPCTL_DEBUG,
2326 		       CTL_EOL);
2327 	sysctl_createv(clog, 0, NULL, NULL,
2328 		       CTLFLAG_PERMANENT,
2329 		       CTLTYPE_INT, "debx",
2330 		       SYSCTL_DESCR("Number of TCP debug sockets messages"),
2331 		       NULL, 0, &tcp_debx, sizeof(tcp_debx),
2332 		       CTL_NET, pf, IPPROTO_TCP, TCPCTL_DEBX,
2333 		       CTL_EOL);
2334 #endif
2335 	sysctl_createv(clog, 0, NULL, NULL,
2336 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2337 		       CTLTYPE_STRUCT, "drop",
2338 		       SYSCTL_DESCR("TCP drop connection"),
2339 		       sysctl_net_inet_tcp_drop, 0, NULL, 0,
2340 		       CTL_NET, pf, IPPROTO_TCP, TCPCTL_DROP, CTL_EOL);
2341 	sysctl_createv(clog, 0, NULL, NULL,
2342 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2343 		       CTLTYPE_INT, "iss_hash",
2344 		       SYSCTL_DESCR("Enable RFC 1948 ISS by cryptographic "
2345 				    "hash computation"),
2346 		       NULL, 0, &tcp_do_rfc1948, sizeof(tcp_do_rfc1948),
2347 		       CTL_NET, pf, IPPROTO_TCP, CTL_CREATE,
2348 		       CTL_EOL);
2349 
2350 	/* ABC subtree */
2351 
2352 	sysctl_createv(clog, 0, NULL, &abc_node,
2353 		       CTLFLAG_PERMANENT, CTLTYPE_NODE, "abc",
2354 		       SYSCTL_DESCR("RFC3465 Appropriate Byte Counting (ABC)"),
2355 		       NULL, 0, NULL, 0,
2356 		       CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL);
2357 	sysctl_createv(clog, 0, &abc_node, NULL,
2358 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2359 		       CTLTYPE_INT, "enable",
2360 		       SYSCTL_DESCR("Enable RFC3465 Appropriate Byte Counting"),
2361 		       NULL, 0, &tcp_do_abc, 0, CTL_CREATE, CTL_EOL);
2362 	sysctl_createv(clog, 0, &abc_node, NULL,
2363 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2364 		       CTLTYPE_INT, "aggressive",
2365 		       SYSCTL_DESCR("1: L=2*SMSS 0: L=1*SMSS"),
2366 		       NULL, 0, &tcp_abc_aggressive, 0, CTL_CREATE, CTL_EOL);
2367 
2368 	/* MSL tuning subtree */
2369 
2370 	sysctl_createv(clog, 0, NULL, &mslt_node,
2371 		       CTLFLAG_PERMANENT, CTLTYPE_NODE, "mslt",
2372 		       SYSCTL_DESCR("MSL Tuning for TIME_WAIT truncation"),
2373 		       NULL, 0, NULL, 0,
2374 		       CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL);
2375 	sysctl_createv(clog, 0, &mslt_node, NULL,
2376 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2377 		       CTLTYPE_INT, "enable",
2378 		       SYSCTL_DESCR("Enable TIME_WAIT truncation"),
2379 		       NULL, 0, &tcp_msl_enable, 0, CTL_CREATE, CTL_EOL);
2380 	sysctl_createv(clog, 0, &mslt_node, NULL,
2381 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2382 		       CTLTYPE_INT, "loopback",
2383 		       SYSCTL_DESCR("MSL value to use for loopback connections"),
2384 		       NULL, 0, &tcp_msl_loop, 0, CTL_CREATE, CTL_EOL);
2385 	sysctl_createv(clog, 0, &mslt_node, NULL,
2386 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2387 		       CTLTYPE_INT, "local",
2388 		       SYSCTL_DESCR("MSL value to use for local connections"),
2389 		       NULL, 0, &tcp_msl_local, 0, CTL_CREATE, CTL_EOL);
2390 	sysctl_createv(clog, 0, &mslt_node, NULL,
2391 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2392 		       CTLTYPE_INT, "remote",
2393 		       SYSCTL_DESCR("MSL value to use for remote connections"),
2394 		       NULL, 0, &tcp_msl_remote, 0, CTL_CREATE, CTL_EOL);
2395 	sysctl_createv(clog, 0, &mslt_node, NULL,
2396 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2397 		       CTLTYPE_INT, "remote_threshold",
2398 		       SYSCTL_DESCR("RTT estimate value to promote local to remote"),
2399 		       NULL, 0, &tcp_msl_remote_threshold, 0, CTL_CREATE, CTL_EOL);
2400 
2401 	/* vestigial TIME_WAIT tuning subtree */
2402 
2403 	sysctl_createv(clog, 0, NULL, &vtw_node,
2404 		       CTLFLAG_PERMANENT, CTLTYPE_NODE, "vtw",
2405 		       SYSCTL_DESCR("Tuning for Vestigial TIME_WAIT"),
2406 		       NULL, 0, NULL, 0,
2407 		       CTL_NET, pf, IPPROTO_TCP, CTL_CREATE, CTL_EOL);
2408 	sysctl_createv(clog, 0, &vtw_node, NULL,
2409 		       CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2410 		       CTLTYPE_INT, "enable",
2411 		       SYSCTL_DESCR("Enable Vestigial TIME_WAIT"),
2412 		       sysctl_tcp_vtw_enable, 0,
2413 	               (pf == AF_INET) ? &tcp4_vtw_enable : &tcp6_vtw_enable,
2414 		       0, CTL_CREATE, CTL_EOL);
2415 	sysctl_createv(clog, 0, &vtw_node, NULL,
2416 		       CTLFLAG_PERMANENT|CTLFLAG_READONLY,
2417 		       CTLTYPE_INT, "entries",
2418 		       SYSCTL_DESCR("Maximum number of vestigial TIME_WAIT entries"),
2419 		       NULL, 0, &tcp_vtw_entries, 0, CTL_CREATE, CTL_EOL);
2420 }
2421 
2422 void
2423 tcp_usrreq_init(void)
2424 {
2425 
2426 #ifdef INET
2427 	sysctl_net_inet_tcp_setup2(NULL, PF_INET, "inet", "tcp");
2428 #endif
2429 #ifdef INET6
2430 	sysctl_net_inet_tcp_setup2(NULL, PF_INET6, "inet6", "tcp6");
2431 #endif
2432 }
2433 
2434 PR_WRAP_USRREQS(tcp)
2435 #define	tcp_attach	tcp_attach_wrapper
2436 #define	tcp_detach	tcp_detach_wrapper
2437 #define	tcp_accept	tcp_accept_wrapper
2438 #define	tcp_bind	tcp_bind_wrapper
2439 #define	tcp_listen	tcp_listen_wrapper
2440 #define	tcp_connect	tcp_connect_wrapper
2441 #define	tcp_connect2	tcp_connect2_wrapper
2442 #define	tcp_disconnect	tcp_disconnect_wrapper
2443 #define	tcp_shutdown	tcp_shutdown_wrapper
2444 #define	tcp_abort	tcp_abort_wrapper
2445 #define	tcp_ioctl	tcp_ioctl_wrapper
2446 #define	tcp_stat	tcp_stat_wrapper
2447 #define	tcp_peeraddr	tcp_peeraddr_wrapper
2448 #define	tcp_sockaddr	tcp_sockaddr_wrapper
2449 #define	tcp_rcvd	tcp_rcvd_wrapper
2450 #define	tcp_recvoob	tcp_recvoob_wrapper
2451 #define	tcp_send	tcp_send_wrapper
2452 #define	tcp_sendoob	tcp_sendoob_wrapper
2453 #define	tcp_purgeif	tcp_purgeif_wrapper
2454 #define	tcp_usrreq	tcp_usrreq_wrapper
2455 
2456 const struct pr_usrreqs tcp_usrreqs = {
2457 	.pr_attach	= tcp_attach,
2458 	.pr_detach	= tcp_detach,
2459 	.pr_accept	= tcp_accept,
2460 	.pr_bind	= tcp_bind,
2461 	.pr_listen	= tcp_listen,
2462 	.pr_connect	= tcp_connect,
2463 	.pr_connect2	= tcp_connect2,
2464 	.pr_disconnect	= tcp_disconnect,
2465 	.pr_shutdown	= tcp_shutdown,
2466 	.pr_abort	= tcp_abort,
2467 	.pr_ioctl	= tcp_ioctl,
2468 	.pr_stat	= tcp_stat,
2469 	.pr_peeraddr	= tcp_peeraddr,
2470 	.pr_sockaddr	= tcp_sockaddr,
2471 	.pr_rcvd	= tcp_rcvd,
2472 	.pr_recvoob	= tcp_recvoob,
2473 	.pr_send	= tcp_send,
2474 	.pr_sendoob	= tcp_sendoob,
2475 	.pr_purgeif	= tcp_purgeif,
2476 	.pr_generic	= tcp_usrreq,
2477 };
2478