xref: /netbsd-src/sys/net/rtsock.c (revision 946379e7b37692fc43f68eb0d1c10daa0a7f3b6c)
1 /*	$NetBSD: rtsock.c,v 1.174 2015/10/13 21:28:34 rjs 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) 1988, 1991, 1993
34  *	The Regents of the University of California.  All rights reserved.
35  *
36  * Redistribution and use in source and binary forms, with or without
37  * modification, are permitted provided that the following conditions
38  * are met:
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. Neither the name of the University nor the names of its contributors
45  *    may be used to endorse or promote products derived from this software
46  *    without specific prior written permission.
47  *
48  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
49  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
50  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
51  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
52  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
53  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
54  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
55  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
56  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
57  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
58  * SUCH DAMAGE.
59  *
60  *	@(#)rtsock.c	8.7 (Berkeley) 10/12/95
61  */
62 
63 #include <sys/cdefs.h>
64 __KERNEL_RCSID(0, "$NetBSD: rtsock.c,v 1.174 2015/10/13 21:28:34 rjs Exp $");
65 
66 #ifdef _KERNEL_OPT
67 #include "opt_inet.h"
68 #include "opt_mpls.h"
69 #include "opt_compat_netbsd.h"
70 #include "opt_sctp.h"
71 #endif
72 
73 #include <sys/param.h>
74 #include <sys/systm.h>
75 #include <sys/proc.h>
76 #include <sys/socket.h>
77 #include <sys/socketvar.h>
78 #include <sys/domain.h>
79 #include <sys/protosw.h>
80 #include <sys/sysctl.h>
81 #include <sys/kauth.h>
82 #include <sys/kmem.h>
83 #include <sys/intr.h>
84 #ifdef RTSOCK_DEBUG
85 #include <netinet/in.h>
86 #endif /* RTSOCK_DEBUG */
87 
88 #include <net/if.h>
89 #include <net/route.h>
90 #include <net/raw_cb.h>
91 
92 #include <netmpls/mpls.h>
93 
94 #ifdef SCTP
95 extern void sctp_add_ip_address(struct ifaddr *);
96 extern void sctp_delete_ip_address(struct ifaddr *);
97 #endif
98 
99 #if defined(COMPAT_14) || defined(COMPAT_50)
100 #include <compat/net/if.h>
101 #include <compat/net/route.h>
102 #endif
103 #ifdef COMPAT_RTSOCK
104 #define	RTM_XVERSION	RTM_OVERSION
105 #define	RT_XADVANCE(a,b) RT_OADVANCE(a,b)
106 #define	RT_XROUNDUP(n)	RT_OROUNDUP(n)
107 #define	PF_XROUTE	PF_OROUTE
108 #define	rt_xmsghdr	rt_msghdr50
109 #define	if_xmsghdr	if_msghdr	/* if_msghdr50 is for RTM_OIFINFO */
110 #define	ifa_xmsghdr	ifa_msghdr50
111 #define	if_xannouncemsghdr	if_announcemsghdr50
112 #define	COMPATNAME(x)	compat_50_ ## x
113 #define	DOMAINNAME	"oroute"
114 CTASSERT(sizeof(struct ifa_xmsghdr) == 20);
115 DOMAIN_DEFINE(compat_50_routedomain); /* forward declare and add to link set */
116 #else /* COMPAT_RTSOCK */
117 #define	RTM_XVERSION	RTM_VERSION
118 #define	RT_XADVANCE(a,b) RT_ADVANCE(a,b)
119 #define	RT_XROUNDUP(n)	RT_ROUNDUP(n)
120 #define	PF_XROUTE	PF_ROUTE
121 #define	rt_xmsghdr	rt_msghdr
122 #define	if_xmsghdr	if_msghdr
123 #define	ifa_xmsghdr	ifa_msghdr
124 #define	if_xannouncemsghdr	if_announcemsghdr
125 #define	COMPATNAME(x)	x
126 #define	DOMAINNAME	"route"
127 CTASSERT(sizeof(struct ifa_xmsghdr) == 24);
128 #ifdef COMPAT_50
129 #define	COMPATCALL(name, args)	compat_50_ ## name args
130 #endif
131 DOMAIN_DEFINE(routedomain); /* forward declare and add to link set */
132 #undef COMPAT_50
133 #undef COMPAT_14
134 #endif /* COMPAT_RTSOCK */
135 
136 #ifndef COMPATCALL
137 #define	COMPATCALL(name, args)	do { } while (/*CONSTCOND*/ 0)
138 #endif
139 
140 #ifdef RTSOCK_DEBUG
141 #define RT_IN_PRINT(b, a) (in_print((b), sizeof(b), \
142     &((const struct sockaddr_in *)info.rti_info[(a)])->sin_addr), (b))
143 #endif /* RTSOCK_DEBUG */
144 
145 struct route_info COMPATNAME(route_info) = {
146 	.ri_dst = { .sa_len = 2, .sa_family = PF_XROUTE, },
147 	.ri_src = { .sa_len = 2, .sa_family = PF_XROUTE, },
148 	.ri_maxqlen = IFQ_MAXLEN,
149 };
150 
151 #define	PRESERVED_RTF	(RTF_UP | RTF_GATEWAY | RTF_HOST | RTF_DONE | RTF_MASK)
152 
153 static void COMPATNAME(route_init)(void);
154 static int COMPATNAME(route_output)(struct mbuf *, ...);
155 
156 static int rt_msg2(int, struct rt_addrinfo *, void *, struct rt_walkarg *, int *);
157 static int rt_xaddrs(u_char, const char *, const char *, struct rt_addrinfo *);
158 static struct mbuf *rt_makeifannouncemsg(struct ifnet *, int, int,
159     struct rt_addrinfo *);
160 static void rt_setmetrics(int, const struct rt_xmsghdr *, struct rtentry *);
161 static void rtm_setmetrics(const struct rtentry *, struct rt_xmsghdr *);
162 static void sysctl_net_route_setup(struct sysctllog **);
163 static int sysctl_dumpentry(struct rtentry *, void *);
164 static int sysctl_iflist(int, struct rt_walkarg *, int);
165 static int sysctl_rtable(SYSCTLFN_PROTO);
166 static void rt_adjustcount(int, int);
167 
168 static void
169 rt_adjustcount(int af, int cnt)
170 {
171 	struct route_cb * const cb = &COMPATNAME(route_info).ri_cb;
172 
173 	cb->any_count += cnt;
174 
175 	switch (af) {
176 	case AF_INET:
177 		cb->ip_count += cnt;
178 		return;
179 #ifdef INET6
180 	case AF_INET6:
181 		cb->ip6_count += cnt;
182 		return;
183 #endif
184 	case AF_MPLS:
185 		cb->mpls_count += cnt;
186 		return;
187 	}
188 }
189 
190 static int
191 COMPATNAME(route_attach)(struct socket *so, int proto)
192 {
193 	struct rawcb *rp;
194 	int s, error;
195 
196 	KASSERT(sotorawcb(so) == NULL);
197 	rp = kmem_zalloc(sizeof(*rp), KM_SLEEP);
198 	rp->rcb_len = sizeof(*rp);
199 	so->so_pcb = rp;
200 
201 	s = splsoftnet();
202 	if ((error = raw_attach(so, proto)) == 0) {
203 		rt_adjustcount(rp->rcb_proto.sp_protocol, 1);
204 		rp->rcb_laddr = &COMPATNAME(route_info).ri_src;
205 		rp->rcb_faddr = &COMPATNAME(route_info).ri_dst;
206 	}
207 	splx(s);
208 
209 	if (error) {
210 		kmem_free(rp, sizeof(*rp));
211 		so->so_pcb = NULL;
212 		return error;
213 	}
214 
215 	soisconnected(so);
216 	so->so_options |= SO_USELOOPBACK;
217 	KASSERT(solocked(so));
218 
219 	return error;
220 }
221 
222 static void
223 COMPATNAME(route_detach)(struct socket *so)
224 {
225 	struct rawcb *rp = sotorawcb(so);
226 	int s;
227 
228 	KASSERT(rp != NULL);
229 	KASSERT(solocked(so));
230 
231 	s = splsoftnet();
232 	rt_adjustcount(rp->rcb_proto.sp_protocol, -1);
233 	raw_detach(so);
234 	splx(s);
235 }
236 
237 static int
238 COMPATNAME(route_accept)(struct socket *so, struct sockaddr *nam)
239 {
240 	KASSERT(solocked(so));
241 
242 	panic("route_accept");
243 
244 	return EOPNOTSUPP;
245 }
246 
247 static int
248 COMPATNAME(route_bind)(struct socket *so, struct sockaddr *nam, struct lwp *l)
249 {
250 	KASSERT(solocked(so));
251 
252 	return EOPNOTSUPP;
253 }
254 
255 static int
256 COMPATNAME(route_listen)(struct socket *so, struct lwp *l)
257 {
258 	KASSERT(solocked(so));
259 
260 	return EOPNOTSUPP;
261 }
262 
263 static int
264 COMPATNAME(route_connect)(struct socket *so, struct sockaddr *nam, struct lwp *l)
265 {
266 	KASSERT(solocked(so));
267 
268 	return EOPNOTSUPP;
269 }
270 
271 static int
272 COMPATNAME(route_connect2)(struct socket *so, struct socket *so2)
273 {
274 	KASSERT(solocked(so));
275 
276 	return EOPNOTSUPP;
277 }
278 
279 static int
280 COMPATNAME(route_disconnect)(struct socket *so)
281 {
282 	struct rawcb *rp = sotorawcb(so);
283 	int s;
284 
285 	KASSERT(solocked(so));
286 	KASSERT(rp != NULL);
287 
288 	s = splsoftnet();
289 	soisdisconnected(so);
290 	raw_disconnect(rp);
291 	splx(s);
292 
293 	return 0;
294 }
295 
296 static int
297 COMPATNAME(route_shutdown)(struct socket *so)
298 {
299 	int s;
300 
301 	KASSERT(solocked(so));
302 
303 	/*
304 	 * Mark the connection as being incapable of further input.
305 	 */
306 	s = splsoftnet();
307 	socantsendmore(so);
308 	splx(s);
309 	return 0;
310 }
311 
312 static int
313 COMPATNAME(route_abort)(struct socket *so)
314 {
315 	KASSERT(solocked(so));
316 
317 	panic("route_abort");
318 
319 	return EOPNOTSUPP;
320 }
321 
322 static int
323 COMPATNAME(route_ioctl)(struct socket *so, u_long cmd, void *nam,
324     struct ifnet * ifp)
325 {
326 	return EOPNOTSUPP;
327 }
328 
329 static int
330 COMPATNAME(route_stat)(struct socket *so, struct stat *ub)
331 {
332 	KASSERT(solocked(so));
333 
334 	return 0;
335 }
336 
337 static int
338 COMPATNAME(route_peeraddr)(struct socket *so, struct sockaddr *nam)
339 {
340 	struct rawcb *rp = sotorawcb(so);
341 
342 	KASSERT(solocked(so));
343 	KASSERT(rp != NULL);
344 	KASSERT(nam != NULL);
345 
346 	if (rp->rcb_faddr == NULL)
347 		return ENOTCONN;
348 
349 	raw_setpeeraddr(rp, nam);
350 	return 0;
351 }
352 
353 static int
354 COMPATNAME(route_sockaddr)(struct socket *so, struct sockaddr *nam)
355 {
356 	struct rawcb *rp = sotorawcb(so);
357 
358 	KASSERT(solocked(so));
359 	KASSERT(rp != NULL);
360 	KASSERT(nam != NULL);
361 
362 	if (rp->rcb_faddr == NULL)
363 		return ENOTCONN;
364 
365 	raw_setsockaddr(rp, nam);
366 	return 0;
367 }
368 
369 static int
370 COMPATNAME(route_rcvd)(struct socket *so, int flags, struct lwp *l)
371 {
372 	KASSERT(solocked(so));
373 
374 	return EOPNOTSUPP;
375 }
376 
377 static int
378 COMPATNAME(route_recvoob)(struct socket *so, struct mbuf *m, int flags)
379 {
380 	KASSERT(solocked(so));
381 
382 	return EOPNOTSUPP;
383 }
384 
385 static int
386 COMPATNAME(route_send)(struct socket *so, struct mbuf *m,
387     struct sockaddr *nam, struct mbuf *control, struct lwp *l)
388 {
389 	int error = 0;
390 	int s;
391 
392 	KASSERT(solocked(so));
393 
394 	s = splsoftnet();
395 	error = raw_send(so, m, nam, control, l);
396 	splx(s);
397 
398 	return error;
399 }
400 
401 static int
402 COMPATNAME(route_sendoob)(struct socket *so, struct mbuf *m,
403     struct mbuf *control)
404 {
405 	KASSERT(solocked(so));
406 
407 	m_freem(m);
408 	m_freem(control);
409 
410 	return EOPNOTSUPP;
411 }
412 static int
413 COMPATNAME(route_purgeif)(struct socket *so, struct ifnet *ifp)
414 {
415 
416 	panic("route_purgeif");
417 
418 	return EOPNOTSUPP;
419 }
420 
421 /*ARGSUSED*/
422 int
423 COMPATNAME(route_output)(struct mbuf *m, ...)
424 {
425 	struct sockproto proto = { .sp_family = PF_XROUTE, };
426 	struct rt_xmsghdr *rtm = NULL;
427 	struct rt_xmsghdr *old_rtm = NULL;
428 	struct rtentry *rt = NULL;
429 	struct rtentry *saved_nrt = NULL;
430 	struct rt_addrinfo info;
431 	int len, error = 0;
432 	struct ifnet *ifp = NULL;
433 	struct ifaddr *ifa = NULL;
434 	struct socket *so;
435 	va_list ap;
436 	sa_family_t family;
437 
438 	va_start(ap, m);
439 	so = va_arg(ap, struct socket *);
440 	va_end(ap);
441 
442 #define senderr(e) do { error = e; goto flush;} while (/*CONSTCOND*/ 0)
443 	if (m == NULL || ((m->m_len < sizeof(int32_t)) &&
444 	   (m = m_pullup(m, sizeof(int32_t))) == NULL))
445 		return ENOBUFS;
446 	if ((m->m_flags & M_PKTHDR) == 0)
447 		panic("%s", __func__);
448 	len = m->m_pkthdr.len;
449 	if (len < sizeof(*rtm) ||
450 	    len != mtod(m, struct rt_xmsghdr *)->rtm_msglen) {
451 		info.rti_info[RTAX_DST] = NULL;
452 		senderr(EINVAL);
453 	}
454 	R_Malloc(rtm, struct rt_xmsghdr *, len);
455 	if (rtm == NULL) {
456 		info.rti_info[RTAX_DST] = NULL;
457 		senderr(ENOBUFS);
458 	}
459 	m_copydata(m, 0, len, rtm);
460 	if (rtm->rtm_version != RTM_XVERSION) {
461 		info.rti_info[RTAX_DST] = NULL;
462 		senderr(EPROTONOSUPPORT);
463 	}
464 	rtm->rtm_pid = curproc->p_pid;
465 	memset(&info, 0, sizeof(info));
466 	info.rti_addrs = rtm->rtm_addrs;
467 	if (rt_xaddrs(rtm->rtm_type, (const char *)(rtm + 1), len + (char *)rtm,
468 	    &info)) {
469 		senderr(EINVAL);
470 	}
471 	info.rti_flags = rtm->rtm_flags;
472 #ifdef RTSOCK_DEBUG
473 	if (info.rti_info[RTAX_DST]->sa_family == AF_INET) {
474 		char abuf[INET_ADDRSTRLEN];
475 		printf("%s: extracted info.rti_info[RTAX_DST] %s\n", __func__,
476 		    RT_IN_PRINT(abuf, RTAX_DST));
477 	}
478 #endif /* RTSOCK_DEBUG */
479 	if (info.rti_info[RTAX_DST] == NULL ||
480 	    (info.rti_info[RTAX_DST]->sa_family >= AF_MAX)) {
481 		senderr(EINVAL);
482 	}
483 	if (info.rti_info[RTAX_GATEWAY] != NULL &&
484 	    (info.rti_info[RTAX_GATEWAY]->sa_family >= AF_MAX)) {
485 		senderr(EINVAL);
486 	}
487 
488 	/*
489 	 * Verify that the caller has the appropriate privilege; RTM_GET
490 	 * is the only operation the non-superuser is allowed.
491 	 */
492 	if (kauth_authorize_network(curlwp->l_cred, KAUTH_NETWORK_ROUTE,
493 	    0, rtm, NULL, NULL) != 0)
494 		senderr(EACCES);
495 
496 	switch (rtm->rtm_type) {
497 
498 	case RTM_ADD:
499 		if (info.rti_info[RTAX_GATEWAY] == NULL) {
500 			senderr(EINVAL);
501 		}
502 		error = rtrequest1(rtm->rtm_type, &info, &saved_nrt);
503 		if (error == 0) {
504 			rt_setmetrics(rtm->rtm_inits, rtm, saved_nrt);
505 			rtfree(saved_nrt);
506 		}
507 		break;
508 
509 	case RTM_DELETE:
510 		error = rtrequest1(rtm->rtm_type, &info, &saved_nrt);
511 		if (error == 0) {
512 			rt = saved_nrt;
513 			goto report;
514 		}
515 		break;
516 
517 	case RTM_GET:
518 	case RTM_CHANGE:
519 	case RTM_LOCK:
520                 /* XXX This will mask info.rti_info[RTAX_DST] with
521 		 * info.rti_info[RTAX_NETMASK] before
522                  * searching.  It did not used to do that.  --dyoung
523 		 */
524 		rt = NULL;
525 		error = rtrequest1(RTM_GET, &info, &rt);
526 		if (error != 0)
527 			senderr(error);
528 		if (rtm->rtm_type != RTM_GET) {/* XXX: too grotty */
529 			if (memcmp(info.rti_info[RTAX_DST], rt_getkey(rt),
530 			    info.rti_info[RTAX_DST]->sa_len) != 0)
531 				senderr(ESRCH);
532 			if (info.rti_info[RTAX_NETMASK] == NULL &&
533 			    rt_mask(rt) != NULL)
534 				senderr(ETOOMANYREFS);
535 		}
536 
537 		switch (rtm->rtm_type) {
538 		case RTM_GET:
539 		report:
540 			info.rti_info[RTAX_DST] = rt_getkey(rt);
541 			info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
542 			info.rti_info[RTAX_NETMASK] = rt_mask(rt);
543 			info.rti_info[RTAX_TAG] = rt_gettag(rt);
544 			if ((rtm->rtm_addrs & (RTA_IFP | RTA_IFA)) == 0)
545 				;
546 			else if ((ifp = rt->rt_ifp) != NULL) {
547 				const struct ifaddr *rtifa;
548 				info.rti_info[RTAX_IFP] = ifp->if_dl->ifa_addr;
549                                 /* rtifa used to be simply rt->rt_ifa.
550                                  * If rt->rt_ifa != NULL, then
551                                  * rt_get_ifa() != NULL.  So this
552                                  * ought to still be safe. --dyoung
553 				 */
554 				rtifa = rt_get_ifa(rt);
555 				info.rti_info[RTAX_IFA] = rtifa->ifa_addr;
556 #ifdef RTSOCK_DEBUG
557 				if (info.rti_info[RTAX_IFA]->sa_family ==
558 				    AF_INET) {
559 					char ibuf[INET_ADDRSTRLEN];
560 					char abuf[INET_ADDRSTRLEN];
561 					printf("%s: copying out RTAX_IFA %s "
562 					    "for info.rti_info[RTAX_DST] %s "
563 					    "ifa_getifa %p ifa_seqno %p\n",
564 					    __func__,
565 					    RT_IN_PRINT(ibuf, RTAX_IFA),
566 					    RT_IN_PRINT(abuf, RTAX_DST),
567 					    (void *)rtifa->ifa_getifa,
568 					    rtifa->ifa_seqno);
569 				}
570 #endif /* RTSOCK_DEBUG */
571 				if (ifp->if_flags & IFF_POINTOPOINT) {
572 					info.rti_info[RTAX_BRD] =
573 					    rtifa->ifa_dstaddr;
574 				} else
575 					info.rti_info[RTAX_BRD] = NULL;
576 				rtm->rtm_index = ifp->if_index;
577 			} else {
578 				info.rti_info[RTAX_IFP] = NULL;
579 				info.rti_info[RTAX_IFA] = NULL;
580 			}
581 			(void)rt_msg2(rtm->rtm_type, &info, NULL, NULL, &len);
582 			if (len > rtm->rtm_msglen) {
583 				old_rtm = rtm;
584 				R_Malloc(rtm, struct rt_xmsghdr *, len);
585 				if (rtm == NULL)
586 					senderr(ENOBUFS);
587 				(void)memcpy(rtm, old_rtm, old_rtm->rtm_msglen);
588 			}
589 			(void)rt_msg2(rtm->rtm_type, &info, rtm, NULL, 0);
590 			rtm->rtm_flags = rt->rt_flags;
591 			rtm_setmetrics(rt, rtm);
592 			rtm->rtm_addrs = info.rti_addrs;
593 			break;
594 
595 		case RTM_CHANGE:
596 			/*
597 			 * new gateway could require new ifaddr, ifp;
598 			 * flags may also be different; ifp may be specified
599 			 * by ll sockaddr when protocol address is ambiguous
600 			 */
601 			if ((error = rt_getifa(&info)) != 0)
602 				senderr(error);
603 			if (info.rti_info[RTAX_GATEWAY] &&
604 			    rt_setgate(rt, info.rti_info[RTAX_GATEWAY]))
605 				senderr(EDQUOT);
606 			if (info.rti_info[RTAX_TAG])
607 				rt_settag(rt, info.rti_info[RTAX_TAG]);
608 			/* new gateway could require new ifaddr, ifp;
609 			   flags may also be different; ifp may be specified
610 			   by ll sockaddr when protocol address is ambiguous */
611 			if (info.rti_info[RTAX_IFP] &&
612 			    (ifa = ifa_ifwithnet(info.rti_info[RTAX_IFP])) &&
613 			    (ifp = ifa->ifa_ifp) && (info.rti_info[RTAX_IFA] ||
614 			    info.rti_info[RTAX_GATEWAY])) {
615 				if (info.rti_info[RTAX_IFA] == NULL ||
616 				    (ifa = ifa_ifwithaddr(
617 				    info.rti_info[RTAX_IFA])) == NULL)
618 					ifa = ifaof_ifpforaddr(
619 					    info.rti_info[RTAX_IFA] ?
620 					    info.rti_info[RTAX_IFA] :
621 					    info.rti_info[RTAX_GATEWAY], ifp);
622 			} else if ((info.rti_info[RTAX_IFA] &&
623 			    (ifa = ifa_ifwithaddr(info.rti_info[RTAX_IFA]))) ||
624 			    (info.rti_info[RTAX_GATEWAY] &&
625 			    (ifa = ifa_ifwithroute(rt->rt_flags,
626 			    rt_getkey(rt), info.rti_info[RTAX_GATEWAY])))) {
627 				ifp = ifa->ifa_ifp;
628 			}
629 			if (ifa) {
630 				struct ifaddr *oifa = rt->rt_ifa;
631 				if (oifa != ifa) {
632 					if (oifa && oifa->ifa_rtrequest) {
633 						oifa->ifa_rtrequest(RTM_DELETE,
634 						    rt, &info);
635 					}
636 					rt_replace_ifa(rt, ifa);
637 					rt->rt_ifp = ifp;
638 				}
639 			}
640 			if (ifp && rt->rt_ifp != ifp)
641 				rt->rt_ifp = ifp;
642 			rt_setmetrics(rtm->rtm_inits, rtm, rt);
643 			if (rt->rt_flags != info.rti_flags)
644 				rt->rt_flags = (info.rti_flags & ~PRESERVED_RTF)
645 				    | (rt->rt_flags & PRESERVED_RTF);
646 			if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest)
647 				rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, &info);
648 			/*FALLTHROUGH*/
649 		case RTM_LOCK:
650 			rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits);
651 			rt->rt_rmx.rmx_locks |=
652 			    (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks);
653 			break;
654 		}
655 		break;
656 
657 	default:
658 		senderr(EOPNOTSUPP);
659 	}
660 
661 flush:
662 	if (rtm) {
663 		if (error)
664 			rtm->rtm_errno = error;
665 		else
666 			rtm->rtm_flags |= RTF_DONE;
667 	}
668 	family = info.rti_info[RTAX_DST] ? info.rti_info[RTAX_DST]->sa_family :
669 	    0;
670 	/* We cannot free old_rtm until we have stopped using the
671 	 * pointers in info, some of which may point to sockaddrs
672 	 * in old_rtm.
673 	 */
674 	if (old_rtm != NULL)
675 		Free(old_rtm);
676 	if (rt)
677 		rtfree(rt);
678     {
679 	struct rawcb *rp = NULL;
680 	/*
681 	 * Check to see if we don't want our own messages.
682 	 */
683 	if ((so->so_options & SO_USELOOPBACK) == 0) {
684 		if (COMPATNAME(route_info).ri_cb.any_count <= 1) {
685 			if (rtm)
686 				Free(rtm);
687 			m_freem(m);
688 			return error;
689 		}
690 		/* There is another listener, so construct message */
691 		rp = sotorawcb(so);
692 	}
693 	if (rtm) {
694 		m_copyback(m, 0, rtm->rtm_msglen, rtm);
695 		if (m->m_pkthdr.len < rtm->rtm_msglen) {
696 			m_freem(m);
697 			m = NULL;
698 		} else if (m->m_pkthdr.len > rtm->rtm_msglen)
699 			m_adj(m, rtm->rtm_msglen - m->m_pkthdr.len);
700 		Free(rtm);
701 	}
702 	if (rp)
703 		rp->rcb_proto.sp_family = 0; /* Avoid us */
704 	if (family)
705 		proto.sp_protocol = family;
706 	if (m)
707 		raw_input(m, &proto, &COMPATNAME(route_info).ri_src,
708 		    &COMPATNAME(route_info).ri_dst);
709 	if (rp)
710 		rp->rcb_proto.sp_family = PF_XROUTE;
711     }
712 	return error;
713 }
714 
715 static void
716 rt_setmetrics(int which, const struct rt_xmsghdr *in, struct rtentry *out)
717 {
718 #define metric(f, e) if (which & (f)) out->rt_rmx.e = in->rtm_rmx.e;
719 	metric(RTV_RPIPE, rmx_recvpipe);
720 	metric(RTV_SPIPE, rmx_sendpipe);
721 	metric(RTV_SSTHRESH, rmx_ssthresh);
722 	metric(RTV_RTT, rmx_rtt);
723 	metric(RTV_RTTVAR, rmx_rttvar);
724 	metric(RTV_HOPCOUNT, rmx_hopcount);
725 	metric(RTV_MTU, rmx_mtu);
726 #undef metric
727 	if (which & RTV_EXPIRE) {
728 		out->rt_rmx.rmx_expire = in->rtm_rmx.rmx_expire ?
729 		    time_wall_to_mono(in->rtm_rmx.rmx_expire) : 0;
730 	}
731 }
732 
733 static void
734 rtm_setmetrics(const struct rtentry *in, struct rt_xmsghdr *out)
735 {
736 #define metric(e) out->rtm_rmx.e = in->rt_rmx.e;
737 	metric(rmx_recvpipe);
738 	metric(rmx_sendpipe);
739 	metric(rmx_ssthresh);
740 	metric(rmx_rtt);
741 	metric(rmx_rttvar);
742 	metric(rmx_hopcount);
743 	metric(rmx_mtu);
744 #undef metric
745 	out->rtm_rmx.rmx_expire = in->rt_rmx.rmx_expire ?
746 	    time_mono_to_wall(in->rt_rmx.rmx_expire) : 0;
747 }
748 
749 static int
750 rt_xaddrs(u_char rtmtype, const char *cp, const char *cplim,
751     struct rt_addrinfo *rtinfo)
752 {
753 	const struct sockaddr *sa = NULL;	/* Quell compiler warning */
754 	int i;
755 
756 	for (i = 0; i < RTAX_MAX && cp < cplim; i++) {
757 		if ((rtinfo->rti_addrs & (1 << i)) == 0)
758 			continue;
759 		rtinfo->rti_info[i] = sa = (const struct sockaddr *)cp;
760 		RT_XADVANCE(cp, sa);
761 	}
762 
763 	/*
764 	 * Check for extra addresses specified, except RTM_GET asking
765 	 * for interface info.
766 	 */
767 	if (rtmtype == RTM_GET) {
768 		if (((rtinfo->rti_addrs &
769 		    (~((1 << RTAX_IFP) | (1 << RTAX_IFA)))) & (~0 << i)) != 0)
770 			return 1;
771 	} else if ((rtinfo->rti_addrs & (~0 << i)) != 0)
772 		return 1;
773 	/* Check for bad data length.  */
774 	if (cp != cplim) {
775 		if (i == RTAX_NETMASK + 1 && sa != NULL &&
776 		    cp - RT_XROUNDUP(sa->sa_len) + sa->sa_len == cplim)
777 			/*
778 			 * The last sockaddr was info.rti_info[RTAX_NETMASK].
779 			 * We accept this for now for the sake of old
780 			 * binaries or third party softwares.
781 			 */
782 			;
783 		else
784 			return 1;
785 	}
786 	return 0;
787 }
788 
789 static int
790 rt_getlen(int type)
791 {
792 #ifndef COMPAT_RTSOCK
793 	CTASSERT(__alignof(struct ifa_msghdr) >= sizeof(uint64_t));
794 	CTASSERT(__alignof(struct if_msghdr) >= sizeof(uint64_t));
795 	CTASSERT(__alignof(struct if_announcemsghdr) >= sizeof(uint64_t));
796 	CTASSERT(__alignof(struct rt_msghdr) >= sizeof(uint64_t));
797 #endif
798 
799 	switch (type) {
800 	case RTM_DELADDR:
801 	case RTM_NEWADDR:
802 	case RTM_CHGADDR:
803 		return sizeof(struct ifa_xmsghdr);
804 
805 	case RTM_OOIFINFO:
806 #ifdef COMPAT_14
807 		return sizeof(struct if_msghdr14);
808 #else
809 #ifdef DIAGNOSTIC
810 		printf("RTM_OOIFINFO\n");
811 #endif
812 		return -1;
813 #endif
814 	case RTM_OIFINFO:
815 #ifdef COMPAT_50
816 		return sizeof(struct if_msghdr50);
817 #else
818 #ifdef DIAGNOSTIC
819 		printf("RTM_OIFINFO\n");
820 #endif
821 		return -1;
822 #endif
823 
824 	case RTM_IFINFO:
825 		return sizeof(struct if_xmsghdr);
826 
827 	case RTM_IFANNOUNCE:
828 	case RTM_IEEE80211:
829 		return sizeof(struct if_xannouncemsghdr);
830 
831 	default:
832 		return sizeof(struct rt_xmsghdr);
833 	}
834 }
835 
836 
837 struct mbuf *
838 COMPATNAME(rt_msg1)(int type, struct rt_addrinfo *rtinfo, void *data, int datalen)
839 {
840 	struct rt_xmsghdr *rtm;
841 	struct mbuf *m;
842 	int i;
843 	const struct sockaddr *sa;
844 	int len, dlen;
845 
846 	m = m_gethdr(M_DONTWAIT, MT_DATA);
847 	if (m == NULL)
848 		return m;
849 	MCLAIM(m, &COMPATNAME(routedomain).dom_mowner);
850 
851 	if ((len = rt_getlen(type)) == -1)
852 		goto out;
853 	if (len > MHLEN + MLEN)
854 		panic("%s: message too long", __func__);
855 	else if (len > MHLEN) {
856 		m->m_next = m_get(M_DONTWAIT, MT_DATA);
857 		if (m->m_next == NULL)
858 			goto out;
859 		MCLAIM(m->m_next, m->m_owner);
860 		m->m_pkthdr.len = len;
861 		m->m_len = MHLEN;
862 		m->m_next->m_len = len - MHLEN;
863 	} else {
864 		m->m_pkthdr.len = m->m_len = len;
865 	}
866 	m->m_pkthdr.rcvif = NULL;
867 	m_copyback(m, 0, datalen, data);
868 	if (len > datalen)
869 		(void)memset(mtod(m, char *) + datalen, 0, len - datalen);
870 	rtm = mtod(m, struct rt_xmsghdr *);
871 	for (i = 0; i < RTAX_MAX; i++) {
872 		if ((sa = rtinfo->rti_info[i]) == NULL)
873 			continue;
874 		rtinfo->rti_addrs |= (1 << i);
875 		dlen = RT_XROUNDUP(sa->sa_len);
876 		m_copyback(m, len, sa->sa_len, sa);
877 		if (dlen != sa->sa_len) {
878 			/*
879 			 * Up to 6 + 1 nul's since roundup is to
880 			 * sizeof(uint64_t) (8 bytes)
881 			 */
882 			m_copyback(m, len + sa->sa_len,
883 			    dlen - sa->sa_len, "\0\0\0\0\0\0");
884 		}
885 		len += dlen;
886 	}
887 	if (m->m_pkthdr.len != len)
888 		goto out;
889 	rtm->rtm_msglen = len;
890 	rtm->rtm_version = RTM_XVERSION;
891 	rtm->rtm_type = type;
892 	return m;
893 out:
894 	m_freem(m);
895 	return NULL;
896 }
897 
898 /*
899  * rt_msg2
900  *
901  *	 fills 'cp' or 'w'.w_tmem with the routing socket message and
902  *		returns the length of the message in 'lenp'.
903  *
904  * if walkarg is 0, cp is expected to be 0 or a buffer large enough to hold
905  *	the message
906  * otherwise walkarg's w_needed is updated and if the user buffer is
907  *	specified and w_needed indicates space exists the information is copied
908  *	into the temp space (w_tmem). w_tmem is [re]allocated if necessary,
909  *	if the allocation fails ENOBUFS is returned.
910  */
911 static int
912 rt_msg2(int type, struct rt_addrinfo *rtinfo, void *cpv, struct rt_walkarg *w,
913 	int *lenp)
914 {
915 	int i;
916 	int len, dlen, second_time = 0;
917 	char *cp0, *cp = cpv;
918 
919 	rtinfo->rti_addrs = 0;
920 again:
921 	if ((len = rt_getlen(type)) == -1)
922 		return EINVAL;
923 
924 	if ((cp0 = cp) != NULL)
925 		cp += len;
926 	for (i = 0; i < RTAX_MAX; i++) {
927 		const struct sockaddr *sa;
928 
929 		if ((sa = rtinfo->rti_info[i]) == NULL)
930 			continue;
931 		rtinfo->rti_addrs |= (1 << i);
932 		dlen = RT_XROUNDUP(sa->sa_len);
933 		if (cp) {
934 			int diff = dlen - sa->sa_len;
935 			(void)memcpy(cp, sa, (size_t)sa->sa_len);
936 			cp += sa->sa_len;
937 			if (diff > 0) {
938 				(void)memset(cp, 0, (size_t)diff);
939 				cp += diff;
940 			}
941 		}
942 		len += dlen;
943 	}
944 	if (cp == NULL && w != NULL && !second_time) {
945 		struct rt_walkarg *rw = w;
946 
947 		rw->w_needed += len;
948 		if (rw->w_needed <= 0 && rw->w_where) {
949 			if (rw->w_tmemsize < len) {
950 				if (rw->w_tmem)
951 					free(rw->w_tmem, M_RTABLE);
952 				rw->w_tmem = malloc(len, M_RTABLE, M_NOWAIT);
953 				if (rw->w_tmem)
954 					rw->w_tmemsize = len;
955 				else
956 					rw->w_tmemsize = 0;
957 			}
958 			if (rw->w_tmem) {
959 				cp = rw->w_tmem;
960 				second_time = 1;
961 				goto again;
962 			} else {
963 				rw->w_tmemneeded = len;
964 				return ENOBUFS;
965 			}
966 		}
967 	}
968 	if (cp) {
969 		struct rt_xmsghdr *rtm = (struct rt_xmsghdr *)cp0;
970 
971 		rtm->rtm_version = RTM_XVERSION;
972 		rtm->rtm_type = type;
973 		rtm->rtm_msglen = len;
974 	}
975 	if (lenp)
976 		*lenp = len;
977 	return 0;
978 }
979 
980 /*
981  * This routine is called to generate a message from the routing
982  * socket indicating that a redirect has occurred, a routing lookup
983  * has failed, or that a protocol has detected timeouts to a particular
984  * destination.
985  */
986 void
987 COMPATNAME(rt_missmsg)(int type, const struct rt_addrinfo *rtinfo, int flags,
988     int error)
989 {
990 	struct rt_xmsghdr rtm;
991 	struct mbuf *m;
992 	const struct sockaddr *sa = rtinfo->rti_info[RTAX_DST];
993 	struct rt_addrinfo info = *rtinfo;
994 
995 	COMPATCALL(rt_missmsg, (type, rtinfo, flags, error));
996 	if (COMPATNAME(route_info).ri_cb.any_count == 0)
997 		return;
998 	memset(&rtm, 0, sizeof(rtm));
999 	rtm.rtm_flags = RTF_DONE | flags;
1000 	rtm.rtm_errno = error;
1001 	m = COMPATNAME(rt_msg1)(type, &info, &rtm, sizeof(rtm));
1002 	if (m == NULL)
1003 		return;
1004 	mtod(m, struct rt_xmsghdr *)->rtm_addrs = info.rti_addrs;
1005 	COMPATNAME(route_enqueue)(m, sa ? sa->sa_family : 0);
1006 }
1007 
1008 /*
1009  * This routine is called to generate a message from the routing
1010  * socket indicating that the status of a network interface has changed.
1011  */
1012 void
1013 COMPATNAME(rt_ifmsg)(struct ifnet *ifp)
1014 {
1015 	struct if_xmsghdr ifm;
1016 	struct mbuf *m;
1017 	struct rt_addrinfo info;
1018 
1019 	COMPATCALL(rt_ifmsg, (ifp));
1020 	if (COMPATNAME(route_info).ri_cb.any_count == 0)
1021 		return;
1022 	(void)memset(&info, 0, sizeof(info));
1023 	(void)memset(&ifm, 0, sizeof(ifm));
1024 	ifm.ifm_index = ifp->if_index;
1025 	ifm.ifm_flags = ifp->if_flags;
1026 	ifm.ifm_data = ifp->if_data;
1027 	ifm.ifm_addrs = 0;
1028 	m = COMPATNAME(rt_msg1)(RTM_IFINFO, &info, &ifm, sizeof(ifm));
1029 	if (m == NULL)
1030 		return;
1031 	COMPATNAME(route_enqueue)(m, 0);
1032 #ifdef COMPAT_14
1033 	compat_14_rt_oifmsg(ifp);
1034 #endif
1035 #ifdef COMPAT_50
1036 	compat_50_rt_oifmsg(ifp);
1037 #endif
1038 }
1039 
1040 
1041 /*
1042  * This is called to generate messages from the routing socket
1043  * indicating a network interface has had addresses associated with it.
1044  * if we ever reverse the logic and replace messages TO the routing
1045  * socket indicate a request to configure interfaces, then it will
1046  * be unnecessary as the routing socket will automatically generate
1047  * copies of it.
1048  */
1049 void
1050 COMPATNAME(rt_newaddrmsg)(int cmd, struct ifaddr *ifa, int error,
1051     struct rtentry *rt)
1052 {
1053 #define	cmdpass(__cmd, __pass)	(((__cmd) << 2) | (__pass))
1054 	struct rt_addrinfo info;
1055 	const struct sockaddr *sa;
1056 	int pass;
1057 	struct mbuf *m;
1058 	struct ifnet *ifp;
1059 	struct rt_xmsghdr rtm;
1060 	struct ifa_xmsghdr ifam;
1061 	int ncmd;
1062 
1063 	KASSERT(ifa != NULL);
1064 	ifp = ifa->ifa_ifp;
1065 #ifdef SCTP
1066 	if (cmd == RTM_ADD) {
1067 		sctp_add_ip_address(ifa);
1068 	} else if (cmd == RTM_DELETE) {
1069 		sctp_delete_ip_address(ifa);
1070 	}
1071 #endif
1072 
1073 	COMPATCALL(rt_newaddrmsg, (cmd, ifa, error, rt));
1074 	if (COMPATNAME(route_info).ri_cb.any_count == 0)
1075 		return;
1076 	for (pass = 1; pass < 3; pass++) {
1077 		memset(&info, 0, sizeof(info));
1078 		switch (cmdpass(cmd, pass)) {
1079 		case cmdpass(RTM_ADD, 1):
1080 		case cmdpass(RTM_CHANGE, 1):
1081 		case cmdpass(RTM_DELETE, 2):
1082 		case cmdpass(RTM_NEWADDR, 1):
1083 		case cmdpass(RTM_DELADDR, 1):
1084 		case cmdpass(RTM_CHGADDR, 1):
1085 			switch (cmd) {
1086 			case RTM_ADD:
1087 				ncmd = RTM_NEWADDR;
1088 				break;
1089 			case RTM_DELETE:
1090 				ncmd = RTM_DELADDR;
1091 				break;
1092 			case RTM_CHANGE:
1093 				ncmd = RTM_CHGADDR;
1094 				break;
1095 			default:
1096 				ncmd = cmd;
1097 			}
1098 			info.rti_info[RTAX_IFA] = sa = ifa->ifa_addr;
1099 			KASSERT(ifp->if_dl != NULL);
1100 			info.rti_info[RTAX_IFP] = ifp->if_dl->ifa_addr;
1101 			info.rti_info[RTAX_NETMASK] = ifa->ifa_netmask;
1102 			info.rti_info[RTAX_BRD] = ifa->ifa_dstaddr;
1103 			memset(&ifam, 0, sizeof(ifam));
1104 			ifam.ifam_index = ifp->if_index;
1105 			ifam.ifam_metric = ifa->ifa_metric;
1106 			ifam.ifam_flags = ifa->ifa_flags;
1107 			m = COMPATNAME(rt_msg1)(ncmd, &info, &ifam, sizeof(ifam));
1108 			if (m == NULL)
1109 				continue;
1110 			mtod(m, struct ifa_xmsghdr *)->ifam_addrs =
1111 			    info.rti_addrs;
1112 			break;
1113 		case cmdpass(RTM_ADD, 2):
1114 		case cmdpass(RTM_CHANGE, 2):
1115 		case cmdpass(RTM_DELETE, 1):
1116 			if (rt == NULL)
1117 				continue;
1118 			info.rti_info[RTAX_NETMASK] = rt_mask(rt);
1119 			info.rti_info[RTAX_DST] = sa = rt_getkey(rt);
1120 			info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
1121 			memset(&rtm, 0, sizeof(rtm));
1122 			rtm.rtm_index = ifp->if_index;
1123 			rtm.rtm_flags |= rt->rt_flags;
1124 			rtm.rtm_errno = error;
1125 			m = COMPATNAME(rt_msg1)(cmd, &info, &rtm, sizeof(rtm));
1126 			if (m == NULL)
1127 				continue;
1128 			mtod(m, struct rt_xmsghdr *)->rtm_addrs = info.rti_addrs;
1129 			break;
1130 		default:
1131 			continue;
1132 		}
1133 #ifdef DIAGNOSTIC
1134 		if (m == NULL)
1135 			panic("%s: called with wrong command", __func__);
1136 #endif
1137 		COMPATNAME(route_enqueue)(m, sa ? sa->sa_family : 0);
1138 	}
1139 #undef cmdpass
1140 }
1141 
1142 static struct mbuf *
1143 rt_makeifannouncemsg(struct ifnet *ifp, int type, int what,
1144     struct rt_addrinfo *info)
1145 {
1146 	struct if_xannouncemsghdr ifan;
1147 
1148 	memset(info, 0, sizeof(*info));
1149 	memset(&ifan, 0, sizeof(ifan));
1150 	ifan.ifan_index = ifp->if_index;
1151 	strlcpy(ifan.ifan_name, ifp->if_xname, sizeof(ifan.ifan_name));
1152 	ifan.ifan_what = what;
1153 	return COMPATNAME(rt_msg1)(type, info, &ifan, sizeof(ifan));
1154 }
1155 
1156 /*
1157  * This is called to generate routing socket messages indicating
1158  * network interface arrival and departure.
1159  */
1160 void
1161 COMPATNAME(rt_ifannouncemsg)(struct ifnet *ifp, int what)
1162 {
1163 	struct mbuf *m;
1164 	struct rt_addrinfo info;
1165 
1166 	COMPATCALL(rt_ifannouncemsg, (ifp, what));
1167 	if (COMPATNAME(route_info).ri_cb.any_count == 0)
1168 		return;
1169 	m = rt_makeifannouncemsg(ifp, RTM_IFANNOUNCE, what, &info);
1170 	if (m == NULL)
1171 		return;
1172 	COMPATNAME(route_enqueue)(m, 0);
1173 }
1174 
1175 /*
1176  * This is called to generate routing socket messages indicating
1177  * IEEE80211 wireless events.
1178  * XXX we piggyback on the RTM_IFANNOUNCE msg format in a clumsy way.
1179  */
1180 void
1181 COMPATNAME(rt_ieee80211msg)(struct ifnet *ifp, int what, void *data,
1182 	size_t data_len)
1183 {
1184 	struct mbuf *m;
1185 	struct rt_addrinfo info;
1186 
1187 	COMPATCALL(rt_ieee80211msg, (ifp, what, data, data_len));
1188 	if (COMPATNAME(route_info).ri_cb.any_count == 0)
1189 		return;
1190 	m = rt_makeifannouncemsg(ifp, RTM_IEEE80211, what, &info);
1191 	if (m == NULL)
1192 		return;
1193 	/*
1194 	 * Append the ieee80211 data.  Try to stick it in the
1195 	 * mbuf containing the ifannounce msg; otherwise allocate
1196 	 * a new mbuf and append.
1197 	 *
1198 	 * NB: we assume m is a single mbuf.
1199 	 */
1200 	if (data_len > M_TRAILINGSPACE(m)) {
1201 		struct mbuf *n = m_get(M_NOWAIT, MT_DATA);
1202 		if (n == NULL) {
1203 			m_freem(m);
1204 			return;
1205 		}
1206 		(void)memcpy(mtod(n, void *), data, data_len);
1207 		n->m_len = data_len;
1208 		m->m_next = n;
1209 	} else if (data_len > 0) {
1210 		(void)memcpy(mtod(m, uint8_t *) + m->m_len, data, data_len);
1211 		m->m_len += data_len;
1212 	}
1213 	if (m->m_flags & M_PKTHDR)
1214 		m->m_pkthdr.len += data_len;
1215 	mtod(m, struct if_xannouncemsghdr *)->ifan_msglen += data_len;
1216 	COMPATNAME(route_enqueue)(m, 0);
1217 }
1218 
1219 /*
1220  * This is used in dumping the kernel table via sysctl().
1221  */
1222 static int
1223 sysctl_dumpentry(struct rtentry *rt, void *v)
1224 {
1225 	struct rt_walkarg *w = v;
1226 	int error = 0, size;
1227 	struct rt_addrinfo info;
1228 
1229 	if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg))
1230 		return 0;
1231 	memset(&info, 0, sizeof(info));
1232 	info.rti_info[RTAX_DST] = rt_getkey(rt);
1233 	info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
1234 	info.rti_info[RTAX_NETMASK] = rt_mask(rt);
1235 	info.rti_info[RTAX_TAG] = rt_gettag(rt);
1236 	if (rt->rt_ifp) {
1237 		const struct ifaddr *rtifa;
1238 		info.rti_info[RTAX_IFP] = rt->rt_ifp->if_dl->ifa_addr;
1239 		/* rtifa used to be simply rt->rt_ifa.  If rt->rt_ifa != NULL,
1240 		 * then rt_get_ifa() != NULL.  So this ought to still be safe.
1241 		 * --dyoung
1242 		 */
1243 		rtifa = rt_get_ifa(rt);
1244 		info.rti_info[RTAX_IFA] = rtifa->ifa_addr;
1245 		if (rt->rt_ifp->if_flags & IFF_POINTOPOINT)
1246 			info.rti_info[RTAX_BRD] = rtifa->ifa_dstaddr;
1247 	}
1248 	if ((error = rt_msg2(RTM_GET, &info, 0, w, &size)))
1249 		return error;
1250 	if (w->w_where && w->w_tmem && w->w_needed <= 0) {
1251 		struct rt_xmsghdr *rtm = (struct rt_xmsghdr *)w->w_tmem;
1252 
1253 		rtm->rtm_flags = rt->rt_flags;
1254 		rtm->rtm_use = rt->rt_use;
1255 		rtm_setmetrics(rt, rtm);
1256 		KASSERT(rt->rt_ifp != NULL);
1257 		rtm->rtm_index = rt->rt_ifp->if_index;
1258 		rtm->rtm_errno = rtm->rtm_pid = rtm->rtm_seq = 0;
1259 		rtm->rtm_addrs = info.rti_addrs;
1260 		if ((error = copyout(rtm, w->w_where, size)) != 0)
1261 			w->w_where = NULL;
1262 		else
1263 			w->w_where = (char *)w->w_where + size;
1264 	}
1265 	return error;
1266 }
1267 
1268 static int
1269 sysctl_iflist(int af, struct rt_walkarg *w, int type)
1270 {
1271 	struct ifnet *ifp;
1272 	struct ifaddr *ifa;
1273 	struct	rt_addrinfo info;
1274 	int	len, error = 0;
1275 
1276 	memset(&info, 0, sizeof(info));
1277 	IFNET_FOREACH(ifp) {
1278 		if (w->w_arg && w->w_arg != ifp->if_index)
1279 			continue;
1280 		if (IFADDR_EMPTY(ifp))
1281 			continue;
1282 		info.rti_info[RTAX_IFP] = ifp->if_dl->ifa_addr;
1283 		switch (type) {
1284 		case NET_RT_IFLIST:
1285 			error = rt_msg2(RTM_IFINFO, &info, NULL, w, &len);
1286 			break;
1287 #ifdef COMPAT_14
1288 		case NET_RT_OOIFLIST:
1289 			error = rt_msg2(RTM_OOIFINFO, &info, NULL, w, &len);
1290 			break;
1291 #endif
1292 #ifdef COMPAT_50
1293 		case NET_RT_OIFLIST:
1294 			error = rt_msg2(RTM_OIFINFO, &info, NULL, w, &len);
1295 			break;
1296 #endif
1297 		default:
1298 			panic("sysctl_iflist(1)");
1299 		}
1300 		if (error)
1301 			return error;
1302 		info.rti_info[RTAX_IFP] = NULL;
1303 		if (w->w_where && w->w_tmem && w->w_needed <= 0) {
1304 			switch (type) {
1305 			case NET_RT_IFLIST: {
1306 				struct if_xmsghdr *ifm;
1307 
1308 				ifm = (struct if_xmsghdr *)w->w_tmem;
1309 				ifm->ifm_index = ifp->if_index;
1310 				ifm->ifm_flags = ifp->if_flags;
1311 				ifm->ifm_data = ifp->if_data;
1312 				ifm->ifm_addrs = info.rti_addrs;
1313 				error = copyout(ifm, w->w_where, len);
1314 				if (error)
1315 					return error;
1316 				w->w_where = (char *)w->w_where + len;
1317 				break;
1318 			}
1319 
1320 #ifdef COMPAT_14
1321 			case NET_RT_OOIFLIST:
1322 				error = compat_14_iflist(ifp, w, &info, len);
1323 				if (error)
1324 					return error;
1325 				break;
1326 #endif
1327 #ifdef COMPAT_50
1328 			case NET_RT_OIFLIST:
1329 				error = compat_50_iflist(ifp, w, &info, len);
1330 				if (error)
1331 					return error;
1332 				break;
1333 #endif
1334 			default:
1335 				panic("sysctl_iflist(2)");
1336 			}
1337 		}
1338 		IFADDR_FOREACH(ifa, ifp) {
1339 			if (af && af != ifa->ifa_addr->sa_family)
1340 				continue;
1341 			info.rti_info[RTAX_IFA] = ifa->ifa_addr;
1342 			info.rti_info[RTAX_NETMASK] = ifa->ifa_netmask;
1343 			info.rti_info[RTAX_BRD] = ifa->ifa_dstaddr;
1344 			if ((error = rt_msg2(RTM_NEWADDR, &info, 0, w, &len)))
1345 				return error;
1346 			if (w->w_where && w->w_tmem && w->w_needed <= 0) {
1347 				struct ifa_xmsghdr *ifam;
1348 
1349 				ifam = (struct ifa_xmsghdr *)w->w_tmem;
1350 				ifam->ifam_index = ifa->ifa_ifp->if_index;
1351 				ifam->ifam_flags = ifa->ifa_flags;
1352 				ifam->ifam_metric = ifa->ifa_metric;
1353 				ifam->ifam_addrs = info.rti_addrs;
1354 				error = copyout(w->w_tmem, w->w_where, len);
1355 				if (error)
1356 					return error;
1357 				w->w_where = (char *)w->w_where + len;
1358 			}
1359 		}
1360 		info.rti_info[RTAX_IFA] = info.rti_info[RTAX_NETMASK] =
1361 		    info.rti_info[RTAX_BRD] = NULL;
1362 	}
1363 	return 0;
1364 }
1365 
1366 static int
1367 sysctl_rtable(SYSCTLFN_ARGS)
1368 {
1369 	void 	*where = oldp;
1370 	size_t	*given = oldlenp;
1371 	int	i, s, error = EINVAL;
1372 	u_char  af;
1373 	struct	rt_walkarg w;
1374 
1375 	if (namelen == 1 && name[0] == CTL_QUERY)
1376 		return sysctl_query(SYSCTLFN_CALL(rnode));
1377 
1378 	if (newp)
1379 		return EPERM;
1380 	if (namelen != 3)
1381 		return EINVAL;
1382 	af = name[0];
1383 	w.w_tmemneeded = 0;
1384 	w.w_tmemsize = 0;
1385 	w.w_tmem = NULL;
1386 again:
1387 	/* we may return here if a later [re]alloc of the t_mem buffer fails */
1388 	if (w.w_tmemneeded) {
1389 		w.w_tmem = malloc(w.w_tmemneeded, M_RTABLE, M_WAITOK);
1390 		w.w_tmemsize = w.w_tmemneeded;
1391 		w.w_tmemneeded = 0;
1392 	}
1393 	w.w_op = name[1];
1394 	w.w_arg = name[2];
1395 	w.w_given = *given;
1396 	w.w_needed = 0 - w.w_given;
1397 	w.w_where = where;
1398 
1399 	s = splsoftnet();
1400 	switch (w.w_op) {
1401 
1402 	case NET_RT_DUMP:
1403 	case NET_RT_FLAGS:
1404 		for (i = 1; i <= AF_MAX; i++)
1405 			if ((af == 0 || af == i) &&
1406 			    (error = rt_walktree(i, sysctl_dumpentry, &w)))
1407 				break;
1408 		break;
1409 
1410 #ifdef COMPAT_14
1411 	case NET_RT_OOIFLIST:
1412 		error = sysctl_iflist(af, &w, w.w_op);
1413 		break;
1414 #endif
1415 #ifdef COMPAT_50
1416 	case NET_RT_OIFLIST:
1417 		error = sysctl_iflist(af, &w, w.w_op);
1418 		break;
1419 #endif
1420 	case NET_RT_IFLIST:
1421 		error = sysctl_iflist(af, &w, w.w_op);
1422 		break;
1423 	}
1424 	splx(s);
1425 
1426 	/* check to see if we couldn't allocate memory with NOWAIT */
1427 	if (error == ENOBUFS && w.w_tmem == 0 && w.w_tmemneeded)
1428 		goto again;
1429 
1430 	if (w.w_tmem)
1431 		free(w.w_tmem, M_RTABLE);
1432 	w.w_needed += w.w_given;
1433 	if (where) {
1434 		*given = (char *)w.w_where - (char *)where;
1435 		if (*given < w.w_needed)
1436 			return ENOMEM;
1437 	} else {
1438 		*given = (11 * w.w_needed) / 10;
1439 	}
1440 	return error;
1441 }
1442 
1443 /*
1444  * Routing message software interrupt routine
1445  */
1446 static void
1447 COMPATNAME(route_intr)(void *cookie)
1448 {
1449 	struct sockproto proto = { .sp_family = PF_XROUTE, };
1450 	struct route_info * const ri = &COMPATNAME(route_info);
1451 	struct mbuf *m;
1452 	int s;
1453 
1454 	mutex_enter(softnet_lock);
1455 	KERNEL_LOCK(1, NULL);
1456 	while (!IF_IS_EMPTY(&ri->ri_intrq)) {
1457 		s = splnet();
1458 		IF_DEQUEUE(&ri->ri_intrq, m);
1459 		splx(s);
1460 		if (m == NULL)
1461 			break;
1462 		proto.sp_protocol = M_GETCTX(m, uintptr_t);
1463 		raw_input(m, &proto, &ri->ri_src, &ri->ri_dst);
1464 	}
1465 	KERNEL_UNLOCK_ONE(NULL);
1466 	mutex_exit(softnet_lock);
1467 }
1468 
1469 /*
1470  * Enqueue a message to the software interrupt routine.
1471  */
1472 void
1473 COMPATNAME(route_enqueue)(struct mbuf *m, int family)
1474 {
1475 	struct route_info * const ri = &COMPATNAME(route_info);
1476 	int s, wasempty;
1477 
1478 	s = splnet();
1479 	if (IF_QFULL(&ri->ri_intrq)) {
1480 		IF_DROP(&ri->ri_intrq);
1481 		m_freem(m);
1482 	} else {
1483 		wasempty = IF_IS_EMPTY(&ri->ri_intrq);
1484 		M_SETCTX(m, (uintptr_t)family);
1485 		IF_ENQUEUE(&ri->ri_intrq, m);
1486 		if (wasempty)
1487 			softint_schedule(ri->ri_sih);
1488 	}
1489 	splx(s);
1490 }
1491 
1492 static void
1493 COMPATNAME(route_init)(void)
1494 {
1495 	struct route_info * const ri = &COMPATNAME(route_info);
1496 
1497 #ifndef COMPAT_RTSOCK
1498 	rt_init();
1499 #endif
1500 
1501 	sysctl_net_route_setup(NULL);
1502 	ri->ri_intrq.ifq_maxlen = ri->ri_maxqlen;
1503 	ri->ri_sih = softint_establish(SOFTINT_NET | SOFTINT_MPSAFE,
1504 	    COMPATNAME(route_intr), NULL);
1505 }
1506 
1507 /*
1508  * Definitions of protocols supported in the ROUTE domain.
1509  */
1510 #ifndef COMPAT_RTSOCK
1511 PR_WRAP_USRREQS(route);
1512 #else
1513 PR_WRAP_USRREQS(compat_50_route);
1514 #endif
1515 
1516 static const struct pr_usrreqs route_usrreqs = {
1517 	.pr_attach	= COMPATNAME(route_attach_wrapper),
1518 	.pr_detach	= COMPATNAME(route_detach_wrapper),
1519 	.pr_accept	= COMPATNAME(route_accept_wrapper),
1520 	.pr_bind	= COMPATNAME(route_bind_wrapper),
1521 	.pr_listen	= COMPATNAME(route_listen_wrapper),
1522 	.pr_connect	= COMPATNAME(route_connect_wrapper),
1523 	.pr_connect2	= COMPATNAME(route_connect2_wrapper),
1524 	.pr_disconnect	= COMPATNAME(route_disconnect_wrapper),
1525 	.pr_shutdown	= COMPATNAME(route_shutdown_wrapper),
1526 	.pr_abort	= COMPATNAME(route_abort_wrapper),
1527 	.pr_ioctl	= COMPATNAME(route_ioctl_wrapper),
1528 	.pr_stat	= COMPATNAME(route_stat_wrapper),
1529 	.pr_peeraddr	= COMPATNAME(route_peeraddr_wrapper),
1530 	.pr_sockaddr	= COMPATNAME(route_sockaddr_wrapper),
1531 	.pr_rcvd	= COMPATNAME(route_rcvd_wrapper),
1532 	.pr_recvoob	= COMPATNAME(route_recvoob_wrapper),
1533 	.pr_send	= COMPATNAME(route_send_wrapper),
1534 	.pr_sendoob	= COMPATNAME(route_sendoob_wrapper),
1535 	.pr_purgeif	= COMPATNAME(route_purgeif_wrapper),
1536 };
1537 
1538 static const struct protosw COMPATNAME(route_protosw)[] = {
1539 	{
1540 		.pr_type = SOCK_RAW,
1541 		.pr_domain = &COMPATNAME(routedomain),
1542 		.pr_flags = PR_ATOMIC|PR_ADDR,
1543 		.pr_input = raw_input,
1544 		.pr_output = COMPATNAME(route_output),
1545 		.pr_ctlinput = raw_ctlinput,
1546 		.pr_usrreqs = &route_usrreqs,
1547 		.pr_init = raw_init,
1548 	},
1549 };
1550 
1551 struct domain COMPATNAME(routedomain) = {
1552 	.dom_family = PF_XROUTE,
1553 	.dom_name = DOMAINNAME,
1554 	.dom_init = COMPATNAME(route_init),
1555 	.dom_protosw = COMPATNAME(route_protosw),
1556 	.dom_protoswNPROTOSW =
1557 	    &COMPATNAME(route_protosw)[__arraycount(COMPATNAME(route_protosw))],
1558 };
1559 
1560 static void
1561 sysctl_net_route_setup(struct sysctllog **clog)
1562 {
1563 	const struct sysctlnode *rnode = NULL;
1564 
1565 	sysctl_createv(clog, 0, NULL, &rnode,
1566 		       CTLFLAG_PERMANENT,
1567 		       CTLTYPE_NODE, DOMAINNAME,
1568 		       SYSCTL_DESCR("PF_ROUTE information"),
1569 		       NULL, 0, NULL, 0,
1570 		       CTL_NET, PF_XROUTE, CTL_EOL);
1571 
1572 	sysctl_createv(clog, 0, NULL, NULL,
1573 		       CTLFLAG_PERMANENT,
1574 		       CTLTYPE_NODE, "rtable",
1575 		       SYSCTL_DESCR("Routing table information"),
1576 		       sysctl_rtable, 0, NULL, 0,
1577 		       CTL_NET, PF_XROUTE, 0 /* any protocol */, CTL_EOL);
1578 
1579 	sysctl_createv(clog, 0, &rnode, NULL,
1580 		       CTLFLAG_PERMANENT,
1581 		       CTLTYPE_STRUCT, "stats",
1582 		       SYSCTL_DESCR("Routing statistics"),
1583 		       NULL, 0, &rtstat, sizeof(rtstat),
1584 		       CTL_CREATE, CTL_EOL);
1585 }
1586