xref: /netbsd-src/sys/net/rtsock.c (revision 8b0f9554ff8762542c4defc4f70e1eb76fb508fa)
1 /*	$NetBSD: rtsock.c,v 1.96 2007/12/05 22:56:02 dyoung 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 1.96 2007/12/05 22:56:02 dyoung Exp $");
65 
66 #include "opt_inet.h"
67 
68 #include <sys/param.h>
69 #include <sys/systm.h>
70 #include <sys/proc.h>
71 #include <sys/mbuf.h>
72 #include <sys/socket.h>
73 #include <sys/socketvar.h>
74 #include <sys/domain.h>
75 #include <sys/protosw.h>
76 #include <sys/sysctl.h>
77 #include <sys/kauth.h>
78 #ifdef RTSOCK_DEBUG
79 #include <netinet/in.h>
80 #endif /* RTSOCK_DEBUG */
81 
82 #include <net/if.h>
83 #include <net/route.h>
84 #include <net/raw_cb.h>
85 
86 #include <machine/stdarg.h>
87 
88 DOMAIN_DEFINE(routedomain);	/* forward declare and add to link set */
89 
90 struct	sockaddr route_dst = { .sa_len = 2, .sa_family = PF_ROUTE, };
91 struct	sockaddr route_src = { .sa_len = 2, .sa_family = PF_ROUTE, };
92 struct	sockproto route_proto = { .sp_family = PF_ROUTE, };
93 
94 struct walkarg {
95 	int	w_op;
96 	int	w_arg;
97 	int	w_given;
98 	int	w_needed;
99 	void *	w_where;
100 	int	w_tmemsize;
101 	int	w_tmemneeded;
102 	void *	w_tmem;
103 };
104 
105 static struct mbuf *rt_msg1(int, struct rt_addrinfo *, void *, int);
106 static int rt_msg2(int, struct rt_addrinfo *, void *, struct walkarg *, int *);
107 static int rt_xaddrs(u_char, const char *, const char *, struct rt_addrinfo *);
108 static struct mbuf *rt_makeifannouncemsg(struct ifnet *, int, int,
109     struct rt_addrinfo *);
110 static int sysctl_dumpentry(struct rtentry *, void *);
111 static int sysctl_iflist(int, struct walkarg *, int);
112 static int sysctl_rtable(SYSCTLFN_PROTO);
113 static inline void rt_adjustcount(int, int);
114 
115 /* Sleazy use of local variables throughout file, warning!!!! */
116 #define dst	info.rti_info[RTAX_DST]
117 #define gate	info.rti_info[RTAX_GATEWAY]
118 #define netmask	info.rti_info[RTAX_NETMASK]
119 #define ifpaddr	info.rti_info[RTAX_IFP]
120 #define ifaaddr	info.rti_info[RTAX_IFA]
121 #define brdaddr	info.rti_info[RTAX_BRD]
122 
123 static inline void
124 rt_adjustcount(int af, int cnt)
125 {
126 	route_cb.any_count += cnt;
127 	switch (af) {
128 	case AF_INET:
129 		route_cb.ip_count += cnt;
130 		return;
131 #ifdef INET6
132 	case AF_INET6:
133 		route_cb.ip6_count += cnt;
134 		return;
135 #endif
136 	case AF_IPX:
137 		route_cb.ipx_count += cnt;
138 		return;
139 	case AF_NS:
140 		route_cb.ns_count += cnt;
141 		return;
142 	case AF_ISO:
143 		route_cb.iso_count += cnt;
144 		return;
145 	}
146 }
147 
148 /*ARGSUSED*/
149 int
150 route_usrreq(struct socket *so, int req, struct mbuf *m, struct mbuf *nam,
151 	struct mbuf *control, struct lwp *l)
152 {
153 	int error = 0;
154 	struct rawcb *rp = sotorawcb(so);
155 	int s;
156 
157 	if (req == PRU_ATTACH) {
158 		MALLOC(rp, struct rawcb *, sizeof(*rp), M_PCB, M_WAITOK);
159 		if ((so->so_pcb = rp) != NULL)
160 			memset(so->so_pcb, 0, sizeof(*rp));
161 
162 	}
163 	if (req == PRU_DETACH && rp)
164 		rt_adjustcount(rp->rcb_proto.sp_protocol, -1);
165 	s = splsoftnet();
166 
167 	/*
168 	 * Don't call raw_usrreq() in the attach case, because
169 	 * we want to allow non-privileged processes to listen on
170 	 * and send "safe" commands to the routing socket.
171 	 */
172 	if (req == PRU_ATTACH) {
173 		if (l == NULL)
174 			error = EACCES;
175 		else
176 			error = raw_attach(so, (int)(long)nam);
177 	} else
178 		error = raw_usrreq(so, req, m, nam, control, l);
179 
180 	rp = sotorawcb(so);
181 	if (req == PRU_ATTACH && rp) {
182 		if (error) {
183 			free((void *)rp, M_PCB);
184 			splx(s);
185 			return error;
186 		}
187 		rt_adjustcount(rp->rcb_proto.sp_protocol, 1);
188 		rp->rcb_laddr = &route_src;
189 		rp->rcb_faddr = &route_dst;
190 		soisconnected(so);
191 		so->so_options |= SO_USELOOPBACK;
192 	}
193 	splx(s);
194 	return error;
195 }
196 
197 static const struct sockaddr *
198 intern_netmask(const struct sockaddr *mask)
199 {
200 	struct radix_node *rn;
201 	extern struct radix_node_head *mask_rnhead;
202 
203 	if (mask != NULL &&
204 	    (rn = rn_search(mask, mask_rnhead->rnh_treetop)))
205 		mask = (const struct sockaddr *)rn->rn_key;
206 
207 	return mask;
208 }
209 
210 /*ARGSUSED*/
211 int
212 route_output(struct mbuf *m, ...)
213 {
214 	struct rt_msghdr *rtm = NULL;
215 	struct rtentry *rt = NULL;
216 	struct rtentry *saved_nrt = NULL;
217 	struct rt_addrinfo info;
218 	int len, error = 0;
219 	struct ifnet *ifp = NULL;
220 	struct ifaddr *ifa = NULL;
221 	struct socket *so;
222 	va_list ap;
223 	sa_family_t family;
224 
225 	va_start(ap, m);
226 	so = va_arg(ap, struct socket *);
227 	va_end(ap);
228 
229 #define senderr(e) do { error = e; goto flush;} while (/*CONSTCOND*/ 0)
230 	if (m == NULL || ((m->m_len < sizeof(int32_t)) &&
231 	   (m = m_pullup(m, sizeof(int32_t))) == NULL))
232 		return ENOBUFS;
233 	if ((m->m_flags & M_PKTHDR) == 0)
234 		panic("route_output");
235 	len = m->m_pkthdr.len;
236 	if (len < sizeof(*rtm) ||
237 	    len != mtod(m, struct rt_msghdr *)->rtm_msglen) {
238 		dst = NULL;
239 		senderr(EINVAL);
240 	}
241 	R_Malloc(rtm, struct rt_msghdr *, len);
242 	if (rtm == NULL) {
243 		dst = NULL;
244 		senderr(ENOBUFS);
245 	}
246 	m_copydata(m, 0, len, (void *)rtm);
247 	if (rtm->rtm_version != RTM_VERSION) {
248 		dst = NULL;
249 		senderr(EPROTONOSUPPORT);
250 	}
251 	rtm->rtm_pid = curproc->p_pid;
252 	memset(&info, 0, sizeof(info));
253 	info.rti_addrs = rtm->rtm_addrs;
254 	if (rt_xaddrs(rtm->rtm_type, (void *)(rtm + 1), len + (char *)rtm, &info))
255 		senderr(EINVAL);
256 	info.rti_flags = rtm->rtm_flags;
257 #ifdef RTSOCK_DEBUG
258 	if (dst->sa_family == AF_INET) {
259 		printf("%s: extracted dst %s\n", __func__,
260 		    inet_ntoa(((const struct sockaddr_in *)dst)->sin_addr));
261 	}
262 #endif /* RTSOCK_DEBUG */
263 	if (dst == NULL || (dst->sa_family >= AF_MAX))
264 		senderr(EINVAL);
265 	if (gate != NULL && (gate->sa_family >= AF_MAX))
266 		senderr(EINVAL);
267 
268 	/*
269 	 * Verify that the caller has the appropriate privilege; RTM_GET
270 	 * is the only operation the non-superuser is allowed.
271 	 */
272 	if (kauth_authorize_network(curlwp->l_cred, KAUTH_NETWORK_ROUTE,
273 	    0, rtm, NULL, NULL) != 0)
274 		senderr(EACCES);
275 
276 	switch (rtm->rtm_type) {
277 
278 	case RTM_ADD:
279 		if (gate == NULL)
280 			senderr(EINVAL);
281 		error = rtrequest1(rtm->rtm_type, &info, &saved_nrt);
282 		if (error == 0 && saved_nrt) {
283 			rt_setmetrics(rtm->rtm_inits,
284 			    &rtm->rtm_rmx, &saved_nrt->rt_rmx);
285 			saved_nrt->rt_refcnt--;
286 		}
287 		break;
288 
289 	case RTM_DELETE:
290 		error = rtrequest1(rtm->rtm_type, &info, &saved_nrt);
291 		if (error == 0) {
292 			(rt = saved_nrt)->rt_refcnt++;
293 			goto report;
294 		}
295 		break;
296 
297 	case RTM_GET:
298 	case RTM_CHANGE:
299 	case RTM_LOCK:
300                 /* XXX This will mask dst with netmask before
301                  * searching.  It did not used to do that.  --dyoung
302 		 */
303 		error = rtrequest(RTM_GET, dst, gate, netmask, 0, &rt);
304 		if (error != 0)
305 			senderr(error);
306 		if (rtm->rtm_type != RTM_GET) {/* XXX: too grotty */
307 			struct radix_node *rn;
308 
309 			if (memcmp(dst, rt_getkey(rt), dst->sa_len) != 0)
310 				senderr(ESRCH);
311 			netmask = intern_netmask(netmask);
312 			for (rn = rt->rt_nodes; rn; rn = rn->rn_dupedkey)
313 				if (netmask == (const struct sockaddr *)rn->rn_mask)
314 					break;
315 			if (rn == NULL)
316 				senderr(ETOOMANYREFS);
317 			rt = (struct rtentry *)rn;
318 		}
319 
320 		switch (rtm->rtm_type) {
321 		case RTM_GET:
322 		report:
323 			dst = rt_getkey(rt);
324 			gate = rt->rt_gateway;
325 			netmask = rt_mask(rt);
326 			if ((rtm->rtm_addrs & (RTA_IFP | RTA_IFA)) == 0)
327 				;
328 			else if ((ifp = rt->rt_ifp) != NULL) {
329 				const struct ifaddr *rtifa;
330 				ifpaddr = IFADDR_FIRST(ifp)->ifa_addr;
331                                 /* rtifa used to be simply rt->rt_ifa.
332                                  * If rt->rt_ifa != NULL, then
333                                  * rt_get_ifa() != NULL.  So this
334                                  * ought to still be safe. --dyoung
335 				 */
336 				rtifa = rt_get_ifa(rt);
337 				ifaaddr = rtifa->ifa_addr;
338 #ifdef RTSOCK_DEBUG
339 				if (ifaaddr->sa_family == AF_INET) {
340 					printf("%s: copying out RTAX_IFA %s ",
341 					    __func__,
342 					    inet_ntoa(((const struct sockaddr_in *)ifaaddr)->sin_addr));
343 					printf("for dst %s ifa_getifa %p ifa_seqno %p\n",
344 					    inet_ntoa(((const struct sockaddr_in *)dst)->sin_addr),
345 					    (void *)rtifa->ifa_getifa, rtifa->ifa_seqno);
346 				}
347 #endif /* RTSOCK_DEBUG */
348 				if (ifp->if_flags & IFF_POINTOPOINT)
349 					brdaddr = rtifa->ifa_dstaddr;
350 				else
351 					brdaddr = NULL;
352 				rtm->rtm_index = ifp->if_index;
353 			} else {
354 				ifpaddr = NULL;
355 				ifaaddr = NULL;
356 			}
357 			(void)rt_msg2(rtm->rtm_type, &info, NULL, NULL, &len);
358 			if (len > rtm->rtm_msglen) {
359 				struct rt_msghdr *new_rtm;
360 				R_Malloc(new_rtm, struct rt_msghdr *, len);
361 				if (new_rtm == NULL)
362 					senderr(ENOBUFS);
363 				Bcopy(rtm, new_rtm, rtm->rtm_msglen);
364 				Free(rtm); rtm = new_rtm;
365 			}
366 			(void)rt_msg2(rtm->rtm_type, &info, (void *)rtm,
367 			    NULL, 0);
368 			rtm->rtm_flags = rt->rt_flags;
369 			rtm->rtm_rmx = rt->rt_rmx;
370 			rtm->rtm_addrs = info.rti_addrs;
371 			break;
372 
373 		case RTM_CHANGE:
374 			/*
375 			 * new gateway could require new ifaddr, ifp;
376 			 * flags may also be different; ifp may be specified
377 			 * by ll sockaddr when protocol address is ambiguous
378 			 */
379 			if ((error = rt_getifa(&info)) != 0)
380 				senderr(error);
381 			if (gate && rt_setgate(rt, gate))
382 				senderr(EDQUOT);
383 			/* new gateway could require new ifaddr, ifp;
384 			   flags may also be different; ifp may be specified
385 			   by ll sockaddr when protocol address is ambiguous */
386 			if (ifpaddr && (ifa = ifa_ifwithnet(ifpaddr)) &&
387 			    (ifp = ifa->ifa_ifp) && (ifaaddr || gate))
388 				ifa = ifaof_ifpforaddr(ifaaddr ? ifaaddr : gate,
389 				    ifp);
390 			else if ((ifaaddr && (ifa = ifa_ifwithaddr(ifaaddr))) ||
391 			    (gate && (ifa = ifa_ifwithroute(rt->rt_flags,
392 			    rt_getkey(rt), gate))))
393 				ifp = ifa->ifa_ifp;
394 			if (ifa) {
395 				struct ifaddr *oifa = rt->rt_ifa;
396 				if (oifa != ifa) {
397 					if (oifa && oifa->ifa_rtrequest) {
398 						oifa->ifa_rtrequest(RTM_DELETE,
399 						    rt, &info);
400 					}
401 					rt_replace_ifa(rt, ifa);
402 					rt->rt_ifp = ifp;
403 				}
404 			}
405 			rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx,
406 			    &rt->rt_rmx);
407 			if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest)
408 				rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, &info);
409 			/*
410 			 * Fall into
411 			 */
412 		case RTM_LOCK:
413 			rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits);
414 			rt->rt_rmx.rmx_locks |=
415 			    (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks);
416 			break;
417 		}
418 		break;
419 
420 	default:
421 		senderr(EOPNOTSUPP);
422 	}
423 
424 flush:
425 	if (rtm) {
426 		if (error)
427 			rtm->rtm_errno = error;
428 		else
429 			rtm->rtm_flags |= RTF_DONE;
430 	}
431 	family = dst ? dst->sa_family : 0;
432 	if (rt)
433 		rtfree(rt);
434     {
435 	struct rawcb *rp = NULL;
436 	/*
437 	 * Check to see if we don't want our own messages.
438 	 */
439 	if ((so->so_options & SO_USELOOPBACK) == 0) {
440 		if (route_cb.any_count <= 1) {
441 			if (rtm)
442 				Free(rtm);
443 			m_freem(m);
444 			return error;
445 		}
446 		/* There is another listener, so construct message */
447 		rp = sotorawcb(so);
448 	}
449 	if (rtm) {
450 		m_copyback(m, 0, rtm->rtm_msglen, (void *)rtm);
451 		if (m->m_pkthdr.len < rtm->rtm_msglen) {
452 			m_freem(m);
453 			m = NULL;
454 		} else if (m->m_pkthdr.len > rtm->rtm_msglen)
455 			m_adj(m, rtm->rtm_msglen - m->m_pkthdr.len);
456 		Free(rtm);
457 	}
458 	if (rp)
459 		rp->rcb_proto.sp_family = 0; /* Avoid us */
460 	if (family)
461 		route_proto.sp_protocol = family;
462 	if (m)
463 		raw_input(m, &route_proto, &route_src, &route_dst);
464 	if (rp)
465 		rp->rcb_proto.sp_family = PF_ROUTE;
466     }
467 	return error;
468 }
469 
470 void
471 rt_setmetrics(u_long which, const struct rt_metrics *in, struct rt_metrics *out)
472 {
473 #define metric(f, e) if (which & (f)) out->e = in->e;
474 	metric(RTV_RPIPE, rmx_recvpipe);
475 	metric(RTV_SPIPE, rmx_sendpipe);
476 	metric(RTV_SSTHRESH, rmx_ssthresh);
477 	metric(RTV_RTT, rmx_rtt);
478 	metric(RTV_RTTVAR, rmx_rttvar);
479 	metric(RTV_HOPCOUNT, rmx_hopcount);
480 	metric(RTV_MTU, rmx_mtu);
481 	metric(RTV_EXPIRE, rmx_expire);
482 #undef metric
483 }
484 
485 #define ROUNDUP(a) \
486 	((a) > 0 ? (1 + (((a) - 1) | (sizeof(long) - 1))) : sizeof(long))
487 #define ADVANCE(x, n) (x += ROUNDUP((n)->sa_len))
488 
489 static int
490 rt_xaddrs(u_char rtmtype, const char *cp, const char *cplim, struct rt_addrinfo *rtinfo)
491 {
492 	const struct sockaddr *sa = NULL;	/* Quell compiler warning */
493 	int i;
494 
495 	for (i = 0; (i < RTAX_MAX) && (cp < cplim); i++) {
496 		if ((rtinfo->rti_addrs & (1 << i)) == 0)
497 			continue;
498 		rtinfo->rti_info[i] = sa = (const struct sockaddr *)cp;
499 		ADVANCE(cp, sa);
500 	}
501 
502 	/* Check for extra addresses specified, except RTM_GET asking for interface info.  */
503 	if (rtmtype == RTM_GET) {
504 		if (((rtinfo->rti_addrs & (~((1 << RTAX_IFP) | (1 << RTAX_IFA)))) & (~0 << i)) != 0)
505 			return 1;
506 	} else {
507 		if ((rtinfo->rti_addrs & (~0 << i)) != 0)
508 			return 1;
509 	}
510 	/* Check for bad data length.  */
511 	if (cp != cplim) {
512 		if (i == RTAX_NETMASK + 1 && sa &&
513 		    cp - ROUNDUP(sa->sa_len) + sa->sa_len == cplim)
514 			/*
515 			 * The last sockaddr was netmask.
516 			 * We accept this for now for the sake of old
517 			 * binaries or third party softwares.
518 			 */
519 			;
520 		else
521 			return 1;
522 	}
523 	return 0;
524 }
525 
526 static struct mbuf *
527 rt_msg1(int type, struct rt_addrinfo *rtinfo, void *data, int datalen)
528 {
529 	struct rt_msghdr *rtm;
530 	struct mbuf *m;
531 	int i;
532 	const struct sockaddr *sa;
533 	int len, dlen;
534 
535 	m = m_gethdr(M_DONTWAIT, MT_DATA);
536 	if (m == NULL)
537 		return m;
538 	MCLAIM(m, &routedomain.dom_mowner);
539 	switch (type) {
540 
541 	case RTM_DELADDR:
542 	case RTM_NEWADDR:
543 		len = sizeof(struct ifa_msghdr);
544 		break;
545 
546 #ifdef COMPAT_14
547 	case RTM_OIFINFO:
548 		len = sizeof(struct if_msghdr14);
549 		break;
550 #endif
551 
552 	case RTM_IFINFO:
553 		len = sizeof(struct if_msghdr);
554 		break;
555 
556 	case RTM_IFANNOUNCE:
557 	case RTM_IEEE80211:
558 		len = sizeof(struct if_announcemsghdr);
559 		break;
560 
561 	default:
562 		len = sizeof(struct rt_msghdr);
563 	}
564 	if (len > MHLEN + MLEN)
565 		panic("rt_msg1: message too long");
566 	else if (len > MHLEN) {
567 		m->m_next = m_get(M_DONTWAIT, MT_DATA);
568 		if (m->m_next == NULL) {
569 			m_freem(m);
570 			return NULL;
571 		}
572 		MCLAIM(m->m_next, m->m_owner);
573 		m->m_pkthdr.len = len;
574 		m->m_len = MHLEN;
575 		m->m_next->m_len = len - MHLEN;
576 	} else {
577 		m->m_pkthdr.len = m->m_len = len;
578 	}
579 	m->m_pkthdr.rcvif = NULL;
580 	m_copyback(m, 0, datalen, data);
581 	rtm = mtod(m, struct rt_msghdr *);
582 	for (i = 0; i < RTAX_MAX; i++) {
583 		if ((sa = rtinfo->rti_info[i]) == NULL)
584 			continue;
585 		rtinfo->rti_addrs |= (1 << i);
586 		dlen = ROUNDUP(sa->sa_len);
587 		m_copyback(m, len, dlen, sa);
588 		len += dlen;
589 	}
590 	if (m->m_pkthdr.len != len) {
591 		m_freem(m);
592 		return NULL;
593 	}
594 	rtm->rtm_msglen = len;
595 	rtm->rtm_version = RTM_VERSION;
596 	rtm->rtm_type = type;
597 	return m;
598 }
599 
600 /*
601  * rt_msg2
602  *
603  *	 fills 'cp' or 'w'.w_tmem with the routing socket message and
604  *		returns the length of the message in 'lenp'.
605  *
606  * if walkarg is 0, cp is expected to be 0 or a buffer large enough to hold
607  *	the message
608  * otherwise walkarg's w_needed is updated and if the user buffer is
609  *	specified and w_needed indicates space exists the information is copied
610  *	into the temp space (w_tmem). w_tmem is [re]allocated if necessary,
611  *	if the allocation fails ENOBUFS is returned.
612  */
613 static int
614 rt_msg2(int type, struct rt_addrinfo *rtinfo, void *cpv, struct walkarg *w,
615 	int *lenp)
616 {
617 	int i;
618 	int len, dlen, second_time = 0;
619 	char *cp0, *cp = cpv;
620 
621 	rtinfo->rti_addrs = 0;
622 again:
623 	switch (type) {
624 
625 	case RTM_DELADDR:
626 	case RTM_NEWADDR:
627 		len = sizeof(struct ifa_msghdr);
628 		break;
629 #ifdef COMPAT_14
630 	case RTM_OIFINFO:
631 		len = sizeof(struct if_msghdr14);
632 		break;
633 #endif
634 
635 	case RTM_IFINFO:
636 		len = sizeof(struct if_msghdr);
637 		break;
638 
639 	default:
640 		len = sizeof(struct rt_msghdr);
641 	}
642 	if ((cp0 = cp) != NULL)
643 		cp += len;
644 	for (i = 0; i < RTAX_MAX; i++) {
645 		const struct sockaddr *sa;
646 
647 		if ((sa = rtinfo->rti_info[i]) == NULL)
648 			continue;
649 		rtinfo->rti_addrs |= (1 << i);
650 		dlen = ROUNDUP(sa->sa_len);
651 		if (cp) {
652 			bcopy(sa, cp, (unsigned)dlen);
653 			cp += dlen;
654 		}
655 		len += dlen;
656 	}
657 	if (cp == NULL && w != NULL && !second_time) {
658 		struct walkarg *rw = w;
659 
660 		rw->w_needed += len;
661 		if (rw->w_needed <= 0 && rw->w_where) {
662 			if (rw->w_tmemsize < len) {
663 				if (rw->w_tmem)
664 					free(rw->w_tmem, M_RTABLE);
665 				rw->w_tmem = (void *) malloc(len, M_RTABLE,
666 				    M_NOWAIT);
667 				if (rw->w_tmem)
668 					rw->w_tmemsize = len;
669 			}
670 			if (rw->w_tmem) {
671 				cp = rw->w_tmem;
672 				second_time = 1;
673 				goto again;
674 			} else {
675 				rw->w_tmemneeded = len;
676 				return ENOBUFS;
677 			}
678 		}
679 	}
680 	if (cp) {
681 		struct rt_msghdr *rtm = (struct rt_msghdr *)cp0;
682 
683 		rtm->rtm_version = RTM_VERSION;
684 		rtm->rtm_type = type;
685 		rtm->rtm_msglen = len;
686 	}
687 	if (lenp)
688 		*lenp = len;
689 	return 0;
690 }
691 
692 /*
693  * This routine is called to generate a message from the routing
694  * socket indicating that a redirect has occurred, a routing lookup
695  * has failed, or that a protocol has detected timeouts to a particular
696  * destination.
697  */
698 void
699 rt_missmsg(int type, struct rt_addrinfo *rtinfo, int flags, int error)
700 {
701 	struct rt_msghdr rtm;
702 	struct mbuf *m;
703 	const struct sockaddr *sa = rtinfo->rti_info[RTAX_DST];
704 
705 	if (route_cb.any_count == 0)
706 		return;
707 	memset(&rtm, 0, sizeof(rtm));
708 	rtm.rtm_flags = RTF_DONE | flags;
709 	rtm.rtm_errno = error;
710 	m = rt_msg1(type, rtinfo, (void *)&rtm, sizeof(rtm));
711 	if (m == NULL)
712 		return;
713 	mtod(m, struct rt_msghdr *)->rtm_addrs = rtinfo->rti_addrs;
714 	route_proto.sp_protocol = sa ? sa->sa_family : 0;
715 	raw_input(m, &route_proto, &route_src, &route_dst);
716 }
717 
718 /*
719  * This routine is called to generate a message from the routing
720  * socket indicating that the status of a network interface has changed.
721  */
722 void
723 rt_ifmsg(struct ifnet *ifp)
724 {
725 	struct if_msghdr ifm;
726 #ifdef COMPAT_14
727 	struct if_msghdr14 oifm;
728 #endif
729 	struct mbuf *m;
730 	struct rt_addrinfo info;
731 
732 	if (route_cb.any_count == 0)
733 		return;
734 	memset(&info, 0, sizeof(info));
735 	memset(&ifm, 0, sizeof(ifm));
736 	ifm.ifm_index = ifp->if_index;
737 	ifm.ifm_flags = ifp->if_flags;
738 	ifm.ifm_data = ifp->if_data;
739 	ifm.ifm_addrs = 0;
740 	m = rt_msg1(RTM_IFINFO, &info, (void *)&ifm, sizeof(ifm));
741 	if (m == NULL)
742 		return;
743 	route_proto.sp_protocol = 0;
744 	raw_input(m, &route_proto, &route_src, &route_dst);
745 #ifdef COMPAT_14
746 	memset(&info, 0, sizeof(info));
747 	memset(&oifm, 0, sizeof(oifm));
748 	oifm.ifm_index = ifp->if_index;
749 	oifm.ifm_flags = ifp->if_flags;
750 	oifm.ifm_data.ifi_type = ifp->if_data.ifi_type;
751 	oifm.ifm_data.ifi_addrlen = ifp->if_data.ifi_addrlen;
752 	oifm.ifm_data.ifi_hdrlen = ifp->if_data.ifi_hdrlen;
753 	oifm.ifm_data.ifi_mtu = ifp->if_data.ifi_mtu;
754 	oifm.ifm_data.ifi_metric = ifp->if_data.ifi_metric;
755 	oifm.ifm_data.ifi_baudrate = ifp->if_data.ifi_baudrate;
756 	oifm.ifm_data.ifi_ipackets = ifp->if_data.ifi_ipackets;
757 	oifm.ifm_data.ifi_ierrors = ifp->if_data.ifi_ierrors;
758 	oifm.ifm_data.ifi_opackets = ifp->if_data.ifi_opackets;
759 	oifm.ifm_data.ifi_oerrors = ifp->if_data.ifi_oerrors;
760 	oifm.ifm_data.ifi_collisions = ifp->if_data.ifi_collisions;
761 	oifm.ifm_data.ifi_ibytes = ifp->if_data.ifi_ibytes;
762 	oifm.ifm_data.ifi_obytes = ifp->if_data.ifi_obytes;
763 	oifm.ifm_data.ifi_imcasts = ifp->if_data.ifi_imcasts;
764 	oifm.ifm_data.ifi_omcasts = ifp->if_data.ifi_omcasts;
765 	oifm.ifm_data.ifi_iqdrops = ifp->if_data.ifi_iqdrops;
766 	oifm.ifm_data.ifi_noproto = ifp->if_data.ifi_noproto;
767 	oifm.ifm_data.ifi_lastchange = ifp->if_data.ifi_lastchange;
768 	oifm.ifm_addrs = 0;
769 	m = rt_msg1(RTM_OIFINFO, &info, (void *)&oifm, sizeof(oifm));
770 	if (m == NULL)
771 		return;
772 	route_proto.sp_protocol = 0;
773 	raw_input(m, &route_proto, &route_src, &route_dst);
774 #endif
775 }
776 
777 /*
778  * This is called to generate messages from the routing socket
779  * indicating a network interface has had addresses associated with it.
780  * if we ever reverse the logic and replace messages TO the routing
781  * socket indicate a request to configure interfaces, then it will
782  * be unnecessary as the routing socket will automatically generate
783  * copies of it.
784  */
785 void
786 rt_newaddrmsg(int cmd, struct ifaddr *ifa, int error, struct rtentry *rt)
787 {
788 	struct rt_addrinfo info;
789 	const struct sockaddr *sa = NULL;
790 	int pass;
791 	struct mbuf *m = NULL;
792 	struct ifnet *ifp = ifa->ifa_ifp;
793 
794 	if (route_cb.any_count == 0)
795 		return;
796 	for (pass = 1; pass < 3; pass++) {
797 		memset(&info, 0, sizeof(info));
798 		if ((cmd == RTM_ADD && pass == 1) ||
799 		    (cmd == RTM_DELETE && pass == 2)) {
800 			struct ifa_msghdr ifam;
801 			int ncmd = cmd == RTM_ADD ? RTM_NEWADDR : RTM_DELADDR;
802 
803 			ifaaddr = sa = ifa->ifa_addr;
804 			ifpaddr = IFADDR_FIRST(ifp)->ifa_addr;
805 			netmask = ifa->ifa_netmask;
806 			brdaddr = ifa->ifa_dstaddr;
807 			memset(&ifam, 0, sizeof(ifam));
808 			ifam.ifam_index = ifp->if_index;
809 			ifam.ifam_metric = ifa->ifa_metric;
810 			ifam.ifam_flags = ifa->ifa_flags;
811 			m = rt_msg1(ncmd, &info, (void *)&ifam, sizeof(ifam));
812 			if (m == NULL)
813 				continue;
814 			mtod(m, struct ifa_msghdr *)->ifam_addrs =
815 			    info.rti_addrs;
816 		}
817 		if ((cmd == RTM_ADD && pass == 2) ||
818 		    (cmd == RTM_DELETE && pass == 1)) {
819 			struct rt_msghdr rtm;
820 
821 			if (rt == NULL)
822 				continue;
823 			netmask = rt_mask(rt);
824 			dst = sa = rt_getkey(rt);
825 			gate = rt->rt_gateway;
826 			memset(&rtm, 0, sizeof(rtm));
827 			rtm.rtm_index = ifp->if_index;
828 			rtm.rtm_flags |= rt->rt_flags;
829 			rtm.rtm_errno = error;
830 			m = rt_msg1(cmd, &info, (void *)&rtm, sizeof(rtm));
831 			if (m == NULL)
832 				continue;
833 			mtod(m, struct rt_msghdr *)->rtm_addrs = info.rti_addrs;
834 		}
835 		route_proto.sp_protocol = sa ? sa->sa_family : 0;
836 		raw_input(m, &route_proto, &route_src, &route_dst);
837 	}
838 }
839 
840 static struct mbuf *
841 rt_makeifannouncemsg(struct ifnet *ifp, int type, int what,
842     struct rt_addrinfo *info)
843 {
844 	struct if_announcemsghdr ifan;
845 
846 	memset(info, 0, sizeof(*info));
847 	memset(&ifan, 0, sizeof(ifan));
848 	ifan.ifan_index = ifp->if_index;
849 	strlcpy(ifan.ifan_name, ifp->if_xname, sizeof(ifan.ifan_name));
850 	ifan.ifan_what = what;
851 	return rt_msg1(type, info, (void *)&ifan, sizeof(ifan));
852 }
853 
854 /*
855  * This is called to generate routing socket messages indicating
856  * network interface arrival and departure.
857  */
858 void
859 rt_ifannouncemsg(struct ifnet *ifp, int what)
860 {
861 	struct mbuf *m;
862 	struct rt_addrinfo info;
863 
864 	if (route_cb.any_count == 0)
865 		return;
866 	m = rt_makeifannouncemsg(ifp, RTM_IFANNOUNCE, what, &info);
867 	if (m == NULL)
868 		return;
869 	route_proto.sp_protocol = 0;
870 	raw_input(m, &route_proto, &route_src, &route_dst);
871 }
872 
873 /*
874  * This is called to generate routing socket messages indicating
875  * IEEE80211 wireless events.
876  * XXX we piggyback on the RTM_IFANNOUNCE msg format in a clumsy way.
877  */
878 void
879 rt_ieee80211msg(struct ifnet *ifp, int what, void *data, size_t data_len)
880 {
881 	struct mbuf *m;
882 	struct rt_addrinfo info;
883 
884 	if (route_cb.any_count == 0)
885 		return;
886 	m = rt_makeifannouncemsg(ifp, RTM_IEEE80211, what, &info);
887 	if (m == NULL)
888 		return;
889 	/*
890 	 * Append the ieee80211 data.  Try to stick it in the
891 	 * mbuf containing the ifannounce msg; otherwise allocate
892 	 * a new mbuf and append.
893 	 *
894 	 * NB: we assume m is a single mbuf.
895 	 */
896 	if (data_len > M_TRAILINGSPACE(m)) {
897 		struct mbuf *n = m_get(M_NOWAIT, MT_DATA);
898 		if (n == NULL) {
899 			m_freem(m);
900 			return;
901 		}
902 		(void)memcpy(mtod(n, void *), data, data_len);
903 		n->m_len = data_len;
904 		m->m_next = n;
905 	} else if (data_len > 0) {
906 		(void)memcpy(mtod(m, u_int8_t *) + m->m_len, data, data_len);
907 		m->m_len += data_len;
908 	}
909 	if (m->m_flags & M_PKTHDR)
910 		m->m_pkthdr.len += data_len;
911 	mtod(m, struct if_announcemsghdr *)->ifan_msglen += data_len;
912 	route_proto.sp_protocol = 0;
913 	raw_input(m, &route_proto, &route_src, &route_dst);
914 }
915 
916 /*
917  * This is used in dumping the kernel table via sysctl().
918  */
919 static int
920 sysctl_dumpentry(struct rtentry *rt, void *v)
921 {
922 	struct walkarg *w = v;
923 	int error = 0, size;
924 	struct rt_addrinfo info;
925 
926 	if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg))
927 		return 0;
928 	memset(&info, 0, sizeof(info));
929 	dst = rt_getkey(rt);
930 	gate = rt->rt_gateway;
931 	netmask = rt_mask(rt);
932 	if (rt->rt_ifp) {
933 		const struct ifaddr *rtifa;
934 		ifpaddr = IFADDR_FIRST(rt->rt_ifp)->ifa_addr;
935 		/* rtifa used to be simply rt->rt_ifa.  If rt->rt_ifa != NULL,
936 		 * then rt_get_ifa() != NULL.  So this ought to still be safe.
937 		 * --dyoung
938 		 */
939 		rtifa = rt_get_ifa(rt);
940 		ifaaddr = rtifa->ifa_addr;
941 		if (rt->rt_ifp->if_flags & IFF_POINTOPOINT)
942 			brdaddr = rtifa->ifa_dstaddr;
943 	}
944 	if ((error = rt_msg2(RTM_GET, &info, 0, w, &size)))
945 		return error;
946 	if (w->w_where && w->w_tmem && w->w_needed <= 0) {
947 		struct rt_msghdr *rtm = (struct rt_msghdr *)w->w_tmem;
948 
949 		rtm->rtm_flags = rt->rt_flags;
950 		rtm->rtm_use = rt->rt_use;
951 		rtm->rtm_rmx = rt->rt_rmx;
952 		KASSERT(rt->rt_ifp != NULL);
953 		rtm->rtm_index = rt->rt_ifp->if_index;
954 		rtm->rtm_errno = rtm->rtm_pid = rtm->rtm_seq = 0;
955 		rtm->rtm_addrs = info.rti_addrs;
956 		if ((error = copyout(rtm, w->w_where, size)) != 0)
957 			w->w_where = NULL;
958 		else
959 			w->w_where = (char *)w->w_where + size;
960 	}
961 	return error;
962 }
963 
964 static int
965 sysctl_iflist(int af, struct walkarg *w, int type)
966 {
967 	struct ifnet *ifp;
968 	struct ifaddr *ifa;
969 	struct	rt_addrinfo info;
970 	int	len, error = 0;
971 
972 	memset(&info, 0, sizeof(info));
973 	IFNET_FOREACH(ifp) {
974 		if (w->w_arg && w->w_arg != ifp->if_index)
975 			continue;
976 		ifa = IFADDR_FIRST(ifp);
977 		if (ifa == NULL)
978 			continue;
979 		ifpaddr = ifa->ifa_addr;
980 		switch (type) {
981 		case NET_RT_IFLIST:
982 			error =
983 			    rt_msg2(RTM_IFINFO, &info, NULL, w, &len);
984 			break;
985 #ifdef COMPAT_14
986 		case NET_RT_OIFLIST:
987 			error =
988 			    rt_msg2(RTM_OIFINFO, &info, NULL, w, &len);
989 			break;
990 #endif
991 		default:
992 			panic("sysctl_iflist(1)");
993 		}
994 		if (error)
995 			return error;
996 		ifpaddr = NULL;
997 		if (w->w_where && w->w_tmem && w->w_needed <= 0) {
998 			switch (type) {
999 			case NET_RT_IFLIST: {
1000 				struct if_msghdr *ifm;
1001 
1002 				ifm = (struct if_msghdr *)w->w_tmem;
1003 				ifm->ifm_index = ifp->if_index;
1004 				ifm->ifm_flags = ifp->if_flags;
1005 				ifm->ifm_data = ifp->if_data;
1006 				ifm->ifm_addrs = info.rti_addrs;
1007 				error = copyout(ifm, w->w_where, len);
1008 				if (error)
1009 					return error;
1010 				w->w_where = (char *)w->w_where + len;
1011 				break;
1012 			}
1013 
1014 #ifdef COMPAT_14
1015 			case NET_RT_OIFLIST: {
1016 				struct if_msghdr14 *ifm;
1017 
1018 				ifm = (struct if_msghdr14 *)w->w_tmem;
1019 				ifm->ifm_index = ifp->if_index;
1020 				ifm->ifm_flags = ifp->if_flags;
1021 				ifm->ifm_data.ifi_type = ifp->if_data.ifi_type;
1022 				ifm->ifm_data.ifi_addrlen =
1023 				    ifp->if_data.ifi_addrlen;
1024 				ifm->ifm_data.ifi_hdrlen =
1025 				    ifp->if_data.ifi_hdrlen;
1026 				ifm->ifm_data.ifi_mtu = ifp->if_data.ifi_mtu;
1027 				ifm->ifm_data.ifi_metric =
1028 				    ifp->if_data.ifi_metric;
1029 				ifm->ifm_data.ifi_baudrate =
1030 				    ifp->if_data.ifi_baudrate;
1031 				ifm->ifm_data.ifi_ipackets =
1032 				    ifp->if_data.ifi_ipackets;
1033 				ifm->ifm_data.ifi_ierrors =
1034 				    ifp->if_data.ifi_ierrors;
1035 				ifm->ifm_data.ifi_opackets =
1036 				    ifp->if_data.ifi_opackets;
1037 				ifm->ifm_data.ifi_oerrors =
1038 				    ifp->if_data.ifi_oerrors;
1039 				ifm->ifm_data.ifi_collisions =
1040 				    ifp->if_data.ifi_collisions;
1041 				ifm->ifm_data.ifi_ibytes =
1042 				    ifp->if_data.ifi_ibytes;
1043 				ifm->ifm_data.ifi_obytes =
1044 				    ifp->if_data.ifi_obytes;
1045 				ifm->ifm_data.ifi_imcasts =
1046 				    ifp->if_data.ifi_imcasts;
1047 				ifm->ifm_data.ifi_omcasts =
1048 				    ifp->if_data.ifi_omcasts;
1049 				ifm->ifm_data.ifi_iqdrops =
1050 				    ifp->if_data.ifi_iqdrops;
1051 				ifm->ifm_data.ifi_noproto =
1052 				    ifp->if_data.ifi_noproto;
1053 				ifm->ifm_data.ifi_lastchange =
1054 				    ifp->if_data.ifi_lastchange;
1055 				ifm->ifm_addrs = info.rti_addrs;
1056 				error = copyout(ifm, w->w_where, len);
1057 				if (error)
1058 					return error;
1059 				w->w_where = (char *)w->w_where + len;
1060 				break;
1061 			}
1062 #endif
1063 			default:
1064 				panic("sysctl_iflist(2)");
1065 			}
1066 		}
1067 		while ((ifa = IFADDR_NEXT(ifa)) != NULL) {
1068 			if (af && af != ifa->ifa_addr->sa_family)
1069 				continue;
1070 			ifaaddr = ifa->ifa_addr;
1071 			netmask = ifa->ifa_netmask;
1072 			brdaddr = ifa->ifa_dstaddr;
1073 			if ((error = rt_msg2(RTM_NEWADDR, &info, 0, w, &len)))
1074 				return error;
1075 			if (w->w_where && w->w_tmem && w->w_needed <= 0) {
1076 				struct ifa_msghdr *ifam;
1077 
1078 				ifam = (struct ifa_msghdr *)w->w_tmem;
1079 				ifam->ifam_index = ifa->ifa_ifp->if_index;
1080 				ifam->ifam_flags = ifa->ifa_flags;
1081 				ifam->ifam_metric = ifa->ifa_metric;
1082 				ifam->ifam_addrs = info.rti_addrs;
1083 				error = copyout(w->w_tmem, w->w_where, len);
1084 				if (error)
1085 					return error;
1086 				w->w_where = (char *)w->w_where + len;
1087 			}
1088 		}
1089 		ifaaddr = netmask = brdaddr = NULL;
1090 	}
1091 	return 0;
1092 }
1093 
1094 static int
1095 sysctl_rtable(SYSCTLFN_ARGS)
1096 {
1097 	void 	*where = oldp;
1098 	size_t	*given = oldlenp;
1099 	const void *new = newp;
1100 	int	i, s, error = EINVAL;
1101 	u_char  af;
1102 	struct	walkarg w;
1103 
1104 	if (namelen == 1 && name[0] == CTL_QUERY)
1105 		return sysctl_query(SYSCTLFN_CALL(rnode));
1106 
1107 	if (new)
1108 		return EPERM;
1109 	if (namelen != 3)
1110 		return EINVAL;
1111 	af = name[0];
1112 	w.w_tmemneeded = 0;
1113 	w.w_tmemsize = 0;
1114 	w.w_tmem = NULL;
1115 again:
1116 	/* we may return here if a later [re]alloc of the t_mem buffer fails */
1117 	if (w.w_tmemneeded) {
1118 		w.w_tmem = (void *) malloc(w.w_tmemneeded, M_RTABLE, M_WAITOK);
1119 		w.w_tmemsize = w.w_tmemneeded;
1120 		w.w_tmemneeded = 0;
1121 	}
1122 	w.w_op = name[1];
1123 	w.w_arg = name[2];
1124 	w.w_given = *given;
1125 	w.w_needed = 0 - w.w_given;
1126 	w.w_where = where;
1127 
1128 	s = splsoftnet();
1129 	switch (w.w_op) {
1130 
1131 	case NET_RT_DUMP:
1132 	case NET_RT_FLAGS:
1133 		for (i = 1; i <= AF_MAX; i++)
1134 			if ((af == 0 || af == i) &&
1135 			    (error = rt_walktree(i, sysctl_dumpentry, &w)))
1136 				break;
1137 		break;
1138 
1139 #ifdef COMPAT_14
1140 	case NET_RT_OIFLIST:
1141 		error = sysctl_iflist(af, &w, w.w_op);
1142 		break;
1143 #endif
1144 
1145 	case NET_RT_IFLIST:
1146 		error = sysctl_iflist(af, &w, w.w_op);
1147 	}
1148 	splx(s);
1149 
1150 	/* check to see if we couldn't allocate memory with NOWAIT */
1151 	if (error == ENOBUFS && w.w_tmem == 0 && w.w_tmemneeded)
1152 		goto again;
1153 
1154 	if (w.w_tmem)
1155 		free(w.w_tmem, M_RTABLE);
1156 	w.w_needed += w.w_given;
1157 	if (where) {
1158 		*given = (char *)w.w_where - (char *)where;
1159 		if (*given < w.w_needed)
1160 			return ENOMEM;
1161 	} else {
1162 		*given = (11 * w.w_needed) / 10;
1163 	}
1164 	return error;
1165 }
1166 
1167 /*
1168  * Definitions of protocols supported in the ROUTE domain.
1169  */
1170 
1171 const struct protosw routesw[] = {
1172 	{
1173 		.pr_type = SOCK_RAW,
1174 		.pr_domain = &routedomain,
1175 		.pr_flags = PR_ATOMIC|PR_ADDR,
1176 		.pr_input = raw_input,
1177 		.pr_output = route_output,
1178 		.pr_ctlinput = raw_ctlinput,
1179 		.pr_usrreq = route_usrreq,
1180 		.pr_init = raw_init,
1181 	},
1182 };
1183 
1184 struct domain routedomain = {
1185 	.dom_family = PF_ROUTE,
1186 	.dom_name = "route",
1187 	.dom_init = route_init,
1188 	.dom_protosw = routesw,
1189 	.dom_protoswNPROTOSW = &routesw[__arraycount(routesw)],
1190 };
1191 
1192 SYSCTL_SETUP(sysctl_net_route_setup, "sysctl net.route subtree setup")
1193 {
1194 	const struct sysctlnode *rnode = NULL;
1195 
1196 	sysctl_createv(clog, 0, NULL, NULL,
1197 		       CTLFLAG_PERMANENT,
1198 		       CTLTYPE_NODE, "net", NULL,
1199 		       NULL, 0, NULL, 0,
1200 		       CTL_NET, CTL_EOL);
1201 
1202 	sysctl_createv(clog, 0, NULL, &rnode,
1203 		       CTLFLAG_PERMANENT,
1204 		       CTLTYPE_NODE, "route",
1205 		       SYSCTL_DESCR("PF_ROUTE information"),
1206 		       NULL, 0, NULL, 0,
1207 		       CTL_NET, PF_ROUTE, CTL_EOL);
1208 	sysctl_createv(clog, 0, NULL, NULL,
1209 		       CTLFLAG_PERMANENT,
1210 		       CTLTYPE_NODE, "rtable",
1211 		       SYSCTL_DESCR("Routing table information"),
1212 		       sysctl_rtable, 0, NULL, 0,
1213 		       CTL_NET, PF_ROUTE, 0 /* any protocol */, CTL_EOL);
1214 	sysctl_createv(clog, 0, &rnode, NULL,
1215 		       CTLFLAG_PERMANENT,
1216 		       CTLTYPE_STRUCT, "stats",
1217 		       SYSCTL_DESCR("Routing statistics"),
1218 		       NULL, 0, &rtstat, sizeof(rtstat),
1219 		       CTL_CREATE, CTL_EOL);
1220 }
1221