xref: /openbsd-src/sys/net/rtsock.c (revision b2ea75c1b17e1a9a339660e7ed45cd24946b230e)
1 /*	$OpenBSD: rtsock.c,v 1.16 2001/07/21 12:22:57 itojun Exp $	*/
2 /*	$NetBSD: rtsock.c,v 1.18 1996/03/29 00:32:10 cgd Exp $	*/
3 
4 /*
5  * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
6  * All rights reserved.
7  *
8  * Redistribution and use in source and binary forms, with or without
9  * modification, are permitted provided that the following conditions
10  * are met:
11  * 1. Redistributions of source code must retain the above copyright
12  *    notice, this list of conditions and the following disclaimer.
13  * 2. Redistributions in binary form must reproduce the above copyright
14  *    notice, this list of conditions and the following disclaimer in the
15  *    documentation and/or other materials provided with the distribution.
16  * 3. Neither the name of the project nor the names of its contributors
17  *    may be used to endorse or promote products derived from this software
18  *    without specific prior written permission.
19  *
20  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
21  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
24  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30  * SUCH DAMAGE.
31  */
32 
33 /*
34  * Copyright (c) 1988, 1991, 1993
35  *	The Regents of the University of California.  All rights reserved.
36  *
37  * Redistribution and use in source and binary forms, with or without
38  * modification, are permitted provided that the following conditions
39  * are met:
40  * 1. Redistributions of source code must retain the above copyright
41  *    notice, this list of conditions and the following disclaimer.
42  * 2. Redistributions in binary form must reproduce the above copyright
43  *    notice, this list of conditions and the following disclaimer in the
44  *    documentation and/or other materials provided with the distribution.
45  * 3. All advertising materials mentioning features or use of this software
46  *    must display the following acknowledgement:
47  *	This product includes software developed by the University of
48  *	California, Berkeley and its contributors.
49  * 4. Neither the name of the University nor the names of its contributors
50  *    may be used to endorse or promote products derived from this software
51  *    without specific prior written permission.
52  *
53  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
54  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
55  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
56  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
57  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
58  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
59  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
60  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
61  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
62  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
63  * SUCH DAMAGE.
64  *
65  *	@(#)rtsock.c	8.6 (Berkeley) 2/11/95
66  */
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 
77 #include <vm/vm.h>
78 #include <sys/sysctl.h>
79 
80 #include <net/if.h>
81 #include <net/route.h>
82 #include <net/raw_cb.h>
83 
84 #include <machine/stdarg.h>
85 
86 struct	sockaddr route_dst = { 2, PF_ROUTE, };
87 struct	sockaddr route_src = { 2, PF_ROUTE, };
88 struct	sockproto route_proto = { PF_ROUTE, };
89 
90 struct walkarg {
91 	int	w_op, w_arg, w_given, w_needed, w_tmemsize;
92 	caddr_t	w_where, w_tmem;
93 };
94 
95 static struct mbuf *
96 		rt_msg1 __P((int, struct rt_addrinfo *));
97 static int	rt_msg2 __P((int,
98 		    struct rt_addrinfo *, caddr_t, struct walkarg *));
99 static void	rt_xaddrs __P((caddr_t, caddr_t, struct rt_addrinfo *));
100 
101 /* Sleazy use of local variables throughout file, warning!!!! */
102 #define dst	info.rti_info[RTAX_DST]
103 #define gate	info.rti_info[RTAX_GATEWAY]
104 #define netmask	info.rti_info[RTAX_NETMASK]
105 #define genmask	info.rti_info[RTAX_GENMASK]
106 #define ifpaddr	info.rti_info[RTAX_IFP]
107 #define ifaaddr	info.rti_info[RTAX_IFA]
108 #define brdaddr	info.rti_info[RTAX_BRD]
109 
110 /*ARGSUSED*/
111 int
112 route_usrreq(so, req, m, nam, control)
113 	register struct socket *so;
114 	int req;
115 	struct mbuf *m, *nam, *control;
116 {
117 	register int error = 0;
118 	register struct rawcb *rp = sotorawcb(so);
119 	int s;
120 
121 	if (req == PRU_ATTACH) {
122 		MALLOC(rp, struct rawcb *, sizeof(*rp), M_PCB, M_WAITOK);
123 		so->so_pcb = rp;
124 		bzero(so->so_pcb, sizeof(*rp));
125 	}
126 	if (req == PRU_DETACH && rp) {
127 		int af = rp->rcb_proto.sp_protocol;
128 		if (af == AF_INET)
129 			route_cb.ip_count--;
130 		else if (af == AF_INET6)
131 			route_cb.ip6_count--;
132 		else if (af == AF_NS)
133 			route_cb.ns_count--;
134 		else if (af == AF_ISO)
135 			route_cb.iso_count--;
136 		route_cb.any_count--;
137 	}
138 	s = splsoftnet();
139 	/*
140 	 * Don't call raw_usrreq() in the attach case, because
141 	 * we want to allow non-privileged processes to listen on
142 	 * and send "safe" commands to the routing socket.
143 	 */
144 	if (req == PRU_ATTACH) {
145 		if (curproc == 0)
146 			error = EACCES;
147 		else
148 			error = raw_attach(so, (int)(long)nam);
149 	} else
150 		error = raw_usrreq(so, req, m, nam, control);
151 
152 	rp = sotorawcb(so);
153 	if (req == PRU_ATTACH && rp) {
154 		int af = rp->rcb_proto.sp_protocol;
155 		if (error) {
156 			free((caddr_t)rp, M_PCB);
157 			splx(s);
158 			return (error);
159 		}
160 		if (af == AF_INET)
161 			route_cb.ip_count++;
162 		else if (af == AF_INET6)
163 			route_cb.ip6_count++;
164 		else if (af == AF_NS)
165 			route_cb.ns_count++;
166 		else if (af == AF_ISO)
167 			route_cb.iso_count++;
168 		rp->rcb_faddr = &route_src;
169 		route_cb.any_count++;
170 		soisconnected(so);
171 		so->so_options |= SO_USELOOPBACK;
172 	}
173 	splx(s);
174 	return (error);
175 }
176 
177 /*ARGSUSED*/
178 int
179 #if __STDC__
180 route_output(struct mbuf *m, ...)
181 #else
182 route_output(m, va_alist)
183 	struct mbuf *m;
184 	va_dcl
185 #endif
186 {
187 	register struct rt_msghdr *rtm = 0;
188 	register struct radix_node *rn = 0;
189 	register struct rtentry *rt = 0;
190 	struct rtentry *saved_nrt = 0;
191 	struct radix_node_head *rnh;
192 	struct rt_addrinfo info;
193 	int len, error = 0;
194 	struct ifnet *ifp = 0;
195 	struct ifaddr *ifa = 0;
196 	struct socket *so;
197 	va_list ap;
198 
199 	va_start(ap, m);
200 	so = va_arg(ap, struct socket *);
201 	va_end(ap);
202 
203 #define senderr(e) do { error = e; goto flush;} while (0)
204 	if (m == 0 || ((m->m_len < sizeof(int32_t)) &&
205 		       (m = m_pullup(m, sizeof(int32_t))) == 0))
206 		return (ENOBUFS);
207 	if ((m->m_flags & M_PKTHDR) == 0)
208 		panic("route_output");
209 	len = m->m_pkthdr.len;
210 	if (len < sizeof(*rtm) ||
211 	    len != mtod(m, struct rt_msghdr *)->rtm_msglen) {
212 		dst = 0;
213 		senderr(EINVAL);
214 	}
215 	R_Malloc(rtm, struct rt_msghdr *, len);
216 	if (rtm == 0) {
217 		dst = 0;
218 		senderr(ENOBUFS);
219 	}
220 	m_copydata(m, 0, len, (caddr_t)rtm);
221 	if (rtm->rtm_version != RTM_VERSION) {
222 		dst = 0;
223 		senderr(EPROTONOSUPPORT);
224 	}
225 	rtm->rtm_pid = curproc->p_pid;
226 	bzero(&info, sizeof(info));
227 	info.rti_addrs = rtm->rtm_addrs;
228 	rt_xaddrs((caddr_t)(rtm + 1), len + (caddr_t)rtm, &info);
229 	info.rti_flags = rtm->rtm_flags;
230 	if (dst == 0 || (dst->sa_family >= AF_MAX))
231 		senderr(EINVAL);
232 	if (gate != 0 && (gate->sa_family >= AF_MAX))
233 		senderr(EINVAL);
234 	if (genmask) {
235 		struct radix_node *t;
236 		t = rn_addmask((caddr_t)genmask, 0, 1);
237 		if (t && genmask->sa_len >= ((struct sockaddr *)t->rn_key)->sa_len &&
238 		    Bcmp((caddr_t *)genmask + 1, (caddr_t *)t->rn_key + 1,
239 		    ((struct sockaddr *)t->rn_key)->sa_len) - 1)
240 			genmask = (struct sockaddr *)(t->rn_key);
241 		else
242 			senderr(ENOBUFS);
243 	}
244 
245 	/*
246 	 * Verify that the caller has the appropriate privilege; RTM_GET
247 	 * is the only operation the non-superuser is allowed.
248 	 */
249 	if (rtm->rtm_type != RTM_GET &&
250 	    suser(curproc->p_ucred, &curproc->p_acflag) != 0)
251 		senderr(EACCES);
252 	switch (rtm->rtm_type) {
253 
254 	case RTM_DELETE:
255 		error = rtrequest1(rtm->rtm_type, &info, &saved_nrt);
256 		if (error == 0) {
257 			(rt = saved_nrt)->rt_refcnt++;
258 			goto report;
259 		}
260 		break;
261 
262 	case RTM_ADD:
263 		if (gate == 0)
264 			senderr(EINVAL);
265 		error = rtrequest1(rtm->rtm_type, &info, &saved_nrt);
266 		if (error == 0 && saved_nrt) {
267 			rt_setmetrics(rtm->rtm_inits,
268 				&rtm->rtm_rmx, &saved_nrt->rt_rmx);
269 			saved_nrt->rt_refcnt--;
270 			saved_nrt->rt_genmask = genmask;
271 		}
272 		/* FALLTHROUGH */
273 	case RTM_GET:
274 	case RTM_CHANGE:
275 	case RTM_LOCK:
276 		if ((rnh = rt_tables[dst->sa_family]) == 0) {
277 			senderr(EAFNOSUPPORT);
278 		}
279 		rn = rnh->rnh_lookup(dst, netmask, rnh);
280 		if (rn == NULL || (rn->rn_flags & RNF_ROOT) != 0) {
281 			senderr(ESRCH);
282 		}
283 		rt = (struct rtentry *)rn;
284 		rt->rt_refcnt++;
285 
286 		switch(rtm->rtm_type) {
287 
288 		case RTM_GET:
289 		report:
290 			dst = rt_key(rt);
291 			gate = rt->rt_gateway;
292 			netmask = rt_mask(rt);
293 			genmask = rt->rt_genmask;
294 			if (rtm->rtm_addrs & (RTA_IFP | RTA_IFA)) {
295 				if ((ifp = rt->rt_ifp) != NULL) {
296 					ifpaddr = ifp->if_addrlist.tqh_first->ifa_addr;
297 					ifaaddr = rt->rt_ifa->ifa_addr;
298 					if (ifp->if_flags & IFF_POINTOPOINT)
299 						brdaddr = rt->rt_ifa->ifa_dstaddr;
300 					else
301 						brdaddr = 0;
302 					rtm->rtm_index = ifp->if_index;
303 				} else {
304 					ifpaddr = 0;
305 					ifaaddr = 0;
306 				}
307 			}
308 			len = rt_msg2(rtm->rtm_type, &info, (caddr_t)0,
309 				(struct walkarg *)0);
310 			if (len > rtm->rtm_msglen) {
311 				struct rt_msghdr *new_rtm;
312 				R_Malloc(new_rtm, struct rt_msghdr *, len);
313 				if (new_rtm == 0)
314 					senderr(ENOBUFS);
315 				Bcopy(rtm, new_rtm, rtm->rtm_msglen);
316 				Free(rtm); rtm = new_rtm;
317 			}
318 			(void)rt_msg2(rtm->rtm_type, &info, (caddr_t)rtm,
319 				(struct walkarg *)0);
320 			rtm->rtm_flags = rt->rt_flags;
321 			rtm->rtm_rmx = rt->rt_rmx;
322 			rtm->rtm_addrs = info.rti_addrs;
323 			break;
324 
325 		case RTM_CHANGE:
326 			/*
327 			 * new gateway could require new ifaddr, ifp;
328 			 * flags may also be different; ifp may be specified
329 			 * by ll sockaddr when protocol address is ambiguous
330 			 */
331 			if ((error = rt_getifa(&info)) != 0)
332 				senderr(error);
333 			if (gate && rt_setgate(rt, rt_key(rt), gate))
334 				senderr(EDQUOT);
335 			if (ifpaddr && (ifa = ifa_ifwithnet(ifpaddr)) &&
336 			    (ifp = ifa->ifa_ifp) && (ifaaddr || gate))
337 				ifa = ifaof_ifpforaddr(ifaaddr ? ifaaddr : gate,
338 							ifp);
339 			else if ((ifaaddr && (ifa = ifa_ifwithaddr(ifaaddr))) ||
340 				 (gate && (ifa = ifa_ifwithroute(rt->rt_flags,
341 							rt_key(rt), gate))))
342 				ifp = ifa->ifa_ifp;
343 			if (ifa) {
344 				register struct ifaddr *oifa = rt->rt_ifa;
345 				if (oifa != ifa) {
346 				    if (oifa && oifa->ifa_rtrequest)
347 					oifa->ifa_rtrequest(RTM_DELETE, rt,
348 					    &info);
349 				    IFAFREE(rt->rt_ifa);
350 				    rt->rt_ifa = ifa;
351 				    ifa->ifa_refcnt++;
352 				    rt->rt_ifp = ifp;
353 				}
354 			}
355 			rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx,
356 					&rt->rt_rmx);
357 			if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest)
358 			       rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, &info);
359 			if (genmask)
360 				rt->rt_genmask = genmask;
361 			/*
362 			 * Fall into
363 			 */
364 		case RTM_ADD:
365 		case RTM_LOCK:
366 			rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits);
367 			rt->rt_rmx.rmx_locks |=
368 				(rtm->rtm_inits & rtm->rtm_rmx.rmx_locks);
369 			break;
370 		}
371 		break;
372 
373 	default:
374 		senderr(EOPNOTSUPP);
375 	}
376 
377 flush:
378 	if (rtm) {
379 		if (error)
380 			rtm->rtm_errno = error;
381 		else
382 			rtm->rtm_flags |= RTF_DONE;
383 	}
384 	if (rt)
385 		rtfree(rt);
386     {
387 	register struct rawcb *rp = 0;
388 	/*
389 	 * Check to see if we don't want our own messages.
390 	 */
391 	if ((so->so_options & SO_USELOOPBACK) == 0) {
392 		if (route_cb.any_count <= 1) {
393 			if (rtm)
394 				Free(rtm);
395 			m_freem(m);
396 			return (error);
397 		}
398 		/* There is another listener, so construct message */
399 		rp = sotorawcb(so);
400 	}
401 	if (rtm) {
402 		m_copyback(m, 0, rtm->rtm_msglen, (caddr_t)rtm);
403 		if (m->m_pkthdr.len < rtm->rtm_msglen) {
404 			m_freem(m);
405 			m = NULL;
406 		} else if (m->m_pkthdr.len > rtm->rtm_msglen)
407 			m_adj(m, rtm->rtm_msglen - m->m_pkthdr.len);
408 		Free(rtm);
409 	}
410 	if (rp)
411 		rp->rcb_proto.sp_family = 0; /* Avoid us */
412 	if (dst)
413 		route_proto.sp_protocol = dst->sa_family;
414 	if (m)
415 		raw_input(m, &route_proto, &route_src, &route_dst);
416 	if (rp)
417 		rp->rcb_proto.sp_family = PF_ROUTE;
418     }
419 	return (error);
420 }
421 
422 void
423 rt_setmetrics(which, in, out)
424 	u_long which;
425 	register struct rt_metrics *in, *out;
426 {
427 #define metric(f, e) if (which & (f)) out->e = in->e;
428 	metric(RTV_RPIPE, rmx_recvpipe);
429 	metric(RTV_SPIPE, rmx_sendpipe);
430 	metric(RTV_SSTHRESH, rmx_ssthresh);
431 	metric(RTV_RTT, rmx_rtt);
432 	metric(RTV_RTTVAR, rmx_rttvar);
433 	metric(RTV_HOPCOUNT, rmx_hopcount);
434 	metric(RTV_MTU, rmx_mtu);
435 	metric(RTV_EXPIRE, rmx_expire);
436 #undef metric
437 }
438 
439 #define ROUNDUP(a) \
440 	((a) > 0 ? (1 + (((a) - 1) | (sizeof(long) - 1))) : sizeof(long))
441 #define ADVANCE(x, n) (x += ROUNDUP((n)->sa_len))
442 
443 static void
444 rt_xaddrs(cp, cplim, rtinfo)
445 	register caddr_t cp, cplim;
446 	register struct rt_addrinfo *rtinfo;
447 {
448 	register struct sockaddr *sa;
449 	register int i;
450 
451 	bzero(rtinfo->rti_info, sizeof(rtinfo->rti_info));
452 	for (i = 0; (i < RTAX_MAX) && (cp < cplim); i++) {
453 		if ((rtinfo->rti_addrs & (1 << i)) == 0)
454 			continue;
455 		rtinfo->rti_info[i] = sa = (struct sockaddr *)cp;
456 		ADVANCE(cp, sa);
457 	}
458 }
459 
460 /*
461  * Copy data from a buffer back into the indicated mbuf chain,
462  * starting "off" bytes from the beginning, extending the mbuf
463  * chain if necessary. The mbuf needs to be properly initalized
464  * including the setting of m_len.
465  */
466 void
467 m_copyback(m0, off, len, cp)
468 	struct	mbuf *m0;
469 	register int off;
470 	register int len;
471 	caddr_t cp;
472 {
473 	register int mlen;
474 	register struct mbuf *m = m0, *n;
475 	int totlen = 0;
476 
477 	if (m0 == 0)
478 		return;
479 	while (off > (mlen = m->m_len)) {
480 		off -= mlen;
481 		totlen += mlen;
482 		if (m->m_next == 0) {
483 			n = m_getclr(M_DONTWAIT, m->m_type);
484 			if (n == 0)
485 				goto out;
486 			n->m_len = min(MLEN, len + off);
487 			m->m_next = n;
488 		}
489 		m = m->m_next;
490 	}
491 	while (len > 0) {
492 		mlen = min (m->m_len - off, len);
493 		bcopy(cp, off + mtod(m, caddr_t), (unsigned)mlen);
494 		cp += mlen;
495 		len -= mlen;
496 		mlen += off;
497 		off = 0;
498 		totlen += mlen;
499 		if (len == 0)
500 			break;
501 		if (m->m_next == 0) {
502 			n = m_get(M_DONTWAIT, m->m_type);
503 			if (n == 0)
504 				break;
505 			n->m_len = min(MLEN, len);
506 			m->m_next = n;
507 		}
508 		m = m->m_next;
509 	}
510 out:	if (((m = m0)->m_flags & M_PKTHDR) && (m->m_pkthdr.len < totlen))
511 		m->m_pkthdr.len = totlen;
512 }
513 
514 static struct mbuf *
515 rt_msg1(type, rtinfo)
516 	int type;
517 	register struct rt_addrinfo *rtinfo;
518 {
519 	register struct rt_msghdr *rtm;
520 	register struct mbuf *m;
521 	register int i;
522 	register struct sockaddr *sa;
523 	int len, dlen;
524 
525 	switch (type) {
526 
527 	case RTM_DELADDR:
528 	case RTM_NEWADDR:
529 		len = sizeof(struct ifa_msghdr);
530 		break;
531 
532 	case RTM_IFINFO:
533 		len = sizeof(struct if_msghdr);
534 		break;
535 
536 	default:
537 		len = sizeof(struct rt_msghdr);
538 	}
539 	if (len > MCLBYTES)
540 		panic("rt_msg1");
541 	m = m_gethdr(M_DONTWAIT, MT_DATA);
542 	if (m && len > MHLEN) {
543 		MCLGET(m, M_DONTWAIT);
544 		if ((m->m_flags & M_EXT) == 0) {
545 			m_free(m);
546 			m = NULL;
547 		}
548 	}
549 	if (m == 0)
550 		return (m);
551 	m->m_pkthdr.len = m->m_len = len;
552 	m->m_pkthdr.rcvif = 0;
553 	rtm = mtod(m, struct rt_msghdr *);
554 	bzero((caddr_t)rtm, len);
555 	for (i = 0; i < RTAX_MAX; i++) {
556 		if ((sa = rtinfo->rti_info[i]) == NULL)
557 			continue;
558 		rtinfo->rti_addrs |= (1 << i);
559 		dlen = ROUNDUP(sa->sa_len);
560 		m_copyback(m, len, dlen, (caddr_t)sa);
561 		len += dlen;
562 	}
563 	if (m->m_pkthdr.len != len) {
564 		m_freem(m);
565 		return (NULL);
566 	}
567 	rtm->rtm_msglen = len;
568 	rtm->rtm_version = RTM_VERSION;
569 	rtm->rtm_type = type;
570 	return (m);
571 }
572 
573 static int
574 rt_msg2(type, rtinfo, cp, w)
575 	int type;
576 	register struct rt_addrinfo *rtinfo;
577 	caddr_t cp;
578 	struct walkarg *w;
579 {
580 	register int i;
581 	int len, dlen, second_time = 0;
582 	caddr_t cp0;
583 
584 	rtinfo->rti_addrs = 0;
585 again:
586 	switch (type) {
587 
588 	case RTM_DELADDR:
589 	case RTM_NEWADDR:
590 		len = sizeof(struct ifa_msghdr);
591 		break;
592 
593 	case RTM_IFINFO:
594 		len = sizeof(struct if_msghdr);
595 		break;
596 
597 	default:
598 		len = sizeof(struct rt_msghdr);
599 	}
600 	if ((cp0 = cp) != NULL)
601 		cp += len;
602 	for (i = 0; i < RTAX_MAX; i++) {
603 		register struct sockaddr *sa;
604 
605 		if ((sa = rtinfo->rti_info[i]) == 0)
606 			continue;
607 		rtinfo->rti_addrs |= (1 << i);
608 		dlen = ROUNDUP(sa->sa_len);
609 		if (cp) {
610 			bcopy((caddr_t)sa, cp, (unsigned)dlen);
611 			cp += dlen;
612 		}
613 		len += dlen;
614 	}
615 	if (cp == 0 && w != NULL && !second_time) {
616 		register struct walkarg *rw = w;
617 
618 		rw->w_needed += len;
619 		if (rw->w_needed <= 0 && rw->w_where) {
620 			if (rw->w_tmemsize < len) {
621 				if (rw->w_tmem)
622 					free(rw->w_tmem, M_RTABLE);
623 				rw->w_tmem = (caddr_t) malloc(len, M_RTABLE,
624 							      M_NOWAIT);
625 				if (rw->w_tmem)
626 					rw->w_tmemsize = len;
627 			}
628 			if (rw->w_tmem) {
629 				cp = rw->w_tmem;
630 				second_time = 1;
631 				goto again;
632 			} else
633 				rw->w_where = 0;
634 		}
635 	}
636 	if (cp) {
637 		register struct rt_msghdr *rtm = (struct rt_msghdr *)cp0;
638 
639 		rtm->rtm_version = RTM_VERSION;
640 		rtm->rtm_type = type;
641 		rtm->rtm_msglen = len;
642 	}
643 	return (len);
644 }
645 
646 /*
647  * This routine is called to generate a message from the routing
648  * socket indicating that a redirect has occured, a routing lookup
649  * has failed, or that a protocol has detected timeouts to a particular
650  * destination.
651  */
652 void
653 rt_missmsg(type, rtinfo, flags, error)
654 	int type, flags, error;
655 	register struct rt_addrinfo *rtinfo;
656 {
657 	register struct rt_msghdr *rtm;
658 	register struct mbuf *m;
659 	struct sockaddr *sa = rtinfo->rti_info[RTAX_DST];
660 
661 	if (route_cb.any_count == 0)
662 		return;
663 	m = rt_msg1(type, rtinfo);
664 	if (m == 0)
665 		return;
666 	rtm = mtod(m, struct rt_msghdr *);
667 	rtm->rtm_flags = RTF_DONE | flags;
668 	rtm->rtm_errno = error;
669 	rtm->rtm_addrs = rtinfo->rti_addrs;
670 	route_proto.sp_protocol = sa ? sa->sa_family : 0;
671 	raw_input(m, &route_proto, &route_src, &route_dst);
672 }
673 
674 /*
675  * This routine is called to generate a message from the routing
676  * socket indicating that the status of a network interface has changed.
677  */
678 void
679 rt_ifmsg(ifp)
680 	register struct ifnet *ifp;
681 {
682 	register struct if_msghdr *ifm;
683 	struct mbuf *m;
684 	struct rt_addrinfo info;
685 
686 	if (route_cb.any_count == 0)
687 		return;
688 	bzero((caddr_t)&info, sizeof(info));
689 	m = rt_msg1(RTM_IFINFO, &info);
690 	if (m == 0)
691 		return;
692 	ifm = mtod(m, struct if_msghdr *);
693 	ifm->ifm_index = ifp->if_index;
694 	ifm->ifm_flags = ifp->if_flags;
695 	ifm->ifm_data = ifp->if_data;
696 	ifm->ifm_addrs = 0;
697 	route_proto.sp_protocol = 0;
698 	raw_input(m, &route_proto, &route_src, &route_dst);
699 }
700 
701 /*
702  * This is called to generate messages from the routing socket
703  * indicating a network interface has had addresses associated with it.
704  * if we ever reverse the logic and replace messages TO the routing
705  * socket indicate a request to configure interfaces, then it will
706  * be unnecessary as the routing socket will automatically generate
707  * copies of it.
708  */
709 void
710 rt_newaddrmsg(cmd, ifa, error, rt)
711 	int cmd, error;
712 	register struct ifaddr *ifa;
713 	register struct rtentry *rt;
714 {
715 	struct rt_addrinfo info;
716 	struct sockaddr *sa = NULL;
717 	int pass;
718 	struct mbuf *m = NULL;
719 	struct ifnet *ifp = ifa->ifa_ifp;
720 
721 	if (route_cb.any_count == 0)
722 		return;
723 	for (pass = 1; pass < 3; pass++) {
724 		bzero((caddr_t)&info, sizeof(info));
725 		if ((cmd == RTM_ADD && pass == 1) ||
726 		    (cmd == RTM_DELETE && pass == 2)) {
727 			register struct ifa_msghdr *ifam;
728 			int ncmd = cmd == RTM_ADD ? RTM_NEWADDR : RTM_DELADDR;
729 
730 			ifaaddr = sa = ifa->ifa_addr;
731 			ifpaddr = ifp->if_addrlist.tqh_first->ifa_addr;
732 			netmask = ifa->ifa_netmask;
733 			brdaddr = ifa->ifa_dstaddr;
734 			if ((m = rt_msg1(ncmd, &info)) == NULL)
735 				continue;
736 			ifam = mtod(m, struct ifa_msghdr *);
737 			ifam->ifam_index = ifp->if_index;
738 			ifam->ifam_metric = ifa->ifa_metric;
739 			ifam->ifam_flags = ifa->ifa_flags;
740 			ifam->ifam_addrs = info.rti_addrs;
741 		}
742 		if ((cmd == RTM_ADD && pass == 2) ||
743 		    (cmd == RTM_DELETE && pass == 1)) {
744 			register struct rt_msghdr *rtm;
745 
746 			if (rt == 0)
747 				continue;
748 			netmask = rt_mask(rt);
749 			dst = sa = rt_key(rt);
750 			gate = rt->rt_gateway;
751 			if ((m = rt_msg1(cmd, &info)) == NULL)
752 				continue;
753 			rtm = mtod(m, struct rt_msghdr *);
754 			rtm->rtm_index = ifp->if_index;
755 			rtm->rtm_flags |= rt->rt_flags;
756 			rtm->rtm_errno = error;
757 			rtm->rtm_addrs = info.rti_addrs;
758 		}
759 		route_proto.sp_protocol = sa ? sa->sa_family : 0;
760 		raw_input(m, &route_proto, &route_src, &route_dst);
761 	}
762 }
763 
764 /*
765  * This is used in dumping the kernel table via sysctl().
766  */
767 int
768 sysctl_dumpentry(rn, v)
769 	struct radix_node *rn;
770 	register void *v;
771 {
772 	register struct walkarg *w = v;
773 	register struct rtentry *rt = (struct rtentry *)rn;
774 	int error = 0, size;
775 	struct rt_addrinfo info;
776 
777 	if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg))
778 		return 0;
779 	bzero((caddr_t)&info, sizeof(info));
780 	dst = rt_key(rt);
781 	gate = rt->rt_gateway;
782 	netmask = rt_mask(rt);
783 	genmask = rt->rt_genmask;
784 	if (rt->rt_ifp) {
785 		ifpaddr = rt->rt_ifp->if_addrlist.tqh_first->ifa_addr;
786 		ifaaddr = rt->rt_ifa->ifa_addr;
787 		if (rt->rt_ifp->if_flags & IFF_POINTOPOINT)
788 			brdaddr = rt->rt_ifa->ifa_dstaddr;
789 	}
790 	size = rt_msg2(RTM_GET, &info, 0, w);
791 	if (w->w_where && w->w_tmem) {
792 		register struct rt_msghdr *rtm = (struct rt_msghdr *)w->w_tmem;
793 
794 		rtm->rtm_flags = rt->rt_flags;
795 		rtm->rtm_use = rt->rt_use;
796 		rtm->rtm_rmx = rt->rt_rmx;
797 		rtm->rtm_index = rt->rt_ifp->if_index;
798 		rtm->rtm_errno = rtm->rtm_pid = rtm->rtm_seq = 0;
799 		rtm->rtm_addrs = info.rti_addrs;
800 		if ((error = copyout((caddr_t)rtm, w->w_where, size)) != 0)
801 			w->w_where = NULL;
802 		else
803 			w->w_where += size;
804 	}
805 	return (error);
806 }
807 
808 int
809 sysctl_iflist(af, w)
810 	int	af;
811 	register struct	walkarg *w;
812 {
813 	register struct ifnet *ifp;
814 	register struct ifaddr *ifa;
815 	struct	rt_addrinfo info;
816 	int	len, error = 0;
817 
818 	bzero((caddr_t)&info, sizeof(info));
819 	for (ifp = ifnet.tqh_first; ifp != 0; ifp = ifp->if_list.tqe_next) {
820 		if (w->w_arg && w->w_arg != ifp->if_index)
821 			continue;
822 		ifa = ifp->if_addrlist.tqh_first;
823 		ifpaddr = ifa->ifa_addr;
824 		len = rt_msg2(RTM_IFINFO, &info, (caddr_t)0, w);
825 		ifpaddr = 0;
826 		if (w->w_where && w->w_tmem) {
827 			register struct if_msghdr *ifm;
828 
829 			ifm = (struct if_msghdr *)w->w_tmem;
830 			ifm->ifm_index = ifp->if_index;
831 			ifm->ifm_flags = ifp->if_flags;
832 			ifm->ifm_data = ifp->if_data;
833 			ifm->ifm_addrs = info.rti_addrs;
834 			error = copyout((caddr_t)ifm, w->w_where, len);
835 			if (error)
836 				return (error);
837 			w->w_where += len;
838 		}
839 		while ((ifa = ifa->ifa_list.tqe_next) != NULL) {
840 			if (af && af != ifa->ifa_addr->sa_family)
841 				continue;
842 			ifaaddr = ifa->ifa_addr;
843 			netmask = ifa->ifa_netmask;
844 			brdaddr = ifa->ifa_dstaddr;
845 			len = rt_msg2(RTM_NEWADDR, &info, 0, w);
846 			if (w->w_where && w->w_tmem) {
847 				register struct ifa_msghdr *ifam;
848 
849 				ifam = (struct ifa_msghdr *)w->w_tmem;
850 				ifam->ifam_index = ifa->ifa_ifp->if_index;
851 				ifam->ifam_flags = ifa->ifa_flags;
852 				ifam->ifam_metric = ifa->ifa_metric;
853 				ifam->ifam_addrs = info.rti_addrs;
854 				error = copyout(w->w_tmem, w->w_where, len);
855 				if (error)
856 					return (error);
857 				w->w_where += len;
858 			}
859 		}
860 		ifaaddr = netmask = brdaddr = 0;
861 	}
862 	return (0);
863 }
864 
865 int
866 sysctl_rtable(name, namelen, where, given, new, newlen)
867 	int	*name;
868 	u_int	namelen;
869 	void 	*where;
870 	size_t	*given;
871 	void	*new;
872 	size_t	newlen;
873 {
874 	register struct radix_node_head *rnh;
875 	int	i, s, error = EINVAL;
876 	u_char  af;
877 	struct	walkarg w;
878 
879 	if (new)
880 		return (EPERM);
881 	if (namelen != 3)
882 		return (EINVAL);
883 	af = name[0];
884 	Bzero(&w, sizeof(w));
885 	w.w_where = where;
886 	w.w_given = *given;
887 	w.w_needed = 0 - w.w_given;
888 	w.w_op = name[1];
889 	w.w_arg = name[2];
890 
891 	s = splsoftnet();
892 	switch (w.w_op) {
893 
894 	case NET_RT_DUMP:
895 	case NET_RT_FLAGS:
896 		for (i = 1; i <= AF_MAX; i++)
897 			if ((rnh = rt_tables[i]) && (af == 0 || af == i) &&
898 			    (error = (*rnh->rnh_walktree)(rnh,
899 							  sysctl_dumpentry,
900 							  &w)))
901 				break;
902 		break;
903 
904 	case NET_RT_IFLIST:
905 		error = sysctl_iflist(af, &w);
906 	}
907 	splx(s);
908 	if (w.w_tmem)
909 		free(w.w_tmem, M_RTABLE);
910 	w.w_needed += w.w_given;
911 	if (where) {
912 		*given = w.w_where - (caddr_t) where;
913 		if (*given < w.w_needed)
914 			return (ENOMEM);
915 	} else {
916 		*given = (11 * w.w_needed) / 10;
917 	}
918 	return (error);
919 }
920 
921 /*
922  * Definitions of protocols supported in the ROUTE domain.
923  */
924 
925 extern	struct domain routedomain;		/* or at least forward */
926 
927 struct protosw routesw[] = {
928 { SOCK_RAW,	&routedomain,	0,		PR_ATOMIC|PR_ADDR,
929   raw_input,	route_output,	raw_ctlinput,	0,
930   route_usrreq,
931   raw_init,	0,		0,		0,
932   sysctl_rtable,
933 }
934 };
935 
936 struct domain routedomain =
937     { PF_ROUTE, "route", route_init, 0, 0,
938       routesw, &routesw[sizeof(routesw)/sizeof(routesw[0])] };
939