xref: /freebsd/sys/net/rtsock.c (revision f9218d3d4fd34f082473b3a021c6d4d109fb47cf)
1 /*
2  * Copyright (c) 1988, 1991, 1993
3  *	The Regents of the University of California.  All rights reserved.
4  *
5  * Redistribution and use in source and binary forms, with or without
6  * modification, are permitted provided that the following conditions
7  * are met:
8  * 1. Redistributions of source code must retain the above copyright
9  *    notice, this list of conditions and the following disclaimer.
10  * 2. Redistributions in binary form must reproduce the above copyright
11  *    notice, this list of conditions and the following disclaimer in the
12  *    documentation and/or other materials provided with the distribution.
13  * 3. All advertising materials mentioning features or use of this software
14  *    must display the following acknowledgement:
15  *	This product includes software developed by the University of
16  *	California, Berkeley and its contributors.
17  * 4. Neither the name of the University nor the names of its contributors
18  *    may be used to endorse or promote products derived from this software
19  *    without specific prior written permission.
20  *
21  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
22  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
25  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31  * SUCH DAMAGE.
32  *
33  *	@(#)rtsock.c	8.7 (Berkeley) 10/12/95
34  * $FreeBSD$
35  */
36 
37 
38 #include <sys/param.h>
39 #include <sys/domain.h>
40 #include <sys/kernel.h>
41 #include <sys/jail.h>
42 #include <sys/malloc.h>
43 #include <sys/mbuf.h>
44 #include <sys/proc.h>
45 #include <sys/protosw.h>
46 #include <sys/signalvar.h>
47 #include <sys/socket.h>
48 #include <sys/socketvar.h>
49 #include <sys/sysctl.h>
50 #include <sys/systm.h>
51 
52 #include <net/if.h>
53 #include <net/raw_cb.h>
54 #include <net/route.h>
55 
56 MALLOC_DEFINE(M_RTABLE, "routetbl", "routing tables");
57 
58 static struct	sockaddr route_dst = { 2, PF_ROUTE, };
59 static struct	sockaddr route_src = { 2, PF_ROUTE, };
60 static struct	sockaddr sa_zero   = { sizeof(sa_zero), AF_INET, };
61 static struct	sockproto route_proto = { PF_ROUTE, };
62 
63 struct walkarg {
64 	int	w_tmemsize;
65 	int	w_op, w_arg;
66 	caddr_t	w_tmem;
67 	struct sysctl_req *w_req;
68 };
69 
70 static struct mbuf *
71 		rt_msg1(int, struct rt_addrinfo *);
72 static int	rt_msg2(int, struct rt_addrinfo *, caddr_t, struct walkarg *);
73 static int	rt_xaddrs(caddr_t, caddr_t, struct rt_addrinfo *);
74 static int	sysctl_dumpentry(struct radix_node *rn, void *vw);
75 static int	sysctl_iflist(int af, struct walkarg *w);
76 static int	 route_output(struct mbuf *, struct socket *);
77 static void	 rt_setmetrics(u_long, struct rt_metrics *, struct rt_metrics *);
78 
79 /* Sleazy use of local variables throughout file, warning!!!! */
80 #define dst	info.rti_info[RTAX_DST]
81 #define gate	info.rti_info[RTAX_GATEWAY]
82 #define netmask	info.rti_info[RTAX_NETMASK]
83 #define genmask	info.rti_info[RTAX_GENMASK]
84 #define ifpaddr	info.rti_info[RTAX_IFP]
85 #define ifaaddr	info.rti_info[RTAX_IFA]
86 #define brdaddr	info.rti_info[RTAX_BRD]
87 
88 /*
89  * It really doesn't make any sense at all for this code to share much
90  * with raw_usrreq.c, since its functionality is so restricted.  XXX
91  */
92 static int
93 rts_abort(struct socket *so)
94 {
95 	int s, error;
96 	s = splnet();
97 	error = raw_usrreqs.pru_abort(so);
98 	splx(s);
99 	return error;
100 }
101 
102 /* pru_accept is EOPNOTSUPP */
103 
104 static int
105 rts_attach(struct socket *so, int proto, struct thread *td)
106 {
107 	struct rawcb *rp;
108 	int s, error;
109 
110 	if (sotorawcb(so) != 0)
111 		return EISCONN;	/* XXX panic? */
112 	/* XXX */
113 	MALLOC(rp, struct rawcb *, sizeof *rp, M_PCB, M_WAITOK | M_ZERO);
114 	if (rp == 0)
115 		return ENOBUFS;
116 
117 	/*
118 	 * The splnet() is necessary to block protocols from sending
119 	 * error notifications (like RTM_REDIRECT or RTM_LOSING) while
120 	 * this PCB is extant but incompletely initialized.
121 	 * Probably we should try to do more of this work beforehand and
122 	 * eliminate the spl.
123 	 */
124 	s = splnet();
125 	so->so_pcb = (caddr_t)rp;
126 	error = raw_attach(so, proto);
127 	rp = sotorawcb(so);
128 	if (error) {
129 		splx(s);
130 		so->so_pcb = NULL;
131 		free(rp, M_PCB);
132 		return error;
133 	}
134 	switch(rp->rcb_proto.sp_protocol) {
135 	case AF_INET:
136 		route_cb.ip_count++;
137 		break;
138 	case AF_INET6:
139 		route_cb.ip6_count++;
140 		break;
141 	case AF_IPX:
142 		route_cb.ipx_count++;
143 		break;
144 	case AF_NS:
145 		route_cb.ns_count++;
146 		break;
147 	}
148 	rp->rcb_faddr = &route_src;
149 	route_cb.any_count++;
150 	soisconnected(so);
151 	so->so_options |= SO_USELOOPBACK;
152 	splx(s);
153 	return 0;
154 }
155 
156 static int
157 rts_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
158 {
159 	int s, error;
160 	s = splnet();
161 	error = raw_usrreqs.pru_bind(so, nam, td); /* xxx just EINVAL */
162 	splx(s);
163 	return error;
164 }
165 
166 static int
167 rts_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
168 {
169 	int s, error;
170 	s = splnet();
171 	error = raw_usrreqs.pru_connect(so, nam, td); /* XXX just EINVAL */
172 	splx(s);
173 	return error;
174 }
175 
176 /* pru_connect2 is EOPNOTSUPP */
177 /* pru_control is EOPNOTSUPP */
178 
179 static int
180 rts_detach(struct socket *so)
181 {
182 	struct rawcb *rp = sotorawcb(so);
183 	int s, error;
184 
185 	s = splnet();
186 	if (rp != 0) {
187 		switch(rp->rcb_proto.sp_protocol) {
188 		case AF_INET:
189 			route_cb.ip_count--;
190 			break;
191 		case AF_INET6:
192 			route_cb.ip6_count--;
193 			break;
194 		case AF_IPX:
195 			route_cb.ipx_count--;
196 			break;
197 		case AF_NS:
198 			route_cb.ns_count--;
199 			break;
200 		}
201 		route_cb.any_count--;
202 	}
203 	error = raw_usrreqs.pru_detach(so);
204 	splx(s);
205 	return error;
206 }
207 
208 static int
209 rts_disconnect(struct socket *so)
210 {
211 	int s, error;
212 	s = splnet();
213 	error = raw_usrreqs.pru_disconnect(so);
214 	splx(s);
215 	return error;
216 }
217 
218 /* pru_listen is EOPNOTSUPP */
219 
220 static int
221 rts_peeraddr(struct socket *so, struct sockaddr **nam)
222 {
223 	int s, error;
224 	s = splnet();
225 	error = raw_usrreqs.pru_peeraddr(so, nam);
226 	splx(s);
227 	return error;
228 }
229 
230 /* pru_rcvd is EOPNOTSUPP */
231 /* pru_rcvoob is EOPNOTSUPP */
232 
233 static int
234 rts_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
235 	 struct mbuf *control, struct thread *td)
236 {
237 	int s, error;
238 	s = splnet();
239 	error = raw_usrreqs.pru_send(so, flags, m, nam, control, td);
240 	splx(s);
241 	return error;
242 }
243 
244 /* pru_sense is null */
245 
246 static int
247 rts_shutdown(struct socket *so)
248 {
249 	int s, error;
250 	s = splnet();
251 	error = raw_usrreqs.pru_shutdown(so);
252 	splx(s);
253 	return error;
254 }
255 
256 static int
257 rts_sockaddr(struct socket *so, struct sockaddr **nam)
258 {
259 	int s, error;
260 	s = splnet();
261 	error = raw_usrreqs.pru_sockaddr(so, nam);
262 	splx(s);
263 	return error;
264 }
265 
266 static struct pr_usrreqs route_usrreqs = {
267 	rts_abort, pru_accept_notsupp, rts_attach, rts_bind, rts_connect,
268 	pru_connect2_notsupp, pru_control_notsupp, rts_detach, rts_disconnect,
269 	pru_listen_notsupp, rts_peeraddr, pru_rcvd_notsupp, pru_rcvoob_notsupp,
270 	rts_send, pru_sense_null, rts_shutdown, rts_sockaddr,
271 	sosend, soreceive, sopoll
272 };
273 
274 /*ARGSUSED*/
275 static int
276 route_output(m, so)
277 	register struct mbuf *m;
278 	struct socket *so;
279 {
280 	register struct rt_msghdr *rtm = 0;
281 	register struct rtentry *rt = 0;
282 	struct rtentry *saved_nrt = 0;
283 	struct radix_node_head *rnh;
284 	struct rt_addrinfo info;
285 	int len, error = 0;
286 	struct ifnet *ifp = 0;
287 	struct ifaddr *ifa = 0;
288 
289 #define senderr(e) { error = e; goto flush;}
290 	if (m == 0 || ((m->m_len < sizeof(long)) &&
291 		       (m = m_pullup(m, sizeof(long))) == 0))
292 		return (ENOBUFS);
293 	if ((m->m_flags & M_PKTHDR) == 0)
294 		panic("route_output");
295 	len = m->m_pkthdr.len;
296 	if (len < sizeof(*rtm) ||
297 	    len != mtod(m, struct rt_msghdr *)->rtm_msglen) {
298 		dst = 0;
299 		senderr(EINVAL);
300 	}
301 	R_Malloc(rtm, struct rt_msghdr *, len);
302 	if (rtm == 0) {
303 		dst = 0;
304 		senderr(ENOBUFS);
305 	}
306 	m_copydata(m, 0, len, (caddr_t)rtm);
307 	if (rtm->rtm_version != RTM_VERSION) {
308 		dst = 0;
309 		senderr(EPROTONOSUPPORT);
310 	}
311 	rtm->rtm_pid = curproc->p_pid;
312 	bzero(&info, sizeof(info));
313 	info.rti_addrs = rtm->rtm_addrs;
314 	if (rt_xaddrs((caddr_t)(rtm + 1), len + (caddr_t)rtm, &info)) {
315 		dst = 0;
316 		senderr(EINVAL);
317 	}
318 	info.rti_flags = rtm->rtm_flags;
319 	if (dst == 0 || (dst->sa_family >= AF_MAX)
320 	    || (gate != 0 && (gate->sa_family >= AF_MAX)))
321 		senderr(EINVAL);
322 	if (genmask) {
323 		struct radix_node *t;
324 		t = rn_addmask((caddr_t)genmask, 0, 1);
325 		if (t && Bcmp((caddr_t *)genmask + 1, (caddr_t *)t->rn_key + 1,
326 			      *(u_char *)t->rn_key - 1) == 0)
327 			genmask = (struct sockaddr *)(t->rn_key);
328 		else
329 			senderr(ENOBUFS);
330 	}
331 
332 	/*
333 	 * Verify that the caller has the appropriate privilege; RTM_GET
334 	 * is the only operation the non-superuser is allowed.
335 	 */
336 	if (rtm->rtm_type != RTM_GET && (error = suser(curthread)) != 0)
337 		senderr(error);
338 
339 	switch (rtm->rtm_type) {
340 
341 	case RTM_ADD:
342 		if (gate == 0)
343 			senderr(EINVAL);
344 		error = rtrequest1(RTM_ADD, &info, &saved_nrt);
345 		if (error == 0 && saved_nrt) {
346 			rt_setmetrics(rtm->rtm_inits,
347 				&rtm->rtm_rmx, &saved_nrt->rt_rmx);
348 			saved_nrt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits);
349 			saved_nrt->rt_rmx.rmx_locks |=
350 				(rtm->rtm_inits & rtm->rtm_rmx.rmx_locks);
351 			saved_nrt->rt_refcnt--;
352 			saved_nrt->rt_genmask = genmask;
353 		}
354 		break;
355 
356 	case RTM_DELETE:
357 		error = rtrequest1(RTM_DELETE, &info, &saved_nrt);
358 		if (error == 0) {
359 			rt = saved_nrt;
360 			goto report;
361 		}
362 		break;
363 
364 	case RTM_GET:
365 	case RTM_CHANGE:
366 	case RTM_LOCK:
367 		if ((rnh = rt_tables[dst->sa_family]) == 0) {
368 			senderr(EAFNOSUPPORT);
369 		}
370 		RADIX_NODE_HEAD_LOCK(rnh);
371 		rt = (struct rtentry *) rnh->rnh_lookup(dst, netmask, rnh);
372 		RADIX_NODE_HEAD_UNLOCK(rnh);
373 		if (rt != NULL)
374 			rt->rt_refcnt++;
375 		else
376 			senderr(ESRCH);
377 
378 		switch(rtm->rtm_type) {
379 
380 		case RTM_GET:
381 		report:
382 			dst = rt_key(rt);
383 			gate = rt->rt_gateway;
384 			netmask = rt_mask(rt);
385 			genmask = rt->rt_genmask;
386 			if (rtm->rtm_addrs & (RTA_IFP | RTA_IFA)) {
387 				ifp = rt->rt_ifp;
388 				if (ifp) {
389 					ifpaddr = TAILQ_FIRST(&ifp->if_addrhead)->ifa_addr;
390 					ifaaddr = rt->rt_ifa->ifa_addr;
391 					if (ifp->if_flags & IFF_POINTOPOINT)
392 						brdaddr = rt->rt_ifa->ifa_dstaddr;
393 					rtm->rtm_index = ifp->if_index;
394 				} else {
395 					ifpaddr = 0;
396 					ifaaddr = 0;
397 				}
398 			}
399 			len = rt_msg2(rtm->rtm_type, &info, (caddr_t)0,
400 				(struct walkarg *)0);
401 			if (len > rtm->rtm_msglen) {
402 				struct rt_msghdr *new_rtm;
403 				R_Malloc(new_rtm, struct rt_msghdr *, len);
404 				if (new_rtm == 0)
405 					senderr(ENOBUFS);
406 				Bcopy(rtm, new_rtm, rtm->rtm_msglen);
407 				Free(rtm); rtm = new_rtm;
408 			}
409 			(void)rt_msg2(rtm->rtm_type, &info, (caddr_t)rtm,
410 				(struct walkarg *)0);
411 			rtm->rtm_flags = rt->rt_flags;
412 			rtm->rtm_rmx = rt->rt_rmx;
413 			rtm->rtm_addrs = info.rti_addrs;
414 			break;
415 
416 		case RTM_CHANGE:
417 			/* new gateway could require new ifaddr, ifp;
418 			   flags may also be different; ifp may be specified
419 			   by ll sockaddr when protocol address is ambiguous */
420 /* compare two sockaddr structures */
421 #define	sa_equal(a1, a2) (bcmp((a1), (a2), (a1)->sa_len) == 0)
422 			if ((rt->rt_flags & RTF_GATEWAY && gate != NULL) ||
423 			    ifpaddr != NULL ||
424 			    (ifaaddr != NULL &&
425 			    !sa_equal(ifaaddr, rt->rt_ifa->ifa_addr))) {
426 				if ((error = rt_getifa(&info)) != 0)
427 					senderr(error);
428 			}
429 			if (gate != NULL &&
430 			    (error = rt_setgate(rt, rt_key(rt), gate)) != 0)
431 				senderr(error);
432 			if ((ifa = info.rti_ifa) != NULL) {
433 				register struct ifaddr *oifa = rt->rt_ifa;
434 				if (oifa != ifa) {
435 				    if (oifa) {
436 					if (oifa->ifa_rtrequest)
437 					    oifa->ifa_rtrequest(RTM_DELETE, rt,
438 						&info);
439 				        IFAFREE(oifa);
440 				    }
441 				    IFAREF(ifa);
442 				    rt->rt_ifa = ifa;
443 				    rt->rt_ifp = info.rti_ifp;
444 				}
445 			}
446 			rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx,
447 					&rt->rt_rmx);
448 			if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest)
449 			       rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, &info);
450 			if (genmask)
451 				rt->rt_genmask = genmask;
452 			/* FALLTHROUGH */
453 		case RTM_LOCK:
454 			rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits);
455 			rt->rt_rmx.rmx_locks |=
456 				(rtm->rtm_inits & rtm->rtm_rmx.rmx_locks);
457 			break;
458 		}
459 		break;
460 
461 	default:
462 		senderr(EOPNOTSUPP);
463 	}
464 
465 flush:
466 	if (rtm) {
467 		if (error)
468 			rtm->rtm_errno = error;
469 		else
470 			rtm->rtm_flags |= RTF_DONE;
471 	}
472 	if (rt)
473 		rtfree(rt);
474     {
475 	register struct rawcb *rp = 0;
476 	/*
477 	 * Check to see if we don't want our own messages.
478 	 */
479 	if ((so->so_options & SO_USELOOPBACK) == 0) {
480 		if (route_cb.any_count <= 1) {
481 			if (rtm)
482 				Free(rtm);
483 			m_freem(m);
484 			return (error);
485 		}
486 		/* There is another listener, so construct message */
487 		rp = sotorawcb(so);
488 	}
489 	if (rtm) {
490 		m_copyback(m, 0, rtm->rtm_msglen, (caddr_t)rtm);
491 		if (m->m_pkthdr.len < rtm->rtm_msglen) {
492 			m_freem(m);
493 			m = NULL;
494 		} else if (m->m_pkthdr.len > rtm->rtm_msglen)
495 			m_adj(m, rtm->rtm_msglen - m->m_pkthdr.len);
496 		Free(rtm);
497 	}
498 	if (rp)
499 		rp->rcb_proto.sp_family = 0; /* Avoid us */
500 	if (dst)
501 		route_proto.sp_protocol = dst->sa_family;
502 	if (m)
503 		raw_input(m, &route_proto, &route_src, &route_dst);
504 	if (rp)
505 		rp->rcb_proto.sp_family = PF_ROUTE;
506     }
507 	return (error);
508 }
509 
510 static void
511 rt_setmetrics(which, in, out)
512 	u_long which;
513 	register struct rt_metrics *in, *out;
514 {
515 #define metric(f, e) if (which & (f)) out->e = in->e;
516 	metric(RTV_RPIPE, rmx_recvpipe);
517 	metric(RTV_SPIPE, rmx_sendpipe);
518 	metric(RTV_SSTHRESH, rmx_ssthresh);
519 	metric(RTV_RTT, rmx_rtt);
520 	metric(RTV_RTTVAR, rmx_rttvar);
521 	metric(RTV_HOPCOUNT, rmx_hopcount);
522 	metric(RTV_MTU, rmx_mtu);
523 	metric(RTV_EXPIRE, rmx_expire);
524 #undef metric
525 }
526 
527 #define ROUNDUP(a) \
528 	((a) > 0 ? (1 + (((a) - 1) | (sizeof(long) - 1))) : sizeof(long))
529 #define ADVANCE(x, n) (x += ROUNDUP((n)->sa_len))
530 
531 
532 /*
533  * Extract the addresses of the passed sockaddrs.
534  * Do a little sanity checking so as to avoid bad memory references.
535  * This data is derived straight from userland.
536  */
537 static int
538 rt_xaddrs(cp, cplim, rtinfo)
539 	register caddr_t cp, cplim;
540 	register struct rt_addrinfo *rtinfo;
541 {
542 	register struct sockaddr *sa;
543 	register int i;
544 
545 	for (i = 0; (i < RTAX_MAX) && (cp < cplim); i++) {
546 		if ((rtinfo->rti_addrs & (1 << i)) == 0)
547 			continue;
548 		sa = (struct sockaddr *)cp;
549 		/*
550 		 * It won't fit.
551 		 */
552 		if ( (cp + sa->sa_len) > cplim ) {
553 			return (EINVAL);
554 		}
555 
556 		/*
557 		 * there are no more.. quit now
558 		 * If there are more bits, they are in error.
559 		 * I've seen this. route(1) can evidently generate these.
560 		 * This causes kernel to core dump.
561 		 * for compatibility, If we see this, point to a safe address.
562 		 */
563 		if (sa->sa_len == 0) {
564 			rtinfo->rti_info[i] = &sa_zero;
565 			return (0); /* should be EINVAL but for compat */
566 		}
567 
568 		/* accept it */
569 		rtinfo->rti_info[i] = sa;
570 		ADVANCE(cp, sa);
571 	}
572 	return (0);
573 }
574 
575 static struct mbuf *
576 rt_msg1(type, rtinfo)
577 	int type;
578 	register struct rt_addrinfo *rtinfo;
579 {
580 	register struct rt_msghdr *rtm;
581 	register struct mbuf *m;
582 	register int i;
583 	register struct sockaddr *sa;
584 	int len, dlen;
585 
586 	switch (type) {
587 
588 	case RTM_DELADDR:
589 	case RTM_NEWADDR:
590 		len = sizeof(struct ifa_msghdr);
591 		break;
592 
593 	case RTM_DELMADDR:
594 	case RTM_NEWMADDR:
595 		len = sizeof(struct ifma_msghdr);
596 		break;
597 
598 	case RTM_IFINFO:
599 		len = sizeof(struct if_msghdr);
600 		break;
601 
602 	case RTM_IFANNOUNCE:
603 		len = sizeof(struct if_announcemsghdr);
604 		break;
605 
606 	default:
607 		len = sizeof(struct rt_msghdr);
608 	}
609 	if (len > MCLBYTES)
610 		panic("rt_msg1");
611 	m = m_gethdr(M_DONTWAIT, MT_DATA);
612 	if (m && len > MHLEN) {
613 		MCLGET(m, M_DONTWAIT);
614 		if ((m->m_flags & M_EXT) == 0) {
615 			m_free(m);
616 			m = NULL;
617 		}
618 	}
619 	if (m == 0)
620 		return (m);
621 	m->m_pkthdr.len = m->m_len = len;
622 	m->m_pkthdr.rcvif = 0;
623 	rtm = mtod(m, struct rt_msghdr *);
624 	bzero((caddr_t)rtm, len);
625 	for (i = 0; i < RTAX_MAX; i++) {
626 		if ((sa = rtinfo->rti_info[i]) == NULL)
627 			continue;
628 		rtinfo->rti_addrs |= (1 << i);
629 		dlen = ROUNDUP(sa->sa_len);
630 		m_copyback(m, len, dlen, (caddr_t)sa);
631 		len += dlen;
632 	}
633 	if (m->m_pkthdr.len != len) {
634 		m_freem(m);
635 		return (NULL);
636 	}
637 	rtm->rtm_msglen = len;
638 	rtm->rtm_version = RTM_VERSION;
639 	rtm->rtm_type = type;
640 	return (m);
641 }
642 
643 static int
644 rt_msg2(type, rtinfo, cp, w)
645 	int type;
646 	register struct rt_addrinfo *rtinfo;
647 	caddr_t cp;
648 	struct walkarg *w;
649 {
650 	register int i;
651 	int len, dlen, second_time = 0;
652 	caddr_t cp0;
653 
654 	rtinfo->rti_addrs = 0;
655 again:
656 	switch (type) {
657 
658 	case RTM_DELADDR:
659 	case RTM_NEWADDR:
660 		len = sizeof(struct ifa_msghdr);
661 		break;
662 
663 	case RTM_IFINFO:
664 		len = sizeof(struct if_msghdr);
665 		break;
666 
667 	default:
668 		len = sizeof(struct rt_msghdr);
669 	}
670 	cp0 = cp;
671 	if (cp0)
672 		cp += len;
673 	for (i = 0; i < RTAX_MAX; i++) {
674 		register struct sockaddr *sa;
675 
676 		if ((sa = rtinfo->rti_info[i]) == 0)
677 			continue;
678 		rtinfo->rti_addrs |= (1 << i);
679 		dlen = ROUNDUP(sa->sa_len);
680 		if (cp) {
681 			bcopy((caddr_t)sa, cp, (unsigned)dlen);
682 			cp += dlen;
683 		}
684 		len += dlen;
685 	}
686 	len = ALIGN(len);
687 	if (cp == 0 && w != NULL && !second_time) {
688 		register struct walkarg *rw = w;
689 
690 		if (rw->w_req) {
691 			if (rw->w_tmemsize < len) {
692 				if (rw->w_tmem)
693 					free(rw->w_tmem, M_RTABLE);
694 				rw->w_tmem = (caddr_t)
695 					malloc(len, M_RTABLE, M_NOWAIT);
696 				if (rw->w_tmem)
697 					rw->w_tmemsize = len;
698 			}
699 			if (rw->w_tmem) {
700 				cp = rw->w_tmem;
701 				second_time = 1;
702 				goto again;
703 			}
704 		}
705 	}
706 	if (cp) {
707 		register struct rt_msghdr *rtm = (struct rt_msghdr *)cp0;
708 
709 		rtm->rtm_version = RTM_VERSION;
710 		rtm->rtm_type = type;
711 		rtm->rtm_msglen = len;
712 	}
713 	return (len);
714 }
715 
716 /*
717  * This routine is called to generate a message from the routing
718  * socket indicating that a redirect has occured, a routing lookup
719  * has failed, or that a protocol has detected timeouts to a particular
720  * destination.
721  */
722 void
723 rt_missmsg(type, rtinfo, flags, error)
724 	int type, flags, error;
725 	register struct rt_addrinfo *rtinfo;
726 {
727 	register struct rt_msghdr *rtm;
728 	register struct mbuf *m;
729 	struct sockaddr *sa = rtinfo->rti_info[RTAX_DST];
730 
731 	if (route_cb.any_count == 0)
732 		return;
733 	m = rt_msg1(type, rtinfo);
734 	if (m == 0)
735 		return;
736 	rtm = mtod(m, struct rt_msghdr *);
737 	rtm->rtm_flags = RTF_DONE | flags;
738 	rtm->rtm_errno = error;
739 	rtm->rtm_addrs = rtinfo->rti_addrs;
740 	route_proto.sp_protocol = sa ? sa->sa_family : 0;
741 	raw_input(m, &route_proto, &route_src, &route_dst);
742 }
743 
744 /*
745  * This routine is called to generate a message from the routing
746  * socket indicating that the status of a network interface has changed.
747  */
748 void
749 rt_ifmsg(ifp)
750 	register struct ifnet *ifp;
751 {
752 	register struct if_msghdr *ifm;
753 	struct mbuf *m;
754 	struct rt_addrinfo info;
755 
756 	if (route_cb.any_count == 0)
757 		return;
758 	bzero((caddr_t)&info, sizeof(info));
759 	m = rt_msg1(RTM_IFINFO, &info);
760 	if (m == 0)
761 		return;
762 	ifm = mtod(m, struct if_msghdr *);
763 	ifm->ifm_index = ifp->if_index;
764 	ifm->ifm_flags = ifp->if_flags;
765 	ifm->ifm_data = ifp->if_data;
766 	ifm->ifm_addrs = 0;
767 	route_proto.sp_protocol = 0;
768 	raw_input(m, &route_proto, &route_src, &route_dst);
769 }
770 
771 /*
772  * This is called to generate messages from the routing socket
773  * indicating a network interface has had addresses associated with it.
774  * if we ever reverse the logic and replace messages TO the routing
775  * socket indicate a request to configure interfaces, then it will
776  * be unnecessary as the routing socket will automatically generate
777  * copies of it.
778  */
779 void
780 rt_newaddrmsg(cmd, ifa, error, rt)
781 	int cmd, error;
782 	register struct ifaddr *ifa;
783 	register struct rtentry *rt;
784 {
785 	struct rt_addrinfo info;
786 	struct sockaddr *sa = 0;
787 	int pass;
788 	struct mbuf *m = 0;
789 	struct ifnet *ifp = ifa->ifa_ifp;
790 
791 	if (route_cb.any_count == 0)
792 		return;
793 	for (pass = 1; pass < 3; pass++) {
794 		bzero((caddr_t)&info, sizeof(info));
795 		if ((cmd == RTM_ADD && pass == 1) ||
796 		    (cmd == RTM_DELETE && pass == 2)) {
797 			register struct ifa_msghdr *ifam;
798 			int ncmd = cmd == RTM_ADD ? RTM_NEWADDR : RTM_DELADDR;
799 
800 			ifaaddr = sa = ifa->ifa_addr;
801 			ifpaddr = TAILQ_FIRST(&ifp->if_addrhead)->ifa_addr;
802 			netmask = ifa->ifa_netmask;
803 			brdaddr = ifa->ifa_dstaddr;
804 			if ((m = rt_msg1(ncmd, &info)) == NULL)
805 				continue;
806 			ifam = mtod(m, struct ifa_msghdr *);
807 			ifam->ifam_index = ifp->if_index;
808 			ifam->ifam_metric = ifa->ifa_metric;
809 			ifam->ifam_flags = ifa->ifa_flags;
810 			ifam->ifam_addrs = info.rti_addrs;
811 		}
812 		if ((cmd == RTM_ADD && pass == 2) ||
813 		    (cmd == RTM_DELETE && pass == 1)) {
814 			register struct rt_msghdr *rtm;
815 
816 			if (rt == 0)
817 				continue;
818 			netmask = rt_mask(rt);
819 			dst = sa = rt_key(rt);
820 			gate = rt->rt_gateway;
821 			if ((m = rt_msg1(cmd, &info)) == NULL)
822 				continue;
823 			rtm = mtod(m, struct rt_msghdr *);
824 			rtm->rtm_index = ifp->if_index;
825 			rtm->rtm_flags |= rt->rt_flags;
826 			rtm->rtm_errno = error;
827 			rtm->rtm_addrs = info.rti_addrs;
828 		}
829 		route_proto.sp_protocol = sa ? sa->sa_family : 0;
830 		raw_input(m, &route_proto, &route_src, &route_dst);
831 	}
832 }
833 
834 /*
835  * This is the analogue to the rt_newaddrmsg which performs the same
836  * function but for multicast group memberhips.  This is easier since
837  * there is no route state to worry about.
838  */
839 void
840 rt_newmaddrmsg(cmd, ifma)
841 	int cmd;
842 	struct ifmultiaddr *ifma;
843 {
844 	struct rt_addrinfo info;
845 	struct mbuf *m = 0;
846 	struct ifnet *ifp = ifma->ifma_ifp;
847 	struct ifma_msghdr *ifmam;
848 
849 	if (route_cb.any_count == 0)
850 		return;
851 
852 	bzero((caddr_t)&info, sizeof(info));
853 	ifaaddr = ifma->ifma_addr;
854 	if (ifp && TAILQ_FIRST(&ifp->if_addrhead))
855 		ifpaddr = TAILQ_FIRST(&ifp->if_addrhead)->ifa_addr;
856 	else
857 		ifpaddr = NULL;
858 	/*
859 	 * If a link-layer address is present, present it as a ``gateway''
860 	 * (similarly to how ARP entries, e.g., are presented).
861 	 */
862 	gate = ifma->ifma_lladdr;
863 	if ((m = rt_msg1(cmd, &info)) == NULL)
864 		return;
865 	ifmam = mtod(m, struct ifma_msghdr *);
866 	ifmam->ifmam_index = ifp->if_index;
867 	ifmam->ifmam_addrs = info.rti_addrs;
868 	route_proto.sp_protocol = ifma->ifma_addr->sa_family;
869 	raw_input(m, &route_proto, &route_src, &route_dst);
870 }
871 
872 /*
873  * This is called to generate routing socket messages indicating
874  * network interface arrival and departure.
875  */
876 void
877 rt_ifannouncemsg(ifp, what)
878 	struct ifnet *ifp;
879 	int what;
880 {
881 	struct if_announcemsghdr *ifan;
882 	struct mbuf *m;
883 	struct rt_addrinfo info;
884 
885 	if (route_cb.any_count == 0)
886 		return;
887 	bzero((caddr_t)&info, sizeof(info));
888 	m = rt_msg1(RTM_IFANNOUNCE, &info);
889 	if (m == NULL)
890 		return;
891 	ifan = mtod(m, struct if_announcemsghdr *);
892 	ifan->ifan_index = ifp->if_index;
893 	snprintf(ifan->ifan_name, sizeof(ifan->ifan_name),
894 	    "%s%d", ifp->if_name, ifp->if_unit);
895 	ifan->ifan_what = what;
896 	route_proto.sp_protocol = 0;
897 	raw_input(m, &route_proto, &route_src, &route_dst);
898  }
899 
900 /*
901  * This is used in dumping the kernel table via sysctl().
902  */
903 static int
904 sysctl_dumpentry(rn, vw)
905 	struct radix_node *rn;
906 	void *vw;
907 {
908 	register struct walkarg *w = vw;
909 	register struct rtentry *rt = (struct rtentry *)rn;
910 	int error = 0, size;
911 	struct rt_addrinfo info;
912 
913 	if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg))
914 		return 0;
915 	bzero((caddr_t)&info, sizeof(info));
916 	dst = rt_key(rt);
917 	gate = rt->rt_gateway;
918 	netmask = rt_mask(rt);
919 	genmask = rt->rt_genmask;
920 	if (rt->rt_ifp) {
921 		ifpaddr = TAILQ_FIRST(&rt->rt_ifp->if_addrhead)->ifa_addr;
922 		ifaaddr = rt->rt_ifa->ifa_addr;
923 		if (rt->rt_ifp->if_flags & IFF_POINTOPOINT)
924 			brdaddr = rt->rt_ifa->ifa_dstaddr;
925 	}
926 	size = rt_msg2(RTM_GET, &info, 0, w);
927 	if (w->w_req && w->w_tmem) {
928 		register struct rt_msghdr *rtm = (struct rt_msghdr *)w->w_tmem;
929 
930 		rtm->rtm_flags = rt->rt_flags;
931 		rtm->rtm_use = rt->rt_use;
932 		rtm->rtm_rmx = rt->rt_rmx;
933 		rtm->rtm_index = rt->rt_ifp->if_index;
934 		rtm->rtm_errno = rtm->rtm_pid = rtm->rtm_seq = 0;
935 		rtm->rtm_addrs = info.rti_addrs;
936 		error = SYSCTL_OUT(w->w_req, (caddr_t)rtm, size);
937 		return (error);
938 	}
939 	return (error);
940 }
941 
942 static int
943 sysctl_iflist(af, w)
944 	int	af;
945 	register struct	walkarg *w;
946 {
947 	register struct ifnet *ifp;
948 	register struct ifaddr *ifa;
949 	struct	rt_addrinfo info;
950 	int	len, error = 0;
951 
952 	bzero((caddr_t)&info, sizeof(info));
953 	/* IFNET_RLOCK(); */		/* could sleep XXX */
954 	TAILQ_FOREACH(ifp, &ifnet, if_link) {
955 		if (w->w_arg && w->w_arg != ifp->if_index)
956 			continue;
957 		ifa = TAILQ_FIRST(&ifp->if_addrhead);
958 		ifpaddr = ifa->ifa_addr;
959 		len = rt_msg2(RTM_IFINFO, &info, (caddr_t)0, w);
960 		ifpaddr = 0;
961 		if (w->w_req && w->w_tmem) {
962 			register struct if_msghdr *ifm;
963 
964 			ifm = (struct if_msghdr *)w->w_tmem;
965 			ifm->ifm_index = ifp->if_index;
966 			ifm->ifm_flags = ifp->if_flags;
967 			ifm->ifm_data = ifp->if_data;
968 			ifm->ifm_addrs = info.rti_addrs;
969 			error = SYSCTL_OUT(w->w_req,(caddr_t)ifm, len);
970 			if (error)
971 				goto done;
972 		}
973 		while ((ifa = TAILQ_NEXT(ifa, ifa_link)) != 0) {
974 			if (af && af != ifa->ifa_addr->sa_family)
975 				continue;
976 			if (jailed(curthread->td_ucred) &&
977 			    prison_if(curthread->td_ucred, ifa->ifa_addr))
978 				continue;
979 			ifaaddr = ifa->ifa_addr;
980 			netmask = ifa->ifa_netmask;
981 			brdaddr = ifa->ifa_dstaddr;
982 			len = rt_msg2(RTM_NEWADDR, &info, 0, w);
983 			if (w->w_req && w->w_tmem) {
984 				register struct ifa_msghdr *ifam;
985 
986 				ifam = (struct ifa_msghdr *)w->w_tmem;
987 				ifam->ifam_index = ifa->ifa_ifp->if_index;
988 				ifam->ifam_flags = ifa->ifa_flags;
989 				ifam->ifam_metric = ifa->ifa_metric;
990 				ifam->ifam_addrs = info.rti_addrs;
991 				error = SYSCTL_OUT(w->w_req, w->w_tmem, len);
992 				if (error)
993 					goto done;
994 			}
995 		}
996 		ifaaddr = netmask = brdaddr = 0;
997 	}
998 done:
999 	/* IFNET_RUNLOCK(); */ /* XXX */
1000 	return (error);
1001 }
1002 
1003 static int
1004 sysctl_rtsock(SYSCTL_HANDLER_ARGS)
1005 {
1006 	int	*name = (int *)arg1;
1007 	u_int	namelen = arg2;
1008 	register struct radix_node_head *rnh;
1009 	int	i, s, error = EINVAL;
1010 	u_char  af;
1011 	struct	walkarg w;
1012 
1013 	name ++;
1014 	namelen--;
1015 	if (req->newptr)
1016 		return (EPERM);
1017 	if (namelen != 3)
1018 		return ((namelen < 3) ? EISDIR : ENOTDIR);
1019 	af = name[0];
1020 	if (af > AF_MAX)
1021 		return (EINVAL);
1022 	Bzero(&w, sizeof(w));
1023 	w.w_op = name[1];
1024 	w.w_arg = name[2];
1025 	w.w_req = req;
1026 
1027 	s = splnet();
1028 	switch (w.w_op) {
1029 
1030 	case NET_RT_DUMP:
1031 	case NET_RT_FLAGS:
1032 		if (af != 0) {
1033 			if ((rnh = rt_tables[af]) != NULL) {
1034 				/* RADIX_NODE_HEAD_LOCK(rnh); */
1035 			    	error = rnh->rnh_walktree(rnh,
1036 				    sysctl_dumpentry, &w);/* could sleep XXX */
1037 				/* RADIX_NODE_HEAD_UNLOCK(rnh); */
1038 			} else
1039 				error = EAFNOSUPPORT;
1040 		} else {
1041 			for (i = 1; i <= AF_MAX; i++)
1042 				if ((rnh = rt_tables[i]) != NULL) {
1043 					/* RADIX_NODE_HEAD_LOCK(rnh); */
1044 					error = rnh->rnh_walktree(rnh,
1045 					    sysctl_dumpentry, &w);
1046 					/* RADIX_NODE_HEAD_UNLOCK(rnh); */
1047 					if (error)
1048 						break;
1049 				}
1050 		}
1051 		break;
1052 
1053 	case NET_RT_IFLIST:
1054 		error = sysctl_iflist(af, &w);
1055 	}
1056 	splx(s);
1057 	if (w.w_tmem)
1058 		free(w.w_tmem, M_RTABLE);
1059 	return (error);
1060 }
1061 
1062 SYSCTL_NODE(_net, PF_ROUTE, routetable, CTLFLAG_RD, sysctl_rtsock, "");
1063 
1064 /*
1065  * Definitions of protocols supported in the ROUTE domain.
1066  */
1067 
1068 extern struct domain routedomain;		/* or at least forward */
1069 
1070 static struct protosw routesw[] = {
1071 { SOCK_RAW,	&routedomain,	0,		PR_ATOMIC|PR_ADDR,
1072   0,		route_output,	raw_ctlinput,	0,
1073   0,
1074   raw_init,	0,		0,		0,
1075   &route_usrreqs
1076 }
1077 };
1078 
1079 static struct domain routedomain =
1080     { PF_ROUTE, "route", 0, 0, 0,
1081       routesw, &routesw[sizeof(routesw)/sizeof(routesw[0])] };
1082 
1083 DOMAIN_SET(route);
1084