xref: /dragonfly/sys/net/rtsock.c (revision af79c6e5)
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: src/sys/net/rtsock.c,v 1.44.2.11 2002/12/04 14:05:41 ru Exp $
35  * $DragonFly: src/sys/net/rtsock.c,v 1.7 2003/10/06 06:08:23 hsu Exp $
36  */
37 
38 
39 #include <sys/param.h>
40 #include <sys/systm.h>
41 #include <sys/kernel.h>
42 #include <sys/sysctl.h>
43 #include <sys/proc.h>
44 #include <sys/malloc.h>
45 #include <sys/mbuf.h>
46 #include <sys/socket.h>
47 #include <sys/socketvar.h>
48 #include <sys/domain.h>
49 #include <sys/protosw.h>
50 
51 #include <net/if.h>
52 #include <net/route.h>
53 #include <net/raw_cb.h>
54 
55 MALLOC_DEFINE(M_RTABLE, "routetbl", "routing tables");
56 
57 static struct	sockaddr route_dst = { 2, PF_ROUTE, };
58 static struct	sockaddr route_src = { 2, PF_ROUTE, };
59 static struct	sockaddr sa_zero   = { sizeof(sa_zero), AF_INET, };
60 static struct	sockproto route_proto = { PF_ROUTE, };
61 
62 struct walkarg {
63 	int	w_tmemsize;
64 	int	w_op, w_arg;
65 	caddr_t	w_tmem;
66 	struct sysctl_req *w_req;
67 };
68 
69 static struct mbuf *
70 		rt_msg1 (int, struct rt_addrinfo *);
71 static int	rt_msg2 (int,
72 		    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 	MALLOC(rp, struct rawcb *, sizeof *rp, M_PCB, M_WAITOK|M_ZERO);
113 	if (rp == 0)
114 		return ENOBUFS;
115 
116 	/*
117 	 * The splnet() is necessary to block protocols from sending
118 	 * error notifications (like RTM_REDIRECT or RTM_LOSING) while
119 	 * this PCB is extant but incompletely initialized.
120 	 * Probably we should try to do more of this work beforehand and
121 	 * eliminate the spl.
122 	 */
123 	s = splnet();
124 	so->so_pcb = (caddr_t)rp;
125 	error = raw_attach(so, proto);
126 	rp = sotorawcb(so);
127 	if (error) {
128 		splx(s);
129 		free(rp, M_PCB);
130 		return error;
131 	}
132 	switch(rp->rcb_proto.sp_protocol) {
133 	case AF_INET:
134 		route_cb.ip_count++;
135 		break;
136 	case AF_INET6:
137 		route_cb.ip6_count++;
138 		break;
139 	case AF_IPX:
140 		route_cb.ipx_count++;
141 		break;
142 	case AF_NS:
143 		route_cb.ns_count++;
144 		break;
145 	}
146 	rp->rcb_faddr = &route_src;
147 	route_cb.any_count++;
148 	soisconnected(so);
149 	so->so_options |= SO_USELOOPBACK;
150 	splx(s);
151 	return 0;
152 }
153 
154 static int
155 rts_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
156 {
157 	int s, error;
158 	s = splnet();
159 	error = raw_usrreqs.pru_bind(so, nam, td); /* xxx just EINVAL */
160 	splx(s);
161 	return error;
162 }
163 
164 static int
165 rts_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
166 {
167 	int s, error;
168 	s = splnet();
169 	error = raw_usrreqs.pru_connect(so, nam, td); /* XXX just EINVAL */
170 	splx(s);
171 	return error;
172 }
173 
174 /* pru_connect2 is EOPNOTSUPP */
175 /* pru_control is EOPNOTSUPP */
176 
177 static int
178 rts_detach(struct socket *so)
179 {
180 	struct rawcb *rp = sotorawcb(so);
181 	int s, error;
182 
183 	s = splnet();
184 	if (rp != 0) {
185 		switch(rp->rcb_proto.sp_protocol) {
186 		case AF_INET:
187 			route_cb.ip_count--;
188 			break;
189 		case AF_INET6:
190 			route_cb.ip6_count--;
191 			break;
192 		case AF_IPX:
193 			route_cb.ipx_count--;
194 			break;
195 		case AF_NS:
196 			route_cb.ns_count--;
197 			break;
198 		}
199 		route_cb.any_count--;
200 	}
201 	error = raw_usrreqs.pru_detach(so);
202 	splx(s);
203 	return error;
204 }
205 
206 static int
207 rts_disconnect(struct socket *so)
208 {
209 	int s, error;
210 	s = splnet();
211 	error = raw_usrreqs.pru_disconnect(so);
212 	splx(s);
213 	return error;
214 }
215 
216 /* pru_listen is EOPNOTSUPP */
217 
218 static int
219 rts_peeraddr(struct socket *so, struct sockaddr **nam)
220 {
221 	int s, error;
222 	s = splnet();
223 	error = raw_usrreqs.pru_peeraddr(so, nam);
224 	splx(s);
225 	return error;
226 }
227 
228 /* pru_rcvd is EOPNOTSUPP */
229 /* pru_rcvoob is EOPNOTSUPP */
230 
231 static int
232 rts_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
233 	 struct mbuf *control, struct thread *td)
234 {
235 	int s, error;
236 	s = splnet();
237 	error = raw_usrreqs.pru_send(so, flags, m, nam, control, td);
238 	splx(s);
239 	return error;
240 }
241 
242 /* pru_sense is null */
243 
244 static int
245 rts_shutdown(struct socket *so)
246 {
247 	int s, error;
248 	s = splnet();
249 	error = raw_usrreqs.pru_shutdown(so);
250 	splx(s);
251 	return error;
252 }
253 
254 static int
255 rts_sockaddr(struct socket *so, struct sockaddr **nam)
256 {
257 	int s, error;
258 	s = splnet();
259 	error = raw_usrreqs.pru_sockaddr(so, nam);
260 	splx(s);
261 	return error;
262 }
263 
264 static struct pr_usrreqs route_usrreqs = {
265 	rts_abort, pru_accept_notsupp, rts_attach, rts_bind, rts_connect,
266 	pru_connect2_notsupp, pru_control_notsupp, rts_detach, rts_disconnect,
267 	pru_listen_notsupp, rts_peeraddr, pru_rcvd_notsupp, pru_rcvoob_notsupp,
268 	rts_send, pru_sense_null, rts_shutdown, rts_sockaddr,
269 	sosend, soreceive, sopoll
270 };
271 
272 /*ARGSUSED*/
273 static int
274 route_output(m, so)
275 	struct mbuf *m;
276 	struct socket *so;
277 {
278 	struct rt_msghdr *rtm = 0;
279 	struct rtentry *rt = 0;
280 	struct rtentry *saved_nrt = 0;
281 	struct radix_node_head *rnh;
282 	struct rt_addrinfo info;
283 	int len, error = 0;
284 	struct ifnet *ifp = 0;
285 	struct ifaddr *ifa = 0;
286 
287 #define senderr(e) { error = e; goto flush;}
288 	if (m == 0 || ((m->m_len < sizeof(long)) &&
289 		       (m = m_pullup(m, sizeof(long))) == 0))
290 		return (ENOBUFS);
291 	if ((m->m_flags & M_PKTHDR) == 0)
292 		panic("route_output");
293 	len = m->m_pkthdr.len;
294 	if (len < sizeof(*rtm) ||
295 	    len != mtod(m, struct rt_msghdr *)->rtm_msglen) {
296 		dst = 0;
297 		senderr(EINVAL);
298 	}
299 	R_Malloc(rtm, struct rt_msghdr *, len);
300 	if (rtm == 0) {
301 		dst = 0;
302 		senderr(ENOBUFS);
303 	}
304 	m_copydata(m, 0, len, (caddr_t)rtm);
305 	if (rtm->rtm_version != RTM_VERSION) {
306 		dst = 0;
307 		senderr(EPROTONOSUPPORT);
308 	}
309 	rtm->rtm_pid = curproc->p_pid;
310 	bzero(&info, sizeof(info));
311 	info.rti_addrs = rtm->rtm_addrs;
312 	if (rt_xaddrs((caddr_t)(rtm + 1), len + (caddr_t)rtm, &info)) {
313 		dst = 0;
314 		senderr(EINVAL);
315 	}
316 	info.rti_flags = rtm->rtm_flags;
317 	if (dst == 0 || (dst->sa_family >= AF_MAX)
318 	    || (gate != 0 && (gate->sa_family >= AF_MAX)))
319 		senderr(EINVAL);
320 	if (genmask) {
321 		struct radix_node *t;
322 		t = rn_addmask((caddr_t)genmask, 0, 1);
323 		if (t && Bcmp((caddr_t *)genmask + 1, (caddr_t *)t->rn_key + 1,
324 			      *(u_char *)t->rn_key - 1) == 0)
325 			genmask = (struct sockaddr *)(t->rn_key);
326 		else
327 			senderr(ENOBUFS);
328 	}
329 
330 	/*
331 	 * Verify that the caller has the appropriate privilege; RTM_GET
332 	 * is the only operation the non-superuser is allowed.
333 	 */
334 	if (rtm->rtm_type != RTM_GET && suser_cred(so->so_cred, 0) != 0)
335 		senderr(EPERM);
336 
337 	switch (rtm->rtm_type) {
338 
339 	case RTM_ADD:
340 		if (gate == 0)
341 			senderr(EINVAL);
342 		error = rtrequest1(RTM_ADD, &info, &saved_nrt);
343 		if (error == 0 && saved_nrt) {
344 			rt_setmetrics(rtm->rtm_inits,
345 				&rtm->rtm_rmx, &saved_nrt->rt_rmx);
346 			saved_nrt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits);
347 			saved_nrt->rt_rmx.rmx_locks |=
348 				(rtm->rtm_inits & rtm->rtm_rmx.rmx_locks);
349 			saved_nrt->rt_refcnt--;
350 			saved_nrt->rt_genmask = genmask;
351 		}
352 		break;
353 
354 	case RTM_DELETE:
355 		error = rtrequest1(RTM_DELETE, &info, &saved_nrt);
356 		if (error == 0) {
357 			if ((rt = saved_nrt))
358 				rt->rt_refcnt++;
359 			goto report;
360 		}
361 		break;
362 
363 	case RTM_GET:
364 	case RTM_CHANGE:
365 	case RTM_LOCK:
366 		if ((rnh = rt_tables[dst->sa_family]) == 0) {
367 			senderr(EAFNOSUPPORT);
368 		} else if ((rt = (struct rtentry *)
369 				rnh->rnh_lookup(dst, netmask, rnh)) != NULL)
370 			rt->rt_refcnt++;
371 		else
372 			senderr(ESRCH);
373 		switch(rtm->rtm_type) {
374 
375 		case RTM_GET:
376 		report:
377 			dst = rt_key(rt);
378 			gate = rt->rt_gateway;
379 			netmask = rt_mask(rt);
380 			genmask = rt->rt_genmask;
381 			if (rtm->rtm_addrs & (RTA_IFP | RTA_IFA)) {
382 				ifp = rt->rt_ifp;
383 				if (ifp) {
384 					ifpaddr = TAILQ_FIRST(&ifp->if_addrhead)->ifa_addr;
385 					ifaaddr = rt->rt_ifa->ifa_addr;
386 					if (ifp->if_flags & IFF_POINTOPOINT)
387 						brdaddr = rt->rt_ifa->ifa_dstaddr;
388 					rtm->rtm_index = ifp->if_index;
389 				} else {
390 					ifpaddr = 0;
391 					ifaaddr = 0;
392 			    }
393 			}
394 			len = rt_msg2(rtm->rtm_type, &info, (caddr_t)0,
395 				(struct walkarg *)0);
396 			if (len > rtm->rtm_msglen) {
397 				struct rt_msghdr *new_rtm;
398 				R_Malloc(new_rtm, struct rt_msghdr *, len);
399 				if (new_rtm == 0)
400 					senderr(ENOBUFS);
401 				Bcopy(rtm, new_rtm, rtm->rtm_msglen);
402 				Free(rtm); rtm = new_rtm;
403 			}
404 			(void)rt_msg2(rtm->rtm_type, &info, (caddr_t)rtm,
405 				(struct walkarg *)0);
406 			rtm->rtm_flags = rt->rt_flags;
407 			rtm->rtm_rmx = rt->rt_rmx;
408 			rtm->rtm_addrs = info.rti_addrs;
409 			break;
410 
411 		case RTM_CHANGE:
412 			/* new gateway could require new ifaddr, ifp;
413 			   flags may also be different; ifp may be specified
414 			   by ll sockaddr when protocol address is ambiguous */
415 #define	equal(a1, a2) (bcmp((caddr_t)(a1), (caddr_t)(a2), (a1)->sa_len) == 0)
416 			if ((rt->rt_flags & RTF_GATEWAY && gate != NULL) ||
417 			    ifpaddr != NULL ||
418 			    (ifaaddr != NULL &&
419 			    !equal(ifaaddr, rt->rt_ifa->ifa_addr))) {
420 				if ((error = rt_getifa(&info)) != 0)
421 					senderr(error);
422 			}
423 			if (gate != NULL &&
424 			    (error = rt_setgate(rt, rt_key(rt), gate)) != 0)
425 				senderr(error);
426 			if ((ifa = info.rti_ifa) != NULL) {
427 				struct ifaddr *oifa = rt->rt_ifa;
428 				if (oifa != ifa) {
429 				    if (oifa && oifa->ifa_rtrequest)
430 					oifa->ifa_rtrequest(RTM_DELETE, rt,
431 					    &info);
432 				    IFAFREE(rt->rt_ifa);
433 				    rt->rt_ifa = ifa;
434 				    ifa->ifa_refcnt++;
435 				    rt->rt_ifp = info.rti_ifp;
436 				}
437 			}
438 			rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx,
439 					&rt->rt_rmx);
440 			if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest)
441 			       rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, &info);
442 			if (genmask)
443 				rt->rt_genmask = genmask;
444 			/*
445 			 * Fall into
446 			 */
447 		case RTM_LOCK:
448 			rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits);
449 			rt->rt_rmx.rmx_locks |=
450 				(rtm->rtm_inits & rtm->rtm_rmx.rmx_locks);
451 			break;
452 		}
453 		break;
454 
455 	default:
456 		senderr(EOPNOTSUPP);
457 	}
458 
459 flush:
460 	if (rtm) {
461 		if (error)
462 			rtm->rtm_errno = error;
463 		else
464 			rtm->rtm_flags |= RTF_DONE;
465 	}
466 	if (rt)
467 		rtfree(rt);
468     {
469 	struct rawcb *rp = 0;
470 	/*
471 	 * Check to see if we don't want our own messages.
472 	 */
473 	if ((so->so_options & SO_USELOOPBACK) == 0) {
474 		if (route_cb.any_count <= 1) {
475 			if (rtm)
476 				Free(rtm);
477 			m_freem(m);
478 			return (error);
479 		}
480 		/* There is another listener, so construct message */
481 		rp = sotorawcb(so);
482 	}
483 	if (rtm) {
484 		m_copyback(m, 0, rtm->rtm_msglen, (caddr_t)rtm);
485 		if (m->m_pkthdr.len < rtm->rtm_msglen) {
486 			m_freem(m);
487 			m = NULL;
488 		} else if (m->m_pkthdr.len > rtm->rtm_msglen)
489 			m_adj(m, rtm->rtm_msglen - m->m_pkthdr.len);
490 		Free(rtm);
491 	}
492 	if (rp)
493 		rp->rcb_proto.sp_family = 0; /* Avoid us */
494 	if (dst)
495 		route_proto.sp_protocol = dst->sa_family;
496 	if (m)
497 		raw_input(m, &route_proto, &route_src, &route_dst);
498 	if (rp)
499 		rp->rcb_proto.sp_family = PF_ROUTE;
500     }
501 	return (error);
502 }
503 
504 static void
505 rt_setmetrics(which, in, out)
506 	u_long which;
507 	struct rt_metrics *in, *out;
508 {
509 #define metric(f, e) if (which & (f)) out->e = in->e;
510 	metric(RTV_RPIPE, rmx_recvpipe);
511 	metric(RTV_SPIPE, rmx_sendpipe);
512 	metric(RTV_SSTHRESH, rmx_ssthresh);
513 	metric(RTV_RTT, rmx_rtt);
514 	metric(RTV_RTTVAR, rmx_rttvar);
515 	metric(RTV_HOPCOUNT, rmx_hopcount);
516 	metric(RTV_MTU, rmx_mtu);
517 	metric(RTV_EXPIRE, rmx_expire);
518 #undef metric
519 }
520 
521 #define ROUNDUP(a) \
522 	((a) > 0 ? (1 + (((a) - 1) | (sizeof(long) - 1))) : sizeof(long))
523 #define ADVANCE(x, n) (x += ROUNDUP((n)->sa_len))
524 
525 
526 /*
527  * Extract the addresses of the passed sockaddrs.
528  * Do a little sanity checking so as to avoid bad memory references.
529  * This data is derived straight from userland.
530  */
531 static int
532 rt_xaddrs(cp, cplim, rtinfo)
533 	caddr_t cp, cplim;
534 	struct rt_addrinfo *rtinfo;
535 {
536 	struct sockaddr *sa;
537 	int i;
538 
539 	for (i = 0; (i < RTAX_MAX) && (cp < cplim); i++) {
540 		if ((rtinfo->rti_addrs & (1 << i)) == 0)
541 			continue;
542 		sa = (struct sockaddr *)cp;
543 		/*
544 		 * It won't fit.
545 		 */
546 		if ( (cp + sa->sa_len) > cplim ) {
547 			return (EINVAL);
548 		}
549 
550 		/*
551 		 * there are no more.. quit now
552 		 * If there are more bits, they are in error.
553 		 * I've seen this. route(1) can evidently generate these.
554 		 * This causes kernel to core dump.
555 		 * for compatibility, If we see this, point to a safe address.
556 		 */
557 		if (sa->sa_len == 0) {
558 			rtinfo->rti_info[i] = &sa_zero;
559 			return (0); /* should be EINVAL but for compat */
560 		}
561 
562 		/* accept it */
563 		rtinfo->rti_info[i] = sa;
564 		ADVANCE(cp, sa);
565 	}
566 	return (0);
567 }
568 
569 static struct mbuf *
570 rt_msg1(type, rtinfo)
571 	int type;
572 	struct rt_addrinfo *rtinfo;
573 {
574 	struct rt_msghdr *rtm;
575 	struct mbuf *m;
576 	int i;
577 	struct sockaddr *sa;
578 	int len, dlen;
579 
580 	switch (type) {
581 
582 	case RTM_DELADDR:
583 	case RTM_NEWADDR:
584 		len = sizeof(struct ifa_msghdr);
585 		break;
586 
587 	case RTM_DELMADDR:
588 	case RTM_NEWMADDR:
589 		len = sizeof(struct ifma_msghdr);
590 		break;
591 
592 	case RTM_IFINFO:
593 		len = sizeof(struct if_msghdr);
594 		break;
595 
596 	case RTM_IFANNOUNCE:
597 		len = sizeof(struct if_announcemsghdr);
598 		break;
599 
600 	default:
601 		len = sizeof(struct rt_msghdr);
602 	}
603 	if (len > MCLBYTES)
604 		panic("rt_msg1");
605 	m = m_gethdr(M_DONTWAIT, MT_DATA);
606 	if (m && len > MHLEN) {
607 		MCLGET(m, M_DONTWAIT);
608 		if ((m->m_flags & M_EXT) == 0) {
609 			m_free(m);
610 			m = NULL;
611 		}
612 	}
613 	if (m == 0)
614 		return (m);
615 	m->m_pkthdr.len = m->m_len = len;
616 	m->m_pkthdr.rcvif = 0;
617 	rtm = mtod(m, struct rt_msghdr *);
618 	bzero((caddr_t)rtm, len);
619 	for (i = 0; i < RTAX_MAX; i++) {
620 		if ((sa = rtinfo->rti_info[i]) == NULL)
621 			continue;
622 		rtinfo->rti_addrs |= (1 << i);
623 		dlen = ROUNDUP(sa->sa_len);
624 		m_copyback(m, len, dlen, (caddr_t)sa);
625 		len += dlen;
626 	}
627 	if (m->m_pkthdr.len != len) {
628 		m_freem(m);
629 		return (NULL);
630 	}
631 	rtm->rtm_msglen = len;
632 	rtm->rtm_version = RTM_VERSION;
633 	rtm->rtm_type = type;
634 	return (m);
635 }
636 
637 static int
638 rt_msg2(type, rtinfo, cp, w)
639 	int type;
640 	struct rt_addrinfo *rtinfo;
641 	caddr_t cp;
642 	struct walkarg *w;
643 {
644 	int i;
645 	int len, dlen, second_time = 0;
646 	caddr_t cp0;
647 
648 	rtinfo->rti_addrs = 0;
649 again:
650 	switch (type) {
651 
652 	case RTM_DELADDR:
653 	case RTM_NEWADDR:
654 		len = sizeof(struct ifa_msghdr);
655 		break;
656 
657 	case RTM_IFINFO:
658 		len = sizeof(struct if_msghdr);
659 		break;
660 
661 	default:
662 		len = sizeof(struct rt_msghdr);
663 	}
664 	cp0 = cp;
665 	if (cp0)
666 		cp += len;
667 	for (i = 0; i < RTAX_MAX; i++) {
668 		struct sockaddr *sa;
669 
670 		if ((sa = rtinfo->rti_info[i]) == 0)
671 			continue;
672 		rtinfo->rti_addrs |= (1 << i);
673 		dlen = ROUNDUP(sa->sa_len);
674 		if (cp) {
675 			bcopy((caddr_t)sa, cp, (unsigned)dlen);
676 			cp += dlen;
677 		}
678 		len += dlen;
679 	}
680 	len = ALIGN(len);
681 	if (cp == 0 && w != NULL && !second_time) {
682 		struct walkarg *rw = w;
683 
684 		if (rw->w_req) {
685 			if (rw->w_tmemsize < len) {
686 				if (rw->w_tmem)
687 					free(rw->w_tmem, M_RTABLE);
688 				rw->w_tmem = (caddr_t)
689 					malloc(len, M_RTABLE, M_NOWAIT);
690 				if (rw->w_tmem)
691 					rw->w_tmemsize = len;
692 			}
693 			if (rw->w_tmem) {
694 				cp = rw->w_tmem;
695 				second_time = 1;
696 				goto again;
697 			}
698 		}
699 	}
700 	if (cp) {
701 		struct rt_msghdr *rtm = (struct rt_msghdr *)cp0;
702 
703 		rtm->rtm_version = RTM_VERSION;
704 		rtm->rtm_type = type;
705 		rtm->rtm_msglen = len;
706 	}
707 	return (len);
708 }
709 
710 /*
711  * This routine is called to generate a message from the routing
712  * socket indicating that a redirect has occured, a routing lookup
713  * has failed, or that a protocol has detected timeouts to a particular
714  * destination.
715  */
716 void
717 rt_missmsg(type, rtinfo, flags, error)
718 	int type, flags, error;
719 	struct rt_addrinfo *rtinfo;
720 {
721 	struct rt_msghdr *rtm;
722 	struct mbuf *m;
723 	struct sockaddr *sa = rtinfo->rti_info[RTAX_DST];
724 
725 	if (route_cb.any_count == 0)
726 		return;
727 	m = rt_msg1(type, rtinfo);
728 	if (m == 0)
729 		return;
730 	rtm = mtod(m, struct rt_msghdr *);
731 	rtm->rtm_flags = RTF_DONE | flags;
732 	rtm->rtm_errno = error;
733 	rtm->rtm_addrs = rtinfo->rti_addrs;
734 	route_proto.sp_protocol = sa ? sa->sa_family : 0;
735 	raw_input(m, &route_proto, &route_src, &route_dst);
736 }
737 
738 /*
739  * This routine is called to generate a message from the routing
740  * socket indicating that the status of a network interface has changed.
741  */
742 void
743 rt_ifmsg(ifp)
744 	struct ifnet *ifp;
745 {
746 	struct if_msghdr *ifm;
747 	struct mbuf *m;
748 	struct rt_addrinfo info;
749 
750 	if (route_cb.any_count == 0)
751 		return;
752 	bzero((caddr_t)&info, sizeof(info));
753 	m = rt_msg1(RTM_IFINFO, &info);
754 	if (m == 0)
755 		return;
756 	ifm = mtod(m, struct if_msghdr *);
757 	ifm->ifm_index = ifp->if_index;
758 	ifm->ifm_flags = (u_short)ifp->if_flags;
759 	ifm->ifm_data = ifp->if_data;
760 	ifm->ifm_addrs = 0;
761 	route_proto.sp_protocol = 0;
762 	raw_input(m, &route_proto, &route_src, &route_dst);
763 }
764 
765 static void
766 rt_ifamsg(int cmd, struct ifaddr *ifa)
767 {
768 	struct ifa_msghdr *ifam;
769 	struct rt_addrinfo info;
770 	struct mbuf *m;
771 	struct sockaddr *sa;
772 	struct ifnet *ifp = ifa->ifa_ifp;
773 
774 	bzero(&info, sizeof(info));
775 	ifaaddr = sa = ifa->ifa_addr;
776 	ifpaddr = TAILQ_FIRST(&ifp->if_addrhead)->ifa_addr;
777 	netmask = ifa->ifa_netmask;
778 	brdaddr = ifa->ifa_dstaddr;
779 	if ((m = rt_msg1(cmd, &info)) == NULL)
780 		return;
781 	ifam = mtod(m, struct ifa_msghdr *);
782 	ifam->ifam_index = ifp->if_index;
783 	ifam->ifam_metric = ifa->ifa_metric;
784 	ifam->ifam_flags = ifa->ifa_flags;
785 	ifam->ifam_addrs = info.rti_addrs;
786 
787 	route_proto.sp_protocol = sa ? sa->sa_family : 0;
788 	raw_input(m, &route_proto, &route_src, &route_dst);
789 }
790 
791 static void
792 rt_rtmsg(int cmd, struct ifaddr *ifa, int error, struct rtentry *rt)
793 {
794 	struct rt_msghdr *rtm;
795 	struct rt_addrinfo info;
796 	struct mbuf *m;
797 	struct sockaddr *sa;
798 	struct ifnet *ifp = ifa->ifa_ifp;
799 
800 	if (rt == NULL)
801 		return;
802 	bzero(&info, sizeof(info));
803 	netmask = rt_mask(rt);
804 	dst = sa = rt_key(rt);
805 	gate = rt->rt_gateway;
806 	if ((m = rt_msg1(cmd, &info)) == NULL)
807 		return;
808 	rtm = mtod(m, struct rt_msghdr *);
809 	rtm->rtm_index = ifp->if_index;
810 	rtm->rtm_flags |= rt->rt_flags;
811 	rtm->rtm_errno = error;
812 	rtm->rtm_addrs = info.rti_addrs;
813 
814 	route_proto.sp_protocol = sa ? sa->sa_family : 0;
815 	raw_input(m, &route_proto, &route_src, &route_dst);
816 }
817 
818 /*
819  * This is called to generate messages from the routing socket
820  * indicating a network interface has had addresses associated with it.
821  * if we ever reverse the logic and replace messages TO the routing
822  * socket indicate a request to configure interfaces, then it will
823  * be unnecessary as the routing socket will automatically generate
824  * copies of it.
825  */
826 void
827 rt_newaddrmsg(cmd, ifa, error, rt)
828 	int cmd, error;
829 	struct ifaddr *ifa;
830 	struct rtentry *rt;
831 {
832 	if (route_cb.any_count == 0)
833 		return;
834 
835 	if (cmd == RTM_ADD) {
836 		rt_ifamsg(RTM_NEWADDR, ifa);
837 		rt_rtmsg(RTM_ADD, ifa, error, rt);
838 	} else {
839 		KASSERT((cmd == RTM_DELETE), ("unknown cmd %d", cmd));
840 		rt_rtmsg(RTM_DELETE, ifa, error, rt);
841 		rt_ifamsg(RTM_DELADDR, ifa);
842 	}
843 }
844 
845 /*
846  * This is the analogue to the rt_newaddrmsg which performs the same
847  * function but for multicast group memberhips.  This is easier since
848  * there is no route state to worry about.
849  */
850 void
851 rt_newmaddrmsg(cmd, ifma)
852 	int cmd;
853 	struct ifmultiaddr *ifma;
854 {
855 	struct rt_addrinfo info;
856 	struct mbuf *m = 0;
857 	struct ifnet *ifp = ifma->ifma_ifp;
858 	struct ifma_msghdr *ifmam;
859 
860 	if (route_cb.any_count == 0)
861 		return;
862 
863 	bzero((caddr_t)&info, sizeof(info));
864 	ifaaddr = ifma->ifma_addr;
865 	if (ifp && TAILQ_FIRST(&ifp->if_addrhead))
866 		ifpaddr = TAILQ_FIRST(&ifp->if_addrhead)->ifa_addr;
867 	else
868 		ifpaddr = NULL;
869 	/*
870 	 * If a link-layer address is present, present it as a ``gateway''
871 	 * (similarly to how ARP entries, e.g., are presented).
872 	 */
873 	gate = ifma->ifma_lladdr;
874 	if ((m = rt_msg1(cmd, &info)) == NULL)
875 		return;
876 	ifmam = mtod(m, struct ifma_msghdr *);
877 	ifmam->ifmam_index = ifp->if_index;
878 	ifmam->ifmam_addrs = info.rti_addrs;
879 	route_proto.sp_protocol = ifma->ifma_addr->sa_family;
880 	raw_input(m, &route_proto, &route_src, &route_dst);
881 }
882 
883 /*
884  * This is called to generate routing socket messages indicating
885  * network interface arrival and departure.
886  */
887 void
888 rt_ifannouncemsg(ifp, what)
889 	struct ifnet *ifp;
890 	int what;
891 {
892 	struct if_announcemsghdr *ifan;
893 	struct mbuf *m;
894 	struct rt_addrinfo info;
895 
896 	if (route_cb.any_count == 0)
897 		return;
898 	bzero((caddr_t)&info, sizeof(info));
899 	m = rt_msg1(RTM_IFANNOUNCE, &info);
900 	if (m == NULL)
901 		return;
902 	ifan = mtod(m, struct if_announcemsghdr *);
903 	ifan->ifan_index = ifp->if_index;
904 	snprintf(ifan->ifan_name, sizeof(ifan->ifan_name),
905 	    "%s%d", ifp->if_name, ifp->if_unit);
906 	ifan->ifan_what = what;
907 	route_proto.sp_protocol = 0;
908 	raw_input(m, &route_proto, &route_src, &route_dst);
909  }
910 
911 /*
912  * This is used in dumping the kernel table via sysctl().
913  */
914 int
915 sysctl_dumpentry(rn, vw)
916 	struct radix_node *rn;
917 	void *vw;
918 {
919 	struct walkarg *w = vw;
920 	struct rtentry *rt = (struct rtentry *)rn;
921 	int error = 0, size;
922 	struct rt_addrinfo info;
923 
924 	if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg))
925 		return 0;
926 	bzero((caddr_t)&info, sizeof(info));
927 	dst = rt_key(rt);
928 	gate = rt->rt_gateway;
929 	netmask = rt_mask(rt);
930 	genmask = rt->rt_genmask;
931 	if (rt->rt_ifp) {
932 		ifpaddr = TAILQ_FIRST(&rt->rt_ifp->if_addrhead)->ifa_addr;
933 		ifaaddr = rt->rt_ifa->ifa_addr;
934 		if (rt->rt_ifp->if_flags & IFF_POINTOPOINT)
935 			brdaddr = rt->rt_ifa->ifa_dstaddr;
936 	}
937 	size = rt_msg2(RTM_GET, &info, 0, w);
938 	if (w->w_req && w->w_tmem) {
939 		struct rt_msghdr *rtm = (struct rt_msghdr *)w->w_tmem;
940 
941 		rtm->rtm_flags = rt->rt_flags;
942 		rtm->rtm_use = rt->rt_use;
943 		rtm->rtm_rmx = rt->rt_rmx;
944 		rtm->rtm_index = rt->rt_ifp->if_index;
945 		rtm->rtm_errno = rtm->rtm_pid = rtm->rtm_seq = 0;
946 		rtm->rtm_addrs = info.rti_addrs;
947 		error = SYSCTL_OUT(w->w_req, (caddr_t)rtm, size);
948 		return (error);
949 	}
950 	return (error);
951 }
952 
953 int
954 sysctl_iflist(af, w)
955 	int	af;
956 	struct	walkarg *w;
957 {
958 	struct ifnet *ifp;
959 	struct ifaddr *ifa;
960 	struct	rt_addrinfo info;
961 	int	len, error = 0;
962 
963 	bzero((caddr_t)&info, sizeof(info));
964 	TAILQ_FOREACH(ifp, &ifnet, if_link) {
965 		if (w->w_arg && w->w_arg != ifp->if_index)
966 			continue;
967 		ifa = TAILQ_FIRST(&ifp->if_addrhead);
968 		ifpaddr = ifa->ifa_addr;
969 		len = rt_msg2(RTM_IFINFO, &info, (caddr_t)0, w);
970 		ifpaddr = 0;
971 		if (w->w_req && w->w_tmem) {
972 			struct if_msghdr *ifm;
973 
974 			ifm = (struct if_msghdr *)w->w_tmem;
975 			ifm->ifm_index = ifp->if_index;
976 			ifm->ifm_flags = (u_short)ifp->if_flags;
977 			ifm->ifm_data = ifp->if_data;
978 			ifm->ifm_addrs = info.rti_addrs;
979 			error = SYSCTL_OUT(w->w_req,(caddr_t)ifm, len);
980 			if (error)
981 				return (error);
982 		}
983 		while ((ifa = TAILQ_NEXT(ifa, ifa_link)) != 0) {
984 			if (af && af != ifa->ifa_addr->sa_family)
985 				continue;
986 			if (curproc->p_ucred->cr_prison && prison_if(curthread, ifa->ifa_addr))
987 				continue;
988 			ifaaddr = ifa->ifa_addr;
989 			netmask = ifa->ifa_netmask;
990 			brdaddr = ifa->ifa_dstaddr;
991 			len = rt_msg2(RTM_NEWADDR, &info, 0, w);
992 			if (w->w_req && w->w_tmem) {
993 				struct ifa_msghdr *ifam;
994 
995 				ifam = (struct ifa_msghdr *)w->w_tmem;
996 				ifam->ifam_index = ifa->ifa_ifp->if_index;
997 				ifam->ifam_flags = ifa->ifa_flags;
998 				ifam->ifam_metric = ifa->ifa_metric;
999 				ifam->ifam_addrs = info.rti_addrs;
1000 				error = SYSCTL_OUT(w->w_req, w->w_tmem, len);
1001 				if (error)
1002 					return (error);
1003 			}
1004 		}
1005 		ifaaddr = netmask = brdaddr = 0;
1006 	}
1007 	return (0);
1008 }
1009 
1010 static int
1011 sysctl_rtsock(SYSCTL_HANDLER_ARGS)
1012 {
1013 	int	*name = (int *)arg1;
1014 	u_int	namelen = arg2;
1015 	struct radix_node_head *rnh;
1016 	int	i, s, error = EINVAL;
1017 	u_char  af;
1018 	struct	walkarg w;
1019 
1020 	name ++;
1021 	namelen--;
1022 	if (req->newptr)
1023 		return (EPERM);
1024 	if (namelen != 3)
1025 		return (EINVAL);
1026 	af = name[0];
1027 	Bzero(&w, sizeof(w));
1028 	w.w_op = name[1];
1029 	w.w_arg = name[2];
1030 	w.w_req = req;
1031 
1032 	s = splnet();
1033 	switch (w.w_op) {
1034 
1035 	case NET_RT_DUMP:
1036 	case NET_RT_FLAGS:
1037 		for (i = 1; i <= AF_MAX; i++)
1038 			if ((rnh = rt_tables[i]) && (af == 0 || af == i) &&
1039 			    (error = rnh->rnh_walktree(rnh,
1040 							sysctl_dumpentry, &w)))
1041 				break;
1042 		break;
1043 
1044 	case NET_RT_IFLIST:
1045 		error = sysctl_iflist(af, &w);
1046 	}
1047 	splx(s);
1048 	if (w.w_tmem)
1049 		free(w.w_tmem, M_RTABLE);
1050 	return (error);
1051 }
1052 
1053 SYSCTL_NODE(_net, PF_ROUTE, routetable, CTLFLAG_RD, sysctl_rtsock, "");
1054 
1055 /*
1056  * Definitions of protocols supported in the ROUTE domain.
1057  */
1058 
1059 extern struct domain routedomain;		/* or at least forward */
1060 
1061 static struct protosw routesw[] = {
1062 { SOCK_RAW,	&routedomain,	0,		PR_ATOMIC|PR_ADDR,
1063   0,		route_output,	raw_ctlinput,	0,
1064   0,
1065   raw_init,	0,		0,		0,
1066   &route_usrreqs
1067 }
1068 };
1069 
1070 static struct domain routedomain =
1071     { PF_ROUTE, "route", 0, 0, 0,
1072       routesw, &routesw[sizeof(routesw)/sizeof(routesw[0])] };
1073 
1074 DOMAIN_SET(route);
1075