xref: /freebsd/sys/net/rtsock.c (revision ce834215a70ff69e7e222827437116eee2f9ac6f)
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	8.5 (Berkeley) 11/2/94
34  *	$Id: rtsock.c,v 1.26 1997/02/22 09:41:15 peter Exp $
35  */
36 
37 
38 #include <sys/param.h>
39 #include <sys/queue.h>
40 #include <sys/systm.h>
41 #include <sys/kernel.h>
42 #include <sys/sysctl.h>
43 #include <sys/proc.h>
44 #include <sys/mbuf.h>
45 #include <sys/socket.h>
46 #include <sys/socketvar.h>
47 #include <sys/domain.h>
48 #include <sys/protosw.h>
49 
50 #include <net/if.h>
51 #include <net/route.h>
52 #include <net/raw_cb.h>
53 
54 static struct	sockaddr route_dst = { 2, PF_ROUTE, };
55 static struct	sockaddr route_src = { 2, PF_ROUTE, };
56 static struct	sockproto route_proto = { PF_ROUTE, };
57 
58 struct walkarg {
59 	int	w_tmemsize;
60 	int	w_op, w_arg;
61 	caddr_t	w_tmem;
62 	struct sysctl_req *w_req;
63 };
64 
65 static struct mbuf *
66 		rt_msg1 __P((int, struct rt_addrinfo *));
67 static int	rt_msg2 __P((int,
68 		    struct rt_addrinfo *, caddr_t, struct walkarg *));
69 static void	rt_xaddrs __P((caddr_t, caddr_t, struct rt_addrinfo *));
70 static int	sysctl_dumpentry __P((struct radix_node *rn, void *vw));
71 static int	sysctl_iflist __P((int af, struct walkarg *w));
72 static int	 route_output __P((struct mbuf *, struct socket *));
73 static void	 rt_setmetrics __P((u_long, struct rt_metrics *, struct rt_metrics *));
74 
75 /* Sleazy use of local variables throughout file, warning!!!! */
76 #define dst	info.rti_info[RTAX_DST]
77 #define gate	info.rti_info[RTAX_GATEWAY]
78 #define netmask	info.rti_info[RTAX_NETMASK]
79 #define genmask	info.rti_info[RTAX_GENMASK]
80 #define ifpaddr	info.rti_info[RTAX_IFP]
81 #define ifaaddr	info.rti_info[RTAX_IFA]
82 #define brdaddr	info.rti_info[RTAX_BRD]
83 
84 /*
85  * It really doesn't make any sense at all for this code to share much
86  * with raw_usrreq.c, since its functionality is so restricted.  XXX
87  */
88 static int
89 rts_abort(struct socket *so)
90 {
91 	int s, error;
92 	s = splnet();
93 	error = raw_usrreqs.pru_abort(so);
94 	splx(s);
95 	return error;
96 }
97 
98 /* pru_accept is EOPNOTSUPP */
99 
100 static int
101 rts_attach(struct socket *so, int proto, struct proc *p)
102 {
103 	struct rawcb *rp;
104 	int s, error;
105 
106 	if (sotorawcb(so) != 0)
107 		return EISCONN;	/* XXX panic? */
108 	MALLOC(rp, struct rawcb *, sizeof *rp, M_PCB, M_WAITOK); /* XXX */
109 	if (rp == 0)
110 		return ENOBUFS;
111 	bzero(rp, sizeof *rp);
112 
113 	/*
114 	 * The splnet() is necessary to block protocols from sending
115 	 * error notifications (like RTM_REDIRECT or RTM_LOSING) while
116 	 * this PCB is extant but incompletely initialized.
117 	 * Probably we should try to do more of this work beforehand and
118 	 * eliminate the spl.
119 	 */
120 	s = splnet();
121 	so->so_pcb = (caddr_t)rp;
122 	error = raw_usrreqs.pru_attach(so, proto, p);
123 	rp = sotorawcb(so);
124 	if (error) {
125 		splx(s);
126 		free(rp, M_PCB);
127 		return error;
128 	}
129 	switch(rp->rcb_proto.sp_protocol) {
130 	case AF_INET:
131 		route_cb.ip_count++;
132 		break;
133 	case AF_IPX:
134 		route_cb.ipx_count++;
135 		break;
136 	case AF_NS:
137 		route_cb.ns_count++;
138 		break;
139 	case AF_ISO:
140 		route_cb.iso_count++;
141 		break;
142 	}
143 	rp->rcb_faddr = &route_src;
144 	route_cb.any_count++;
145 	soisconnected(so);
146 	so->so_options |= SO_USELOOPBACK;
147 	splx(s);
148 	return 0;
149 }
150 
151 static int
152 rts_bind(struct socket *so, struct mbuf *nam, struct proc *p)
153 {
154 	int s, error;
155 	s = splnet();
156 	error = raw_usrreqs.pru_bind(so, nam, p); /* xxx just EINVAL */
157 	splx(s);
158 	return error;
159 }
160 
161 static int
162 rts_connect(struct socket *so, struct mbuf *nam, struct proc *p)
163 {
164 	int s, error;
165 	s = splnet();
166 	error = raw_usrreqs.pru_connect(so, nam, p); /* XXX just EINVAL */
167 	splx(s);
168 	return error;
169 }
170 
171 /* pru_connect2 is EOPNOTSUPP */
172 /* pru_control is EOPNOTSUPP */
173 
174 static int
175 rts_detach(struct socket *so)
176 {
177 	struct rawcb *rp = sotorawcb(so);
178 	int s, error;
179 
180 	s = splnet();
181 	if (rp != 0) {
182 		switch(rp->rcb_proto.sp_protocol) {
183 		case AF_INET:
184 			route_cb.ip_count--;
185 			break;
186 		case AF_IPX:
187 			route_cb.ipx_count--;
188 			break;
189 		case AF_NS:
190 			route_cb.ns_count--;
191 			break;
192 		case AF_ISO:
193 			route_cb.iso_count--;
194 			break;
195 		}
196 		route_cb.any_count--;
197 	}
198 	error = raw_usrreqs.pru_detach(so);
199 	splx(s);
200 	return error;
201 }
202 
203 static int
204 rts_disconnect(struct socket *so)
205 {
206 	int s, error;
207 	s = splnet();
208 	error = raw_usrreqs.pru_disconnect(so);
209 	splx(s);
210 	return error;
211 }
212 
213 /* pru_listen is EOPNOTSUPP */
214 
215 static int
216 rts_peeraddr(struct socket *so, struct mbuf *nam)
217 {
218 	int s, error;
219 	s = splnet();
220 	error = raw_usrreqs.pru_peeraddr(so, nam);
221 	splx(s);
222 	return error;
223 }
224 
225 /* pru_rcvd is EOPNOTSUPP */
226 /* pru_rcvoob is EOPNOTSUPP */
227 
228 static int
229 rts_send(struct socket *so, int flags, struct mbuf *m, struct mbuf *nam,
230 	 struct mbuf *control, struct proc *p)
231 {
232 	int s, error;
233 	s = splnet();
234 	error = raw_usrreqs.pru_send(so, flags, m, nam, control, p);
235 	splx(s);
236 	return error;
237 }
238 
239 /* pru_sense is null */
240 
241 static int
242 rts_shutdown(struct socket *so)
243 {
244 	int s, error;
245 	s = splnet();
246 	error = raw_usrreqs.pru_shutdown(so);
247 	splx(s);
248 	return error;
249 }
250 
251 static int
252 rts_sockaddr(struct socket *so, struct mbuf *nam)
253 {
254 	int s, error;
255 	s = splnet();
256 	error = raw_usrreqs.pru_sockaddr(so, nam);
257 	splx(s);
258 	return error;
259 }
260 
261 static struct pr_usrreqs route_usrreqs = {
262 	rts_abort, pru_accept_notsupp, rts_attach, rts_bind, rts_connect,
263 	pru_connect2_notsupp, pru_control_notsupp, rts_detach, rts_disconnect,
264 	pru_listen_notsupp, rts_peeraddr, pru_rcvd_notsupp, pru_rcvoob_notsupp,
265 	rts_send, pru_sense_null, rts_shutdown, rts_sockaddr,
266 	sosend, soreceive, soselect
267 };
268 
269 /*ARGSUSED*/
270 static int
271 route_output(m, so)
272 	register struct mbuf *m;
273 	struct socket *so;
274 {
275 	register struct rt_msghdr *rtm = 0;
276 	register struct rtentry *rt = 0;
277 	struct rtentry *saved_nrt = 0;
278 	struct radix_node_head *rnh;
279 	struct rt_addrinfo info;
280 	int len, error = 0;
281 	struct ifnet *ifp = 0;
282 	struct ifaddr *ifa = 0;
283 
284 #define senderr(e) { error = e; goto flush;}
285 	if (m == 0 || ((m->m_len < sizeof(long)) &&
286 		       (m = m_pullup(m, sizeof(long))) == 0))
287 		return (ENOBUFS);
288 	if ((m->m_flags & M_PKTHDR) == 0)
289 		panic("route_output");
290 	len = m->m_pkthdr.len;
291 	if (len < sizeof(*rtm) ||
292 	    len != mtod(m, struct rt_msghdr *)->rtm_msglen) {
293 		dst = 0;
294 		senderr(EINVAL);
295 	}
296 	R_Malloc(rtm, struct rt_msghdr *, len);
297 	if (rtm == 0) {
298 		dst = 0;
299 		senderr(ENOBUFS);
300 	}
301 	m_copydata(m, 0, len, (caddr_t)rtm);
302 	if (rtm->rtm_version != RTM_VERSION) {
303 		dst = 0;
304 		senderr(EPROTONOSUPPORT);
305 	}
306 	rtm->rtm_pid = curproc->p_pid;
307 	info.rti_addrs = rtm->rtm_addrs;
308 	rt_xaddrs((caddr_t)(rtm + 1), len + (caddr_t)rtm, &info);
309 	if (dst == 0 || (dst->sa_family >= AF_MAX)
310 	    || (gate != 0 && (gate->sa_family >= AF_MAX)))
311 		senderr(EINVAL);
312 	if (genmask) {
313 		struct radix_node *t;
314 		t = rn_addmask((caddr_t)genmask, 0, 1);
315 		if (t && Bcmp(genmask, t->rn_key, *(u_char *)genmask) == 0)
316 			genmask = (struct sockaddr *)(t->rn_key);
317 		else
318 			senderr(ENOBUFS);
319 	}
320 	switch (rtm->rtm_type) {
321 
322 	case RTM_ADD:
323 		if (gate == 0)
324 			senderr(EINVAL);
325 		error = rtrequest(RTM_ADD, dst, gate, netmask,
326 					rtm->rtm_flags, &saved_nrt);
327 		if (error == 0 && saved_nrt) {
328 			rt_setmetrics(rtm->rtm_inits,
329 				&rtm->rtm_rmx, &saved_nrt->rt_rmx);
330 			saved_nrt->rt_refcnt--;
331 			saved_nrt->rt_genmask = genmask;
332 		}
333 		break;
334 
335 	case RTM_DELETE:
336 		error = rtrequest(RTM_DELETE, dst, gate, netmask,
337 				rtm->rtm_flags, &saved_nrt);
338 		if (error == 0) {
339 			if ((rt = saved_nrt))
340 				rt->rt_refcnt++;
341 			goto report;
342 		}
343 		break;
344 
345 	case RTM_GET:
346 	case RTM_CHANGE:
347 	case RTM_LOCK:
348 		if ((rnh = rt_tables[dst->sa_family]) == 0) {
349 			senderr(EAFNOSUPPORT);
350 		} else if (rt = (struct rtentry *)
351 				rnh->rnh_lookup(dst, netmask, rnh))
352 			rt->rt_refcnt++;
353 		else
354 			senderr(ESRCH);
355 		switch(rtm->rtm_type) {
356 
357 		case RTM_GET:
358 		report:
359 			dst = rt_key(rt);
360 			gate = rt->rt_gateway;
361 			netmask = rt_mask(rt);
362 			genmask = rt->rt_genmask;
363 			if (rtm->rtm_addrs & (RTA_IFP | RTA_IFA)) {
364 				ifp = rt->rt_ifp;
365 				if (ifp) {
366 					ifpaddr = ifp->if_addrhead.tqh_first->ifa_addr;
367 					ifaaddr = rt->rt_ifa->ifa_addr;
368 					rtm->rtm_index = ifp->if_index;
369 				} else {
370 					ifpaddr = 0;
371 					ifaaddr = 0;
372 			    }
373 			}
374 			len = rt_msg2(rtm->rtm_type, &info, (caddr_t)0,
375 				(struct walkarg *)0);
376 			if (len > rtm->rtm_msglen) {
377 				struct rt_msghdr *new_rtm;
378 				R_Malloc(new_rtm, struct rt_msghdr *, len);
379 				if (new_rtm == 0)
380 					senderr(ENOBUFS);
381 				Bcopy(rtm, new_rtm, rtm->rtm_msglen);
382 				Free(rtm); rtm = new_rtm;
383 			}
384 			(void)rt_msg2(rtm->rtm_type, &info, (caddr_t)rtm,
385 				(struct walkarg *)0);
386 			rtm->rtm_flags = rt->rt_flags;
387 			rtm->rtm_rmx = rt->rt_rmx;
388 			rtm->rtm_addrs = info.rti_addrs;
389 			break;
390 
391 		case RTM_CHANGE:
392 			if (gate && (error = rt_setgate(rt, rt_key(rt), gate)))
393 				senderr(error);
394 
395 			/*
396 			 * If they tried to change things but didn't specify
397 			 * the required gateway, then just use the old one.
398 			 * This can happen if the user tries to change the
399 			 * flags on the default route without changing the
400 			 * default gateway.  Changing flags still doesn't work.
401 			 */
402 			if ((rt->rt_flags & RTF_GATEWAY) && !gate)
403 				gate = rt->rt_gateway;
404 
405 			/* new gateway could require new ifaddr, ifp;
406 			   flags may also be different; ifp may be specified
407 			   by ll sockaddr when protocol address is ambiguous */
408 			if (ifpaddr && (ifa = ifa_ifwithnet(ifpaddr)) &&
409 			    (ifp = ifa->ifa_ifp) && (ifaaddr || gate))
410 				ifa = ifaof_ifpforaddr(ifaaddr ? ifaaddr : gate,
411 							ifp);
412 			else if ((ifaaddr && (ifa = ifa_ifwithaddr(ifaaddr))) ||
413 				 (gate && (ifa = ifa_ifwithroute(rt->rt_flags,
414 							rt_key(rt), gate))))
415 				ifp = ifa->ifa_ifp;
416 			if (ifa) {
417 				register struct ifaddr *oifa = rt->rt_ifa;
418 				if (oifa != ifa) {
419 				    if (oifa && oifa->ifa_rtrequest)
420 					oifa->ifa_rtrequest(RTM_DELETE,
421 								rt, gate);
422 				    IFAFREE(rt->rt_ifa);
423 				    rt->rt_ifa = ifa;
424 				    ifa->ifa_refcnt++;
425 				    rt->rt_ifp = ifp;
426 				}
427 			}
428 			rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx,
429 					&rt->rt_rmx);
430 			if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest)
431 			       rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, gate);
432 			if (genmask)
433 				rt->rt_genmask = genmask;
434 			/*
435 			 * Fall into
436 			 */
437 		case RTM_LOCK:
438 			rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits);
439 			rt->rt_rmx.rmx_locks |=
440 				(rtm->rtm_inits & rtm->rtm_rmx.rmx_locks);
441 			break;
442 		}
443 		break;
444 
445 	default:
446 		senderr(EOPNOTSUPP);
447 	}
448 
449 flush:
450 	if (rtm) {
451 		if (error)
452 			rtm->rtm_errno = error;
453 		else
454 			rtm->rtm_flags |= RTF_DONE;
455 	}
456 	if (rt)
457 		rtfree(rt);
458     {
459 	register struct rawcb *rp = 0;
460 	/*
461 	 * Check to see if we don't want our own messages.
462 	 */
463 	if ((so->so_options & SO_USELOOPBACK) == 0) {
464 		if (route_cb.any_count <= 1) {
465 			if (rtm)
466 				Free(rtm);
467 			m_freem(m);
468 			return (error);
469 		}
470 		/* There is another listener, so construct message */
471 		rp = sotorawcb(so);
472 	}
473 	if (rtm) {
474 		m_copyback(m, 0, rtm->rtm_msglen, (caddr_t)rtm);
475 		Free(rtm);
476 	}
477 	if (rp)
478 		rp->rcb_proto.sp_family = 0; /* Avoid us */
479 	if (dst)
480 		route_proto.sp_protocol = dst->sa_family;
481 	raw_input(m, &route_proto, &route_src, &route_dst);
482 	if (rp)
483 		rp->rcb_proto.sp_family = PF_ROUTE;
484     }
485 	return (error);
486 }
487 
488 static void
489 rt_setmetrics(which, in, out)
490 	u_long which;
491 	register struct rt_metrics *in, *out;
492 {
493 #define metric(f, e) if (which & (f)) out->e = in->e;
494 	metric(RTV_RPIPE, rmx_recvpipe);
495 	metric(RTV_SPIPE, rmx_sendpipe);
496 	metric(RTV_SSTHRESH, rmx_ssthresh);
497 	metric(RTV_RTT, rmx_rtt);
498 	metric(RTV_RTTVAR, rmx_rttvar);
499 	metric(RTV_HOPCOUNT, rmx_hopcount);
500 	metric(RTV_MTU, rmx_mtu);
501 	metric(RTV_EXPIRE, rmx_expire);
502 #undef metric
503 }
504 
505 #define ROUNDUP(a) \
506 	((a) > 0 ? (1 + (((a) - 1) | (sizeof(long) - 1))) : sizeof(long))
507 #define ADVANCE(x, n) (x += ROUNDUP((n)->sa_len))
508 
509 static void
510 rt_xaddrs(cp, cplim, rtinfo)
511 	register caddr_t cp, cplim;
512 	register struct rt_addrinfo *rtinfo;
513 {
514 	register struct sockaddr *sa;
515 	register int i;
516 
517 	bzero(rtinfo->rti_info, sizeof(rtinfo->rti_info));
518 	for (i = 0; (i < RTAX_MAX) && (cp < cplim); i++) {
519 		if ((rtinfo->rti_addrs & (1 << i)) == 0)
520 			continue;
521 		rtinfo->rti_info[i] = sa = (struct sockaddr *)cp;
522 		ADVANCE(cp, sa);
523 	}
524 }
525 
526 static struct mbuf *
527 rt_msg1(type, rtinfo)
528 	int type;
529 	register struct rt_addrinfo *rtinfo;
530 {
531 	register struct rt_msghdr *rtm;
532 	register struct mbuf *m;
533 	register int i;
534 	register struct sockaddr *sa;
535 	int len, dlen;
536 
537 	m = m_gethdr(M_DONTWAIT, MT_DATA);
538 	if (m == 0)
539 		return (m);
540 	switch (type) {
541 
542 	case RTM_DELADDR:
543 	case RTM_NEWADDR:
544 		len = sizeof(struct ifa_msghdr);
545 		break;
546 
547 	case RTM_DELMADDR:
548 	case RTM_NEWMADDR:
549 		len = sizeof(struct ifma_msghdr);
550 		break;
551 
552 	case RTM_IFINFO:
553 		len = sizeof(struct if_msghdr);
554 		break;
555 
556 	default:
557 		len = sizeof(struct rt_msghdr);
558 	}
559 	if (len > MHLEN)
560 		panic("rt_msg1");
561 	m->m_pkthdr.len = m->m_len = len;
562 	m->m_pkthdr.rcvif = 0;
563 	rtm = mtod(m, struct rt_msghdr *);
564 	bzero((caddr_t)rtm, len);
565 	for (i = 0; i < RTAX_MAX; i++) {
566 		if ((sa = rtinfo->rti_info[i]) == NULL)
567 			continue;
568 		rtinfo->rti_addrs |= (1 << i);
569 		dlen = ROUNDUP(sa->sa_len);
570 		m_copyback(m, len, dlen, (caddr_t)sa);
571 		len += dlen;
572 	}
573 	if (m->m_pkthdr.len != len) {
574 		m_freem(m);
575 		return (NULL);
576 	}
577 	rtm->rtm_msglen = len;
578 	rtm->rtm_version = RTM_VERSION;
579 	rtm->rtm_type = type;
580 	return (m);
581 }
582 
583 static int
584 rt_msg2(type, rtinfo, cp, w)
585 	int type;
586 	register struct rt_addrinfo *rtinfo;
587 	caddr_t cp;
588 	struct walkarg *w;
589 {
590 	register int i;
591 	int len, dlen, second_time = 0;
592 	caddr_t cp0;
593 
594 	rtinfo->rti_addrs = 0;
595 again:
596 	switch (type) {
597 
598 	case RTM_DELADDR:
599 	case RTM_NEWADDR:
600 		len = sizeof(struct ifa_msghdr);
601 		break;
602 
603 	case RTM_IFINFO:
604 		len = sizeof(struct if_msghdr);
605 		break;
606 
607 	default:
608 		len = sizeof(struct rt_msghdr);
609 	}
610 	cp0 = cp;
611 	if (cp0)
612 		cp += len;
613 	for (i = 0; i < RTAX_MAX; i++) {
614 		register struct sockaddr *sa;
615 
616 		if ((sa = rtinfo->rti_info[i]) == 0)
617 			continue;
618 		rtinfo->rti_addrs |= (1 << i);
619 		dlen = ROUNDUP(sa->sa_len);
620 		if (cp) {
621 			bcopy((caddr_t)sa, cp, (unsigned)dlen);
622 			cp += dlen;
623 		}
624 		len += dlen;
625 	}
626 	if (cp == 0 && w != NULL && !second_time) {
627 		register struct walkarg *rw = w;
628 
629 		if (rw->w_req) {
630 			if (rw->w_tmemsize < len) {
631 				if (rw->w_tmem)
632 					free(rw->w_tmem, M_RTABLE);
633 				rw->w_tmem = (caddr_t)
634 					malloc(len, M_RTABLE, M_NOWAIT);
635 				if (rw->w_tmem)
636 					rw->w_tmemsize = len;
637 			}
638 			if (rw->w_tmem) {
639 				cp = rw->w_tmem;
640 				second_time = 1;
641 				goto again;
642 			}
643 		}
644 	}
645 	if (cp) {
646 		register struct rt_msghdr *rtm = (struct rt_msghdr *)cp0;
647 
648 		rtm->rtm_version = RTM_VERSION;
649 		rtm->rtm_type = type;
650 		rtm->rtm_msglen = len;
651 	}
652 	return (len);
653 }
654 
655 /*
656  * This routine is called to generate a message from the routing
657  * socket indicating that a redirect has occured, a routing lookup
658  * has failed, or that a protocol has detected timeouts to a particular
659  * destination.
660  */
661 void
662 rt_missmsg(type, rtinfo, flags, error)
663 	int type, flags, error;
664 	register struct rt_addrinfo *rtinfo;
665 {
666 	register struct rt_msghdr *rtm;
667 	register struct mbuf *m;
668 	struct sockaddr *sa = rtinfo->rti_info[RTAX_DST];
669 
670 	if (route_cb.any_count == 0)
671 		return;
672 	m = rt_msg1(type, rtinfo);
673 	if (m == 0)
674 		return;
675 	rtm = mtod(m, struct rt_msghdr *);
676 	rtm->rtm_flags = RTF_DONE | flags;
677 	rtm->rtm_errno = error;
678 	rtm->rtm_addrs = rtinfo->rti_addrs;
679 	route_proto.sp_protocol = sa ? sa->sa_family : 0;
680 	raw_input(m, &route_proto, &route_src, &route_dst);
681 }
682 
683 /*
684  * This routine is called to generate a message from the routing
685  * socket indicating that the status of a network interface has changed.
686  */
687 void
688 rt_ifmsg(ifp)
689 	register struct ifnet *ifp;
690 {
691 	register struct if_msghdr *ifm;
692 	struct mbuf *m;
693 	struct rt_addrinfo info;
694 
695 	if (route_cb.any_count == 0)
696 		return;
697 	bzero((caddr_t)&info, sizeof(info));
698 	m = rt_msg1(RTM_IFINFO, &info);
699 	if (m == 0)
700 		return;
701 	ifm = mtod(m, struct if_msghdr *);
702 	ifm->ifm_index = ifp->if_index;
703 	ifm->ifm_flags = (u_short)ifp->if_flags;
704 	ifm->ifm_data = ifp->if_data;
705 	ifm->ifm_addrs = 0;
706 	route_proto.sp_protocol = 0;
707 	raw_input(m, &route_proto, &route_src, &route_dst);
708 }
709 
710 /*
711  * This is called to generate messages from the routing socket
712  * indicating a network interface has had addresses associated with it.
713  * if we ever reverse the logic and replace messages TO the routing
714  * socket indicate a request to configure interfaces, then it will
715  * be unnecessary as the routing socket will automatically generate
716  * copies of it.
717  */
718 void
719 rt_newaddrmsg(cmd, ifa, error, rt)
720 	int cmd, error;
721 	register struct ifaddr *ifa;
722 	register struct rtentry *rt;
723 {
724 	struct rt_addrinfo info;
725 	struct sockaddr *sa = 0;
726 	int pass;
727 	struct mbuf *m = 0;
728 	struct ifnet *ifp = ifa->ifa_ifp;
729 
730 	if (route_cb.any_count == 0)
731 		return;
732 	for (pass = 1; pass < 3; pass++) {
733 		bzero((caddr_t)&info, sizeof(info));
734 		if ((cmd == RTM_ADD && pass == 1) ||
735 		    (cmd == RTM_DELETE && pass == 2)) {
736 			register struct ifa_msghdr *ifam;
737 			int ncmd = cmd == RTM_ADD ? RTM_NEWADDR : RTM_DELADDR;
738 
739 			ifaaddr = sa = ifa->ifa_addr;
740 			ifpaddr = ifp->if_addrhead.tqh_first->ifa_addr;
741 			netmask = ifa->ifa_netmask;
742 			brdaddr = ifa->ifa_dstaddr;
743 			if ((m = rt_msg1(ncmd, &info)) == NULL)
744 				continue;
745 			ifam = mtod(m, struct ifa_msghdr *);
746 			ifam->ifam_index = ifp->if_index;
747 			ifam->ifam_metric = ifa->ifa_metric;
748 			ifam->ifam_flags = ifa->ifa_flags;
749 			ifam->ifam_addrs = info.rti_addrs;
750 		}
751 		if ((cmd == RTM_ADD && pass == 2) ||
752 		    (cmd == RTM_DELETE && pass == 1)) {
753 			register struct rt_msghdr *rtm;
754 
755 			if (rt == 0)
756 				continue;
757 			netmask = rt_mask(rt);
758 			dst = sa = rt_key(rt);
759 			gate = rt->rt_gateway;
760 			if ((m = rt_msg1(cmd, &info)) == NULL)
761 				continue;
762 			rtm = mtod(m, struct rt_msghdr *);
763 			rtm->rtm_index = ifp->if_index;
764 			rtm->rtm_flags |= rt->rt_flags;
765 			rtm->rtm_errno = error;
766 			rtm->rtm_addrs = info.rti_addrs;
767 		}
768 		route_proto.sp_protocol = sa ? sa->sa_family : 0;
769 		raw_input(m, &route_proto, &route_src, &route_dst);
770 	}
771 }
772 
773 /*
774  * This is the analogue to the rt_newaddrmsg which performs the same
775  * function but for multicast group memberhips.  This is easier since
776  * there is no route state to worry about.
777  */
778 void
779 rt_newmaddrmsg(cmd, ifma)
780 	int cmd;
781 	struct ifmultiaddr *ifma;
782 {
783 	struct rt_addrinfo info;
784 	struct mbuf *m = 0;
785 	struct ifnet *ifp = ifma->ifma_ifp;
786 	struct ifma_msghdr *ifmam;
787 
788 	if (route_cb.any_count == 0)
789 		return;
790 
791 	bzero((caddr_t)&info, sizeof(info));
792 	ifaaddr = ifma->ifma_addr;
793 	ifpaddr = ifp->if_addrhead.tqh_first->ifa_addr;
794 	/*
795 	 * If a link-layer address is present, present it as a ``gateway''
796 	 * (similarly to how ARP entries, e.g., are presented).
797 	 */
798 	gate = ifma->ifma_lladdr;
799 	if ((m = rt_msg1(cmd, &info)) == NULL)
800 		return;
801 	ifmam = mtod(m, struct ifma_msghdr *);
802 	ifmam->ifmam_index = ifp->if_index;
803 	ifmam->ifmam_addrs = info.rti_addrs;
804 	route_proto.sp_protocol = ifma->ifma_addr->sa_family;
805 	raw_input(m, &route_proto, &route_src, &route_dst);
806 }
807 
808 /*
809  * This is used in dumping the kernel table via sysctl().
810  */
811 int
812 sysctl_dumpentry(rn, vw)
813 	struct radix_node *rn;
814 	void *vw;
815 {
816 	register struct walkarg *w = vw;
817 	register struct rtentry *rt = (struct rtentry *)rn;
818 	int error = 0, size;
819 	struct rt_addrinfo info;
820 
821 	if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg))
822 		return 0;
823 	bzero((caddr_t)&info, sizeof(info));
824 	dst = rt_key(rt);
825 	gate = rt->rt_gateway;
826 	netmask = rt_mask(rt);
827 	genmask = rt->rt_genmask;
828 	size = rt_msg2(RTM_GET, &info, 0, w);
829 	if (w->w_req && w->w_tmem) {
830 		register struct rt_msghdr *rtm = (struct rt_msghdr *)w->w_tmem;
831 
832 		rtm->rtm_flags = rt->rt_flags;
833 		rtm->rtm_use = rt->rt_use;
834 		rtm->rtm_rmx = rt->rt_rmx;
835 		rtm->rtm_index = rt->rt_ifp->if_index;
836 		rtm->rtm_errno = rtm->rtm_pid = rtm->rtm_seq = 0;
837 		rtm->rtm_addrs = info.rti_addrs;
838 		error = SYSCTL_OUT(w->w_req, (caddr_t)rtm, size);
839 		return (error);
840 	}
841 	return (error);
842 }
843 
844 int
845 sysctl_iflist(af, w)
846 	int	af;
847 	register struct	walkarg *w;
848 {
849 	register struct ifnet *ifp;
850 	register struct ifaddr *ifa;
851 	struct	rt_addrinfo info;
852 	int	len, error = 0;
853 
854 	bzero((caddr_t)&info, sizeof(info));
855 	for (ifp = ifnet.tqh_first; ifp; ifp = ifp->if_link.tqe_next) {
856 		if (w->w_arg && w->w_arg != ifp->if_index)
857 			continue;
858 		ifa = ifp->if_addrhead.tqh_first;
859 		ifpaddr = ifa->ifa_addr;
860 		len = rt_msg2(RTM_IFINFO, &info, (caddr_t)0, w);
861 		ifpaddr = 0;
862 		if (w->w_req && w->w_tmem) {
863 			register struct if_msghdr *ifm;
864 
865 			ifm = (struct if_msghdr *)w->w_tmem;
866 			ifm->ifm_index = ifp->if_index;
867 			ifm->ifm_flags = (u_short)ifp->if_flags;
868 			ifm->ifm_data = ifp->if_data;
869 			ifm->ifm_addrs = info.rti_addrs;
870 			error = SYSCTL_OUT(w->w_req,(caddr_t)ifm, len);
871 			if (error)
872 				return (error);
873 		}
874 		while ((ifa = ifa->ifa_link.tqe_next) != 0) {
875 			if (af && af != ifa->ifa_addr->sa_family)
876 				continue;
877 			ifaaddr = ifa->ifa_addr;
878 			netmask = ifa->ifa_netmask;
879 			brdaddr = ifa->ifa_dstaddr;
880 			len = rt_msg2(RTM_NEWADDR, &info, 0, w);
881 			if (w->w_req && w->w_tmem) {
882 				register struct ifa_msghdr *ifam;
883 
884 				ifam = (struct ifa_msghdr *)w->w_tmem;
885 				ifam->ifam_index = ifa->ifa_ifp->if_index;
886 				ifam->ifam_flags = ifa->ifa_flags;
887 				ifam->ifam_metric = ifa->ifa_metric;
888 				ifam->ifam_addrs = info.rti_addrs;
889 				error = SYSCTL_OUT(w->w_req, w->w_tmem, len);
890 				if (error)
891 					return (error);
892 			}
893 		}
894 		ifaaddr = netmask = brdaddr = 0;
895 	}
896 	return (0);
897 }
898 
899 static int
900 sysctl_rtsock SYSCTL_HANDLER_ARGS
901 {
902 	int	*name = (int *)arg1;
903 	u_int	namelen = arg2;
904 	register struct radix_node_head *rnh;
905 	int	i, s, error = EINVAL;
906 	u_char  af;
907 	struct	walkarg w;
908 
909 	name ++;
910 	namelen--;
911 	if (req->newptr)
912 		return (EPERM);
913 	if (namelen != 3)
914 		return (EINVAL);
915 	af = name[0];
916 	Bzero(&w, sizeof(w));
917 	w.w_op = name[1];
918 	w.w_arg = name[2];
919 	w.w_req = req;
920 
921 	s = splnet();
922 	switch (w.w_op) {
923 
924 	case NET_RT_DUMP:
925 	case NET_RT_FLAGS:
926 		for (i = 1; i <= AF_MAX; i++)
927 			if ((rnh = rt_tables[i]) && (af == 0 || af == i) &&
928 			    (error = rnh->rnh_walktree(rnh,
929 							sysctl_dumpentry, &w)))
930 				break;
931 		break;
932 
933 	case NET_RT_IFLIST:
934 		error = sysctl_iflist(af, &w);
935 	}
936 	splx(s);
937 	if (w.w_tmem)
938 		free(w.w_tmem, M_RTABLE);
939 	return (error);
940 }
941 
942 SYSCTL_NODE(_net, PF_ROUTE, routetable, CTLFLAG_RD, sysctl_rtsock, "");
943 
944 /*
945  * Definitions of protocols supported in the ROUTE domain.
946  */
947 
948 extern struct domain routedomain;		/* or at least forward */
949 
950 static struct protosw routesw[] = {
951 { SOCK_RAW,	&routedomain,	0,		PR_ATOMIC|PR_ADDR,
952   0,		route_output,	raw_ctlinput,	0,
953   0,
954   raw_init,	0,		0,		0,
955   &route_usrreqs
956 }
957 };
958 
959 static struct domain routedomain =
960     { PF_ROUTE, "route", route_init, 0, 0,
961       routesw, &routesw[sizeof(routesw)/sizeof(routesw[0])] };
962 
963 DOMAIN_SET(route);
964