xref: /freebsd/sys/net/rtsock.c (revision 02f2e93b60c2b91feac8f45c4c889a5a8e40d8a2)
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 1.34 1997/09/14 03:10:38 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/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 __P((int, struct rt_addrinfo *));
71 static int	rt_msg2 __P((int,
72 		    struct rt_addrinfo *, caddr_t, struct walkarg *));
73 static int	rt_xaddrs __P((caddr_t, caddr_t, struct rt_addrinfo *));
74 static int	sysctl_dumpentry __P((struct radix_node *rn, void *vw));
75 static int	sysctl_iflist __P((int af, struct walkarg *w));
76 static int	 route_output __P((struct mbuf *, struct socket *));
77 static void	 rt_setmetrics __P((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 proc *p)
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); /* XXX */
113 	if (rp == 0)
114 		return ENOBUFS;
115 	bzero(rp, sizeof *rp);
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_usrreqs.pru_attach(so, proto, p);
127 	rp = sotorawcb(so);
128 	if (error) {
129 		splx(s);
130 		free(rp, M_PCB);
131 		return error;
132 	}
133 	switch(rp->rcb_proto.sp_protocol) {
134 	case AF_INET:
135 		route_cb.ip_count++;
136 		break;
137 	case AF_IPX:
138 		route_cb.ipx_count++;
139 		break;
140 	case AF_NS:
141 		route_cb.ns_count++;
142 		break;
143 	case AF_ISO:
144 		route_cb.iso_count++;
145 		break;
146 	}
147 	rp->rcb_faddr = &route_src;
148 	route_cb.any_count++;
149 	soisconnected(so);
150 	so->so_options |= SO_USELOOPBACK;
151 	splx(s);
152 	return 0;
153 }
154 
155 static int
156 rts_bind(struct socket *so, struct sockaddr *nam, struct proc *p)
157 {
158 	int s, error;
159 	s = splnet();
160 	error = raw_usrreqs.pru_bind(so, nam, p); /* xxx just EINVAL */
161 	splx(s);
162 	return error;
163 }
164 
165 static int
166 rts_connect(struct socket *so, struct sockaddr *nam, struct proc *p)
167 {
168 	int s, error;
169 	s = splnet();
170 	error = raw_usrreqs.pru_connect(so, nam, p); /* XXX just EINVAL */
171 	splx(s);
172 	return error;
173 }
174 
175 /* pru_connect2 is EOPNOTSUPP */
176 /* pru_control is EOPNOTSUPP */
177 
178 static int
179 rts_detach(struct socket *so)
180 {
181 	struct rawcb *rp = sotorawcb(so);
182 	int s, error;
183 
184 	s = splnet();
185 	if (rp != 0) {
186 		switch(rp->rcb_proto.sp_protocol) {
187 		case AF_INET:
188 			route_cb.ip_count--;
189 			break;
190 		case AF_IPX:
191 			route_cb.ipx_count--;
192 			break;
193 		case AF_NS:
194 			route_cb.ns_count--;
195 			break;
196 		case AF_ISO:
197 			route_cb.iso_count--;
198 			break;
199 		}
200 		route_cb.any_count--;
201 	}
202 	error = raw_usrreqs.pru_detach(so);
203 	splx(s);
204 	return error;
205 }
206 
207 static int
208 rts_disconnect(struct socket *so)
209 {
210 	int s, error;
211 	s = splnet();
212 	error = raw_usrreqs.pru_disconnect(so);
213 	splx(s);
214 	return error;
215 }
216 
217 /* pru_listen is EOPNOTSUPP */
218 
219 static int
220 rts_peeraddr(struct socket *so, struct sockaddr **nam)
221 {
222 	int s, error;
223 	s = splnet();
224 	error = raw_usrreqs.pru_peeraddr(so, nam);
225 	splx(s);
226 	return error;
227 }
228 
229 /* pru_rcvd is EOPNOTSUPP */
230 /* pru_rcvoob is EOPNOTSUPP */
231 
232 static int
233 rts_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
234 	 struct mbuf *control, struct proc *p)
235 {
236 	int s, error;
237 	s = splnet();
238 	error = raw_usrreqs.pru_send(so, flags, m, nam, control, p);
239 	splx(s);
240 	return error;
241 }
242 
243 /* pru_sense is null */
244 
245 static int
246 rts_shutdown(struct socket *so)
247 {
248 	int s, error;
249 	s = splnet();
250 	error = raw_usrreqs.pru_shutdown(so);
251 	splx(s);
252 	return error;
253 }
254 
255 static int
256 rts_sockaddr(struct socket *so, struct sockaddr **nam)
257 {
258 	int s, error;
259 	s = splnet();
260 	error = raw_usrreqs.pru_sockaddr(so, nam);
261 	splx(s);
262 	return error;
263 }
264 
265 static struct pr_usrreqs route_usrreqs = {
266 	rts_abort, pru_accept_notsupp, rts_attach, rts_bind, rts_connect,
267 	pru_connect2_notsupp, pru_control_notsupp, rts_detach, rts_disconnect,
268 	pru_listen_notsupp, rts_peeraddr, pru_rcvd_notsupp, pru_rcvoob_notsupp,
269 	rts_send, pru_sense_null, rts_shutdown, rts_sockaddr,
270 	sosend, soreceive, sopoll
271 };
272 
273 /*ARGSUSED*/
274 static int
275 route_output(m, so)
276 	register struct mbuf *m;
277 	struct socket *so;
278 {
279 	register struct rt_msghdr *rtm = 0;
280 	register struct rtentry *rt = 0;
281 	struct rtentry *saved_nrt = 0;
282 	struct radix_node_head *rnh;
283 	struct rt_addrinfo info;
284 	int len, error = 0;
285 	struct ifnet *ifp = 0;
286 	struct ifaddr *ifa = 0;
287 
288 #define senderr(e) { error = e; goto flush;}
289 	if (m == 0 || ((m->m_len < sizeof(long)) &&
290 		       (m = m_pullup(m, sizeof(long))) == 0))
291 		return (ENOBUFS);
292 	if ((m->m_flags & M_PKTHDR) == 0)
293 		panic("route_output");
294 	len = m->m_pkthdr.len;
295 	if (len < sizeof(*rtm) ||
296 	    len != mtod(m, struct rt_msghdr *)->rtm_msglen) {
297 		dst = 0;
298 		senderr(EINVAL);
299 	}
300 	R_Malloc(rtm, struct rt_msghdr *, len);
301 	if (rtm == 0) {
302 		dst = 0;
303 		senderr(ENOBUFS);
304 	}
305 	m_copydata(m, 0, len, (caddr_t)rtm);
306 	if (rtm->rtm_version != RTM_VERSION) {
307 		dst = 0;
308 		senderr(EPROTONOSUPPORT);
309 	}
310 	rtm->rtm_pid = curproc->p_pid;
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 	if (dst == 0 || (dst->sa_family >= AF_MAX)
317 	    || (gate != 0 && (gate->sa_family >= AF_MAX)))
318 		senderr(EINVAL);
319 	if (genmask) {
320 		struct radix_node *t;
321 		t = rn_addmask((caddr_t)genmask, 0, 1);
322 		if (t && Bcmp(genmask, t->rn_key, *(u_char *)genmask) == 0)
323 			genmask = (struct sockaddr *)(t->rn_key);
324 		else
325 			senderr(ENOBUFS);
326 	}
327 	switch (rtm->rtm_type) {
328 
329 	case RTM_ADD:
330 		if (gate == 0)
331 			senderr(EINVAL);
332 		error = rtrequest(RTM_ADD, dst, gate, netmask,
333 					rtm->rtm_flags, &saved_nrt);
334 		if (error == 0 && saved_nrt) {
335 			rt_setmetrics(rtm->rtm_inits,
336 				&rtm->rtm_rmx, &saved_nrt->rt_rmx);
337 			saved_nrt->rt_refcnt--;
338 			saved_nrt->rt_genmask = genmask;
339 		}
340 		break;
341 
342 	case RTM_DELETE:
343 		error = rtrequest(RTM_DELETE, dst, gate, netmask,
344 				rtm->rtm_flags, &saved_nrt);
345 		if (error == 0) {
346 			if ((rt = saved_nrt))
347 				rt->rt_refcnt++;
348 			goto report;
349 		}
350 		break;
351 
352 	case RTM_GET:
353 	case RTM_CHANGE:
354 	case RTM_LOCK:
355 		if ((rnh = rt_tables[dst->sa_family]) == 0) {
356 			senderr(EAFNOSUPPORT);
357 		} else if (rt = (struct rtentry *)
358 				rnh->rnh_lookup(dst, netmask, rnh))
359 			rt->rt_refcnt++;
360 		else
361 			senderr(ESRCH);
362 		switch(rtm->rtm_type) {
363 
364 		case RTM_GET:
365 		report:
366 			dst = rt_key(rt);
367 			gate = rt->rt_gateway;
368 			netmask = rt_mask(rt);
369 			genmask = rt->rt_genmask;
370 			if (rtm->rtm_addrs & (RTA_IFP | RTA_IFA)) {
371 				ifp = rt->rt_ifp;
372 				if (ifp) {
373 					ifpaddr = ifp->if_addrhead.tqh_first->ifa_addr;
374 					ifaaddr = rt->rt_ifa->ifa_addr;
375 					rtm->rtm_index = ifp->if_index;
376 				} else {
377 					ifpaddr = 0;
378 					ifaaddr = 0;
379 			    }
380 			}
381 			len = rt_msg2(rtm->rtm_type, &info, (caddr_t)0,
382 				(struct walkarg *)0);
383 			if (len > rtm->rtm_msglen) {
384 				struct rt_msghdr *new_rtm;
385 				R_Malloc(new_rtm, struct rt_msghdr *, len);
386 				if (new_rtm == 0)
387 					senderr(ENOBUFS);
388 				Bcopy(rtm, new_rtm, rtm->rtm_msglen);
389 				Free(rtm); rtm = new_rtm;
390 			}
391 			(void)rt_msg2(rtm->rtm_type, &info, (caddr_t)rtm,
392 				(struct walkarg *)0);
393 			rtm->rtm_flags = rt->rt_flags;
394 			rtm->rtm_rmx = rt->rt_rmx;
395 			rtm->rtm_addrs = info.rti_addrs;
396 			break;
397 
398 		case RTM_CHANGE:
399 			if (gate && (error = rt_setgate(rt, rt_key(rt), gate)))
400 				senderr(error);
401 
402 			/*
403 			 * If they tried to change things but didn't specify
404 			 * the required gateway, then just use the old one.
405 			 * This can happen if the user tries to change the
406 			 * flags on the default route without changing the
407 			 * default gateway.  Changing flags still doesn't work.
408 			 */
409 			if ((rt->rt_flags & RTF_GATEWAY) && !gate)
410 				gate = rt->rt_gateway;
411 
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 			if (ifpaddr && (ifa = ifa_ifwithnet(ifpaddr)) &&
416 			    (ifp = ifa->ifa_ifp) && (ifaaddr || gate))
417 				ifa = ifaof_ifpforaddr(ifaaddr ? ifaaddr : gate,
418 							ifp);
419 			else if ((ifaaddr && (ifa = ifa_ifwithaddr(ifaaddr))) ||
420 				 (gate && (ifa = ifa_ifwithroute(rt->rt_flags,
421 							rt_key(rt), gate))))
422 				ifp = ifa->ifa_ifp;
423 			if (ifa) {
424 				register struct ifaddr *oifa = rt->rt_ifa;
425 				if (oifa != ifa) {
426 				    if (oifa && oifa->ifa_rtrequest)
427 					oifa->ifa_rtrequest(RTM_DELETE,
428 								rt, gate);
429 				    IFAFREE(rt->rt_ifa);
430 				    rt->rt_ifa = ifa;
431 				    ifa->ifa_refcnt++;
432 				    rt->rt_ifp = ifp;
433 				}
434 			}
435 			rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx,
436 					&rt->rt_rmx);
437 			if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest)
438 			       rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, gate);
439 			if (genmask)
440 				rt->rt_genmask = genmask;
441 			/*
442 			 * Fall into
443 			 */
444 		case RTM_LOCK:
445 			rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits);
446 			rt->rt_rmx.rmx_locks |=
447 				(rtm->rtm_inits & rtm->rtm_rmx.rmx_locks);
448 			break;
449 		}
450 		break;
451 
452 	default:
453 		senderr(EOPNOTSUPP);
454 	}
455 
456 flush:
457 	if (rtm) {
458 		if (error)
459 			rtm->rtm_errno = error;
460 		else
461 			rtm->rtm_flags |= RTF_DONE;
462 	}
463 	if (rt)
464 		rtfree(rt);
465     {
466 	register struct rawcb *rp = 0;
467 	/*
468 	 * Check to see if we don't want our own messages.
469 	 */
470 	if ((so->so_options & SO_USELOOPBACK) == 0) {
471 		if (route_cb.any_count <= 1) {
472 			if (rtm)
473 				Free(rtm);
474 			m_freem(m);
475 			return (error);
476 		}
477 		/* There is another listener, so construct message */
478 		rp = sotorawcb(so);
479 	}
480 	if (rtm) {
481 		m_copyback(m, 0, rtm->rtm_msglen, (caddr_t)rtm);
482 		Free(rtm);
483 	}
484 	if (rp)
485 		rp->rcb_proto.sp_family = 0; /* Avoid us */
486 	if (dst)
487 		route_proto.sp_protocol = dst->sa_family;
488 	raw_input(m, &route_proto, &route_src, &route_dst);
489 	if (rp)
490 		rp->rcb_proto.sp_family = PF_ROUTE;
491     }
492 	return (error);
493 }
494 
495 static void
496 rt_setmetrics(which, in, out)
497 	u_long which;
498 	register struct rt_metrics *in, *out;
499 {
500 #define metric(f, e) if (which & (f)) out->e = in->e;
501 	metric(RTV_RPIPE, rmx_recvpipe);
502 	metric(RTV_SPIPE, rmx_sendpipe);
503 	metric(RTV_SSTHRESH, rmx_ssthresh);
504 	metric(RTV_RTT, rmx_rtt);
505 	metric(RTV_RTTVAR, rmx_rttvar);
506 	metric(RTV_HOPCOUNT, rmx_hopcount);
507 	metric(RTV_MTU, rmx_mtu);
508 	metric(RTV_EXPIRE, rmx_expire);
509 #undef metric
510 }
511 
512 #define ROUNDUP(a) \
513 	((a) > 0 ? (1 + (((a) - 1) | (sizeof(long) - 1))) : sizeof(long))
514 #define ADVANCE(x, n) (x += ROUNDUP((n)->sa_len))
515 
516 
517 /*
518  * Extract the addresses of the passed sockaddrs.
519  * Do a little sanity checking so as to avoid bad memory references.
520  * This data is derived straight from userland.
521  */
522 static int
523 rt_xaddrs(cp, cplim, rtinfo)
524 	register caddr_t cp, cplim;
525 	register struct rt_addrinfo *rtinfo;
526 {
527 	register struct sockaddr *sa;
528 	register int i;
529 
530 	bzero(rtinfo->rti_info, sizeof(rtinfo->rti_info));
531 	for (i = 0; (i < RTAX_MAX) && (cp < cplim); i++) {
532 		if ((rtinfo->rti_addrs & (1 << i)) == 0)
533 			continue;
534 		sa = (struct sockaddr *)cp;
535 		/*
536 		 * It won't fit.
537 		 */
538 		if ( (cp + sa->sa_len) > cplim ) {
539 			return (EINVAL);
540 		}
541 
542 		/*
543 		 * there are no more.. quit now
544 		 * If there are more bits, they are in error.
545 		 * I've seen this. route(1) can evidently generate these.
546 		 * This causes kernel to core dump.
547 		 * for compatibility, If we see this, point to a safe address.
548 		 */
549 		if (sa->sa_len == 0) {
550 			rtinfo->rti_info[i] = &sa_zero;
551 			return (0); /* should be EINVAL but for compat */
552 		}
553 
554 		/* accept it */
555 		rtinfo->rti_info[i] = sa;
556 		ADVANCE(cp, sa);
557 	}
558 	return (0);
559 }
560 
561 static struct mbuf *
562 rt_msg1(type, rtinfo)
563 	int type;
564 	register struct rt_addrinfo *rtinfo;
565 {
566 	register struct rt_msghdr *rtm;
567 	register struct mbuf *m;
568 	register int i;
569 	register struct sockaddr *sa;
570 	int len, dlen;
571 
572 	m = m_gethdr(M_DONTWAIT, MT_DATA);
573 	if (m == 0)
574 		return (m);
575 	switch (type) {
576 
577 	case RTM_DELADDR:
578 	case RTM_NEWADDR:
579 		len = sizeof(struct ifa_msghdr);
580 		break;
581 
582 	case RTM_DELMADDR:
583 	case RTM_NEWMADDR:
584 		len = sizeof(struct ifma_msghdr);
585 		break;
586 
587 	case RTM_IFINFO:
588 		len = sizeof(struct if_msghdr);
589 		break;
590 
591 	default:
592 		len = sizeof(struct rt_msghdr);
593 	}
594 	if (len > MHLEN)
595 		panic("rt_msg1");
596 	m->m_pkthdr.len = m->m_len = len;
597 	m->m_pkthdr.rcvif = 0;
598 	rtm = mtod(m, struct rt_msghdr *);
599 	bzero((caddr_t)rtm, len);
600 	for (i = 0; i < RTAX_MAX; i++) {
601 		if ((sa = rtinfo->rti_info[i]) == NULL)
602 			continue;
603 		rtinfo->rti_addrs |= (1 << i);
604 		dlen = ROUNDUP(sa->sa_len);
605 		m_copyback(m, len, dlen, (caddr_t)sa);
606 		len += dlen;
607 	}
608 	if (m->m_pkthdr.len != len) {
609 		m_freem(m);
610 		return (NULL);
611 	}
612 	rtm->rtm_msglen = len;
613 	rtm->rtm_version = RTM_VERSION;
614 	rtm->rtm_type = type;
615 	return (m);
616 }
617 
618 static int
619 rt_msg2(type, rtinfo, cp, w)
620 	int type;
621 	register struct rt_addrinfo *rtinfo;
622 	caddr_t cp;
623 	struct walkarg *w;
624 {
625 	register int i;
626 	int len, dlen, second_time = 0;
627 	caddr_t cp0;
628 
629 	rtinfo->rti_addrs = 0;
630 again:
631 	switch (type) {
632 
633 	case RTM_DELADDR:
634 	case RTM_NEWADDR:
635 		len = sizeof(struct ifa_msghdr);
636 		break;
637 
638 	case RTM_IFINFO:
639 		len = sizeof(struct if_msghdr);
640 		break;
641 
642 	default:
643 		len = sizeof(struct rt_msghdr);
644 	}
645 	cp0 = cp;
646 	if (cp0)
647 		cp += len;
648 	for (i = 0; i < RTAX_MAX; i++) {
649 		register struct sockaddr *sa;
650 
651 		if ((sa = rtinfo->rti_info[i]) == 0)
652 			continue;
653 		rtinfo->rti_addrs |= (1 << i);
654 		dlen = ROUNDUP(sa->sa_len);
655 		if (cp) {
656 			bcopy((caddr_t)sa, cp, (unsigned)dlen);
657 			cp += dlen;
658 		}
659 		len += dlen;
660 	}
661 	if (cp == 0 && w != NULL && !second_time) {
662 		register struct walkarg *rw = w;
663 
664 		if (rw->w_req) {
665 			if (rw->w_tmemsize < len) {
666 				if (rw->w_tmem)
667 					free(rw->w_tmem, M_RTABLE);
668 				rw->w_tmem = (caddr_t)
669 					malloc(len, M_RTABLE, M_NOWAIT);
670 				if (rw->w_tmem)
671 					rw->w_tmemsize = len;
672 			}
673 			if (rw->w_tmem) {
674 				cp = rw->w_tmem;
675 				second_time = 1;
676 				goto again;
677 			}
678 		}
679 	}
680 	if (cp) {
681 		register struct rt_msghdr *rtm = (struct rt_msghdr *)cp0;
682 
683 		rtm->rtm_version = RTM_VERSION;
684 		rtm->rtm_type = type;
685 		rtm->rtm_msglen = len;
686 	}
687 	return (len);
688 }
689 
690 /*
691  * This routine is called to generate a message from the routing
692  * socket indicating that a redirect has occured, a routing lookup
693  * has failed, or that a protocol has detected timeouts to a particular
694  * destination.
695  */
696 void
697 rt_missmsg(type, rtinfo, flags, error)
698 	int type, flags, error;
699 	register struct rt_addrinfo *rtinfo;
700 {
701 	register struct rt_msghdr *rtm;
702 	register struct mbuf *m;
703 	struct sockaddr *sa = rtinfo->rti_info[RTAX_DST];
704 
705 	if (route_cb.any_count == 0)
706 		return;
707 	m = rt_msg1(type, rtinfo);
708 	if (m == 0)
709 		return;
710 	rtm = mtod(m, struct rt_msghdr *);
711 	rtm->rtm_flags = RTF_DONE | flags;
712 	rtm->rtm_errno = error;
713 	rtm->rtm_addrs = rtinfo->rti_addrs;
714 	route_proto.sp_protocol = sa ? sa->sa_family : 0;
715 	raw_input(m, &route_proto, &route_src, &route_dst);
716 }
717 
718 /*
719  * This routine is called to generate a message from the routing
720  * socket indicating that the status of a network interface has changed.
721  */
722 void
723 rt_ifmsg(ifp)
724 	register struct ifnet *ifp;
725 {
726 	register struct if_msghdr *ifm;
727 	struct mbuf *m;
728 	struct rt_addrinfo info;
729 
730 	if (route_cb.any_count == 0)
731 		return;
732 	bzero((caddr_t)&info, sizeof(info));
733 	m = rt_msg1(RTM_IFINFO, &info);
734 	if (m == 0)
735 		return;
736 	ifm = mtod(m, struct if_msghdr *);
737 	ifm->ifm_index = ifp->if_index;
738 	ifm->ifm_flags = (u_short)ifp->if_flags;
739 	ifm->ifm_data = ifp->if_data;
740 	ifm->ifm_addrs = 0;
741 	route_proto.sp_protocol = 0;
742 	raw_input(m, &route_proto, &route_src, &route_dst);
743 }
744 
745 /*
746  * This is called to generate messages from the routing socket
747  * indicating a network interface has had addresses associated with it.
748  * if we ever reverse the logic and replace messages TO the routing
749  * socket indicate a request to configure interfaces, then it will
750  * be unnecessary as the routing socket will automatically generate
751  * copies of it.
752  */
753 void
754 rt_newaddrmsg(cmd, ifa, error, rt)
755 	int cmd, error;
756 	register struct ifaddr *ifa;
757 	register struct rtentry *rt;
758 {
759 	struct rt_addrinfo info;
760 	struct sockaddr *sa = 0;
761 	int pass;
762 	struct mbuf *m = 0;
763 	struct ifnet *ifp = ifa->ifa_ifp;
764 
765 	if (route_cb.any_count == 0)
766 		return;
767 	for (pass = 1; pass < 3; pass++) {
768 		bzero((caddr_t)&info, sizeof(info));
769 		if ((cmd == RTM_ADD && pass == 1) ||
770 		    (cmd == RTM_DELETE && pass == 2)) {
771 			register struct ifa_msghdr *ifam;
772 			int ncmd = cmd == RTM_ADD ? RTM_NEWADDR : RTM_DELADDR;
773 
774 			ifaaddr = sa = ifa->ifa_addr;
775 			ifpaddr = ifp->if_addrhead.tqh_first->ifa_addr;
776 			netmask = ifa->ifa_netmask;
777 			brdaddr = ifa->ifa_dstaddr;
778 			if ((m = rt_msg1(ncmd, &info)) == NULL)
779 				continue;
780 			ifam = mtod(m, struct ifa_msghdr *);
781 			ifam->ifam_index = ifp->if_index;
782 			ifam->ifam_metric = ifa->ifa_metric;
783 			ifam->ifam_flags = ifa->ifa_flags;
784 			ifam->ifam_addrs = info.rti_addrs;
785 		}
786 		if ((cmd == RTM_ADD && pass == 2) ||
787 		    (cmd == RTM_DELETE && pass == 1)) {
788 			register struct rt_msghdr *rtm;
789 
790 			if (rt == 0)
791 				continue;
792 			netmask = rt_mask(rt);
793 			dst = sa = rt_key(rt);
794 			gate = rt->rt_gateway;
795 			if ((m = rt_msg1(cmd, &info)) == NULL)
796 				continue;
797 			rtm = mtod(m, struct rt_msghdr *);
798 			rtm->rtm_index = ifp->if_index;
799 			rtm->rtm_flags |= rt->rt_flags;
800 			rtm->rtm_errno = error;
801 			rtm->rtm_addrs = info.rti_addrs;
802 		}
803 		route_proto.sp_protocol = sa ? sa->sa_family : 0;
804 		raw_input(m, &route_proto, &route_src, &route_dst);
805 	}
806 }
807 
808 /*
809  * This is the analogue to the rt_newaddrmsg which performs the same
810  * function but for multicast group memberhips.  This is easier since
811  * there is no route state to worry about.
812  */
813 void
814 rt_newmaddrmsg(cmd, ifma)
815 	int cmd;
816 	struct ifmultiaddr *ifma;
817 {
818 	struct rt_addrinfo info;
819 	struct mbuf *m = 0;
820 	struct ifnet *ifp = ifma->ifma_ifp;
821 	struct ifma_msghdr *ifmam;
822 
823 	if (route_cb.any_count == 0)
824 		return;
825 
826 	bzero((caddr_t)&info, sizeof(info));
827 	ifaaddr = ifma->ifma_addr;
828 	ifpaddr = ifp->if_addrhead.tqh_first->ifa_addr;
829 	/*
830 	 * If a link-layer address is present, present it as a ``gateway''
831 	 * (similarly to how ARP entries, e.g., are presented).
832 	 */
833 	gate = ifma->ifma_lladdr;
834 	if ((m = rt_msg1(cmd, &info)) == NULL)
835 		return;
836 	ifmam = mtod(m, struct ifma_msghdr *);
837 	ifmam->ifmam_index = ifp->if_index;
838 	ifmam->ifmam_addrs = info.rti_addrs;
839 	route_proto.sp_protocol = ifma->ifma_addr->sa_family;
840 	raw_input(m, &route_proto, &route_src, &route_dst);
841 }
842 
843 /*
844  * This is used in dumping the kernel table via sysctl().
845  */
846 int
847 sysctl_dumpentry(rn, vw)
848 	struct radix_node *rn;
849 	void *vw;
850 {
851 	register struct walkarg *w = vw;
852 	register struct rtentry *rt = (struct rtentry *)rn;
853 	int error = 0, size;
854 	struct rt_addrinfo info;
855 
856 	if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg))
857 		return 0;
858 	bzero((caddr_t)&info, sizeof(info));
859 	dst = rt_key(rt);
860 	gate = rt->rt_gateway;
861 	netmask = rt_mask(rt);
862 	genmask = rt->rt_genmask;
863 	size = rt_msg2(RTM_GET, &info, 0, w);
864 	if (w->w_req && w->w_tmem) {
865 		register struct rt_msghdr *rtm = (struct rt_msghdr *)w->w_tmem;
866 
867 		rtm->rtm_flags = rt->rt_flags;
868 		rtm->rtm_use = rt->rt_use;
869 		rtm->rtm_rmx = rt->rt_rmx;
870 		rtm->rtm_index = rt->rt_ifp->if_index;
871 		rtm->rtm_errno = rtm->rtm_pid = rtm->rtm_seq = 0;
872 		rtm->rtm_addrs = info.rti_addrs;
873 		error = SYSCTL_OUT(w->w_req, (caddr_t)rtm, size);
874 		return (error);
875 	}
876 	return (error);
877 }
878 
879 int
880 sysctl_iflist(af, w)
881 	int	af;
882 	register struct	walkarg *w;
883 {
884 	register struct ifnet *ifp;
885 	register struct ifaddr *ifa;
886 	struct	rt_addrinfo info;
887 	int	len, error = 0;
888 
889 	bzero((caddr_t)&info, sizeof(info));
890 	for (ifp = ifnet.tqh_first; ifp; ifp = ifp->if_link.tqe_next) {
891 		if (w->w_arg && w->w_arg != ifp->if_index)
892 			continue;
893 		ifa = ifp->if_addrhead.tqh_first;
894 		ifpaddr = ifa->ifa_addr;
895 		len = rt_msg2(RTM_IFINFO, &info, (caddr_t)0, w);
896 		ifpaddr = 0;
897 		if (w->w_req && w->w_tmem) {
898 			register struct if_msghdr *ifm;
899 
900 			ifm = (struct if_msghdr *)w->w_tmem;
901 			ifm->ifm_index = ifp->if_index;
902 			ifm->ifm_flags = (u_short)ifp->if_flags;
903 			ifm->ifm_data = ifp->if_data;
904 			ifm->ifm_addrs = info.rti_addrs;
905 			error = SYSCTL_OUT(w->w_req,(caddr_t)ifm, len);
906 			if (error)
907 				return (error);
908 		}
909 		while ((ifa = ifa->ifa_link.tqe_next) != 0) {
910 			if (af && af != ifa->ifa_addr->sa_family)
911 				continue;
912 			ifaaddr = ifa->ifa_addr;
913 			netmask = ifa->ifa_netmask;
914 			brdaddr = ifa->ifa_dstaddr;
915 			len = rt_msg2(RTM_NEWADDR, &info, 0, w);
916 			if (w->w_req && w->w_tmem) {
917 				register struct ifa_msghdr *ifam;
918 
919 				ifam = (struct ifa_msghdr *)w->w_tmem;
920 				ifam->ifam_index = ifa->ifa_ifp->if_index;
921 				ifam->ifam_flags = ifa->ifa_flags;
922 				ifam->ifam_metric = ifa->ifa_metric;
923 				ifam->ifam_addrs = info.rti_addrs;
924 				error = SYSCTL_OUT(w->w_req, w->w_tmem, len);
925 				if (error)
926 					return (error);
927 			}
928 		}
929 		ifaaddr = netmask = brdaddr = 0;
930 	}
931 	return (0);
932 }
933 
934 static int
935 sysctl_rtsock SYSCTL_HANDLER_ARGS
936 {
937 	int	*name = (int *)arg1;
938 	u_int	namelen = arg2;
939 	register struct radix_node_head *rnh;
940 	int	i, s, error = EINVAL;
941 	u_char  af;
942 	struct	walkarg w;
943 
944 	name ++;
945 	namelen--;
946 	if (req->newptr)
947 		return (EPERM);
948 	if (namelen != 3)
949 		return (EINVAL);
950 	af = name[0];
951 	Bzero(&w, sizeof(w));
952 	w.w_op = name[1];
953 	w.w_arg = name[2];
954 	w.w_req = req;
955 
956 	s = splnet();
957 	switch (w.w_op) {
958 
959 	case NET_RT_DUMP:
960 	case NET_RT_FLAGS:
961 		for (i = 1; i <= AF_MAX; i++)
962 			if ((rnh = rt_tables[i]) && (af == 0 || af == i) &&
963 			    (error = rnh->rnh_walktree(rnh,
964 							sysctl_dumpentry, &w)))
965 				break;
966 		break;
967 
968 	case NET_RT_IFLIST:
969 		error = sysctl_iflist(af, &w);
970 	}
971 	splx(s);
972 	if (w.w_tmem)
973 		free(w.w_tmem, M_RTABLE);
974 	return (error);
975 }
976 
977 SYSCTL_NODE(_net, PF_ROUTE, routetable, CTLFLAG_RD, sysctl_rtsock, "");
978 
979 /*
980  * Definitions of protocols supported in the ROUTE domain.
981  */
982 
983 extern struct domain routedomain;		/* or at least forward */
984 
985 static struct protosw routesw[] = {
986 { SOCK_RAW,	&routedomain,	0,		PR_ATOMIC|PR_ADDR,
987   0,		route_output,	raw_ctlinput,	0,
988   0,
989   raw_init,	0,		0,		0,
990   &route_usrreqs
991 }
992 };
993 
994 static struct domain routedomain =
995     { PF_ROUTE, "route", route_init, 0, 0,
996       routesw, &routesw[sizeof(routesw)/sizeof(routesw[0])] };
997 
998 DOMAIN_SET(route);
999