xref: /freebsd/sys/net/rtsock.c (revision ee2ea5ceafed78a5bd9810beb9e3ca927180c226)
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_locked(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 			if ((rt = saved_nrt))
360 				rt->rt_refcnt++;
361 			goto report;
362 		}
363 		break;
364 
365 	case RTM_GET:
366 	case RTM_CHANGE:
367 	case RTM_LOCK:
368 		if ((rnh = rt_tables[dst->sa_family]) == 0) {
369 			senderr(EAFNOSUPPORT);
370 		} else if ((rt = (struct rtentry *)
371 				rnh->rnh_lookup(dst, netmask, rnh)) != NULL)
372 			rt->rt_refcnt++;
373 		else
374 			senderr(ESRCH);
375 		switch(rtm->rtm_type) {
376 
377 		case RTM_GET:
378 		report:
379 			dst = rt_key(rt);
380 			gate = rt->rt_gateway;
381 			netmask = rt_mask(rt);
382 			genmask = rt->rt_genmask;
383 			if (rtm->rtm_addrs & (RTA_IFP | RTA_IFA)) {
384 				ifp = rt->rt_ifp;
385 				if (ifp) {
386 					ifpaddr = TAILQ_FIRST(&ifp->if_addrhead)->ifa_addr;
387 					ifaaddr = rt->rt_ifa->ifa_addr;
388 					if (ifp->if_flags & IFF_POINTOPOINT)
389 						brdaddr = rt->rt_ifa->ifa_dstaddr;
390 					rtm->rtm_index = ifp->if_index;
391 				} else {
392 					ifpaddr = 0;
393 					ifaaddr = 0;
394 			    }
395 			}
396 			len = rt_msg2(rtm->rtm_type, &info, (caddr_t)0,
397 				(struct walkarg *)0);
398 			if (len > rtm->rtm_msglen) {
399 				struct rt_msghdr *new_rtm;
400 				R_Malloc(new_rtm, struct rt_msghdr *, len);
401 				if (new_rtm == 0)
402 					senderr(ENOBUFS);
403 				Bcopy(rtm, new_rtm, rtm->rtm_msglen);
404 				Free(rtm); rtm = new_rtm;
405 			}
406 			(void)rt_msg2(rtm->rtm_type, &info, (caddr_t)rtm,
407 				(struct walkarg *)0);
408 			rtm->rtm_flags = rt->rt_flags;
409 			rtm->rtm_rmx = rt->rt_rmx;
410 			rtm->rtm_addrs = info.rti_addrs;
411 			break;
412 
413 		case RTM_CHANGE:
414 			/* new gateway could require new ifaddr, ifp;
415 			   flags may also be different; ifp may be specified
416 			   by ll sockaddr when protocol address is ambiguous */
417 #define	equal(a1, a2) (bcmp((caddr_t)(a1), (caddr_t)(a2), (a1)->sa_len) == 0)
418 			if ((rt->rt_flags & RTF_GATEWAY && gate != NULL) ||
419 			    ifpaddr != NULL ||
420 			    (ifaaddr != NULL &&
421 			    !equal(ifaaddr, rt->rt_ifa->ifa_addr))) {
422 				if ((error = rt_getifa(&info)) != 0)
423 					senderr(error);
424 			}
425 			if (gate != NULL &&
426 			    (error = rt_setgate(rt, rt_key(rt), gate)) != 0)
427 				senderr(error);
428 			if ((ifa = info.rti_ifa) != NULL) {
429 				register struct ifaddr *oifa = rt->rt_ifa;
430 				if (oifa != ifa) {
431 				    if (oifa && oifa->ifa_rtrequest)
432 					oifa->ifa_rtrequest(RTM_DELETE, rt,
433 					    &info);
434 				    IFAFREE(rt->rt_ifa);
435 				    rt->rt_ifa = ifa;
436 				    ifa->ifa_refcnt++;
437 				    rt->rt_ifp = info.rti_ifp;
438 				}
439 			}
440 			rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx,
441 					&rt->rt_rmx);
442 			if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest)
443 			       rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, &info);
444 			if (genmask)
445 				rt->rt_genmask = genmask;
446 			/*
447 			 * Fall into
448 			 */
449 		case RTM_LOCK:
450 			rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits);
451 			rt->rt_rmx.rmx_locks |=
452 				(rtm->rtm_inits & rtm->rtm_rmx.rmx_locks);
453 			break;
454 		}
455 		break;
456 
457 	default:
458 		senderr(EOPNOTSUPP);
459 	}
460 
461 flush:
462 	if (rtm) {
463 		if (error)
464 			rtm->rtm_errno = error;
465 		else
466 			rtm->rtm_flags |= RTF_DONE;
467 	}
468 	if (rt)
469 		rtfree(rt);
470     {
471 	register struct rawcb *rp = 0;
472 	/*
473 	 * Check to see if we don't want our own messages.
474 	 */
475 	if ((so->so_options & SO_USELOOPBACK) == 0) {
476 		if (route_cb.any_count <= 1) {
477 			if (rtm)
478 				Free(rtm);
479 			m_freem(m);
480 			return (error);
481 		}
482 		/* There is another listener, so construct message */
483 		rp = sotorawcb(so);
484 	}
485 	if (rtm) {
486 		m_copyback(m, 0, rtm->rtm_msglen, (caddr_t)rtm);
487 		if (m->m_pkthdr.len < rtm->rtm_msglen) {
488 			m_freem(m);
489 			m = NULL;
490 		} else if (m->m_pkthdr.len > rtm->rtm_msglen)
491 			m_adj(m, rtm->rtm_msglen - m->m_pkthdr.len);
492 		Free(rtm);
493 	}
494 	if (rp)
495 		rp->rcb_proto.sp_family = 0; /* Avoid us */
496 	if (dst)
497 		route_proto.sp_protocol = dst->sa_family;
498 	if (m)
499 		raw_input(m, &route_proto, &route_src, &route_dst);
500 	if (rp)
501 		rp->rcb_proto.sp_family = PF_ROUTE;
502     }
503 	return (error);
504 }
505 
506 static void
507 rt_setmetrics(which, in, out)
508 	u_long which;
509 	register struct rt_metrics *in, *out;
510 {
511 #define metric(f, e) if (which & (f)) out->e = in->e;
512 	metric(RTV_RPIPE, rmx_recvpipe);
513 	metric(RTV_SPIPE, rmx_sendpipe);
514 	metric(RTV_SSTHRESH, rmx_ssthresh);
515 	metric(RTV_RTT, rmx_rtt);
516 	metric(RTV_RTTVAR, rmx_rttvar);
517 	metric(RTV_HOPCOUNT, rmx_hopcount);
518 	metric(RTV_MTU, rmx_mtu);
519 	metric(RTV_EXPIRE, rmx_expire);
520 #undef metric
521 }
522 
523 #define ROUNDUP(a) \
524 	((a) > 0 ? (1 + (((a) - 1) | (sizeof(long) - 1))) : sizeof(long))
525 #define ADVANCE(x, n) (x += ROUNDUP((n)->sa_len))
526 
527 
528 /*
529  * Extract the addresses of the passed sockaddrs.
530  * Do a little sanity checking so as to avoid bad memory references.
531  * This data is derived straight from userland.
532  */
533 static int
534 rt_xaddrs(cp, cplim, rtinfo)
535 	register caddr_t cp, cplim;
536 	register struct rt_addrinfo *rtinfo;
537 {
538 	register struct sockaddr *sa;
539 	register int i;
540 
541 	for (i = 0; (i < RTAX_MAX) && (cp < cplim); i++) {
542 		if ((rtinfo->rti_addrs & (1 << i)) == 0)
543 			continue;
544 		sa = (struct sockaddr *)cp;
545 		/*
546 		 * It won't fit.
547 		 */
548 		if ( (cp + sa->sa_len) > cplim ) {
549 			return (EINVAL);
550 		}
551 
552 		/*
553 		 * there are no more.. quit now
554 		 * If there are more bits, they are in error.
555 		 * I've seen this. route(1) can evidently generate these.
556 		 * This causes kernel to core dump.
557 		 * for compatibility, If we see this, point to a safe address.
558 		 */
559 		if (sa->sa_len == 0) {
560 			rtinfo->rti_info[i] = &sa_zero;
561 			return (0); /* should be EINVAL but for compat */
562 		}
563 
564 		/* accept it */
565 		rtinfo->rti_info[i] = sa;
566 		ADVANCE(cp, sa);
567 	}
568 	return (0);
569 }
570 
571 static struct mbuf *
572 rt_msg1(type, rtinfo)
573 	int type;
574 	register struct rt_addrinfo *rtinfo;
575 {
576 	register struct rt_msghdr *rtm;
577 	register struct mbuf *m;
578 	register int i;
579 	register struct sockaddr *sa;
580 	int len, dlen;
581 
582 	switch (type) {
583 
584 	case RTM_DELADDR:
585 	case RTM_NEWADDR:
586 		len = sizeof(struct ifa_msghdr);
587 		break;
588 
589 	case RTM_DELMADDR:
590 	case RTM_NEWMADDR:
591 		len = sizeof(struct ifma_msghdr);
592 		break;
593 
594 	case RTM_IFINFO:
595 		len = sizeof(struct if_msghdr);
596 		break;
597 
598 	case RTM_IFANNOUNCE:
599 		len = sizeof(struct if_announcemsghdr);
600 		break;
601 
602 	default:
603 		len = sizeof(struct rt_msghdr);
604 	}
605 	if (len > MCLBYTES)
606 		panic("rt_msg1");
607 	m = m_gethdr(M_DONTWAIT, MT_DATA);
608 	if (m && len > MHLEN) {
609 		MCLGET(m, M_DONTWAIT);
610 		if ((m->m_flags & M_EXT) == 0) {
611 			m_free(m);
612 			m = NULL;
613 		}
614 	}
615 	if (m == 0)
616 		return (m);
617 	m->m_pkthdr.len = m->m_len = len;
618 	m->m_pkthdr.rcvif = 0;
619 	rtm = mtod(m, struct rt_msghdr *);
620 	bzero((caddr_t)rtm, len);
621 	for (i = 0; i < RTAX_MAX; i++) {
622 		if ((sa = rtinfo->rti_info[i]) == NULL)
623 			continue;
624 		rtinfo->rti_addrs |= (1 << i);
625 		dlen = ROUNDUP(sa->sa_len);
626 		m_copyback(m, len, dlen, (caddr_t)sa);
627 		len += dlen;
628 	}
629 	if (m->m_pkthdr.len != len) {
630 		m_freem(m);
631 		return (NULL);
632 	}
633 	rtm->rtm_msglen = len;
634 	rtm->rtm_version = RTM_VERSION;
635 	rtm->rtm_type = type;
636 	return (m);
637 }
638 
639 static int
640 rt_msg2(type, rtinfo, cp, w)
641 	int type;
642 	register struct rt_addrinfo *rtinfo;
643 	caddr_t cp;
644 	struct walkarg *w;
645 {
646 	register int i;
647 	int len, dlen, second_time = 0;
648 	caddr_t cp0;
649 
650 	rtinfo->rti_addrs = 0;
651 again:
652 	switch (type) {
653 
654 	case RTM_DELADDR:
655 	case RTM_NEWADDR:
656 		len = sizeof(struct ifa_msghdr);
657 		break;
658 
659 	case RTM_IFINFO:
660 		len = sizeof(struct if_msghdr);
661 		break;
662 
663 	default:
664 		len = sizeof(struct rt_msghdr);
665 	}
666 	cp0 = cp;
667 	if (cp0)
668 		cp += len;
669 	for (i = 0; i < RTAX_MAX; i++) {
670 		register struct sockaddr *sa;
671 
672 		if ((sa = rtinfo->rti_info[i]) == 0)
673 			continue;
674 		rtinfo->rti_addrs |= (1 << i);
675 		dlen = ROUNDUP(sa->sa_len);
676 		if (cp) {
677 			bcopy((caddr_t)sa, cp, (unsigned)dlen);
678 			cp += dlen;
679 		}
680 		len += dlen;
681 	}
682 	len = ALIGN(len);
683 	if (cp == 0 && w != NULL && !second_time) {
684 		register struct walkarg *rw = w;
685 
686 		if (rw->w_req) {
687 			if (rw->w_tmemsize < len) {
688 				if (rw->w_tmem)
689 					free(rw->w_tmem, M_RTABLE);
690 				rw->w_tmem = (caddr_t)
691 					malloc(len, M_RTABLE, M_NOWAIT);
692 				if (rw->w_tmem)
693 					rw->w_tmemsize = len;
694 			}
695 			if (rw->w_tmem) {
696 				cp = rw->w_tmem;
697 				second_time = 1;
698 				goto again;
699 			}
700 		}
701 	}
702 	if (cp) {
703 		register struct rt_msghdr *rtm = (struct rt_msghdr *)cp0;
704 
705 		rtm->rtm_version = RTM_VERSION;
706 		rtm->rtm_type = type;
707 		rtm->rtm_msglen = len;
708 	}
709 	return (len);
710 }
711 
712 /*
713  * This routine is called to generate a message from the routing
714  * socket indicating that a redirect has occured, a routing lookup
715  * has failed, or that a protocol has detected timeouts to a particular
716  * destination.
717  */
718 void
719 rt_missmsg(type, rtinfo, flags, error)
720 	int type, flags, error;
721 	register struct rt_addrinfo *rtinfo;
722 {
723 	register struct rt_msghdr *rtm;
724 	register struct mbuf *m;
725 	struct sockaddr *sa = rtinfo->rti_info[RTAX_DST];
726 
727 	if (route_cb.any_count == 0)
728 		return;
729 	m = rt_msg1(type, rtinfo);
730 	if (m == 0)
731 		return;
732 	rtm = mtod(m, struct rt_msghdr *);
733 	rtm->rtm_flags = RTF_DONE | flags;
734 	rtm->rtm_errno = error;
735 	rtm->rtm_addrs = rtinfo->rti_addrs;
736 	route_proto.sp_protocol = sa ? sa->sa_family : 0;
737 	raw_input(m, &route_proto, &route_src, &route_dst);
738 }
739 
740 /*
741  * This routine is called to generate a message from the routing
742  * socket indicating that the status of a network interface has changed.
743  */
744 void
745 rt_ifmsg(ifp)
746 	register struct ifnet *ifp;
747 {
748 	register struct if_msghdr *ifm;
749 	struct mbuf *m;
750 	struct rt_addrinfo info;
751 
752 	if (route_cb.any_count == 0)
753 		return;
754 	bzero((caddr_t)&info, sizeof(info));
755 	m = rt_msg1(RTM_IFINFO, &info);
756 	if (m == 0)
757 		return;
758 	ifm = mtod(m, struct if_msghdr *);
759 	ifm->ifm_index = ifp->if_index;
760 	ifm->ifm_flags = (u_short)ifp->if_flags;
761 	ifm->ifm_data = ifp->if_data;
762 	ifm->ifm_addrs = 0;
763 	route_proto.sp_protocol = 0;
764 	raw_input(m, &route_proto, &route_src, &route_dst);
765 }
766 
767 /*
768  * This is called to generate messages from the routing socket
769  * indicating a network interface has had addresses associated with it.
770  * if we ever reverse the logic and replace messages TO the routing
771  * socket indicate a request to configure interfaces, then it will
772  * be unnecessary as the routing socket will automatically generate
773  * copies of it.
774  */
775 void
776 rt_newaddrmsg(cmd, ifa, error, rt)
777 	int cmd, error;
778 	register struct ifaddr *ifa;
779 	register struct rtentry *rt;
780 {
781 	struct rt_addrinfo info;
782 	struct sockaddr *sa = 0;
783 	int pass;
784 	struct mbuf *m = 0;
785 	struct ifnet *ifp = ifa->ifa_ifp;
786 
787 	if (route_cb.any_count == 0)
788 		return;
789 	for (pass = 1; pass < 3; pass++) {
790 		bzero((caddr_t)&info, sizeof(info));
791 		if ((cmd == RTM_ADD && pass == 1) ||
792 		    (cmd == RTM_DELETE && pass == 2)) {
793 			register struct ifa_msghdr *ifam;
794 			int ncmd = cmd == RTM_ADD ? RTM_NEWADDR : RTM_DELADDR;
795 
796 			ifaaddr = sa = ifa->ifa_addr;
797 			ifpaddr = TAILQ_FIRST(&ifp->if_addrhead)->ifa_addr;
798 			netmask = ifa->ifa_netmask;
799 			brdaddr = ifa->ifa_dstaddr;
800 			if ((m = rt_msg1(ncmd, &info)) == NULL)
801 				continue;
802 			ifam = mtod(m, struct ifa_msghdr *);
803 			ifam->ifam_index = ifp->if_index;
804 			ifam->ifam_metric = ifa->ifa_metric;
805 			ifam->ifam_flags = ifa->ifa_flags;
806 			ifam->ifam_addrs = info.rti_addrs;
807 		}
808 		if ((cmd == RTM_ADD && pass == 2) ||
809 		    (cmd == RTM_DELETE && pass == 1)) {
810 			register struct rt_msghdr *rtm;
811 
812 			if (rt == 0)
813 				continue;
814 			netmask = rt_mask(rt);
815 			dst = sa = rt_key(rt);
816 			gate = rt->rt_gateway;
817 			if ((m = rt_msg1(cmd, &info)) == NULL)
818 				continue;
819 			rtm = mtod(m, struct rt_msghdr *);
820 			rtm->rtm_index = ifp->if_index;
821 			rtm->rtm_flags |= rt->rt_flags;
822 			rtm->rtm_errno = error;
823 			rtm->rtm_addrs = info.rti_addrs;
824 		}
825 		route_proto.sp_protocol = sa ? sa->sa_family : 0;
826 		raw_input(m, &route_proto, &route_src, &route_dst);
827 	}
828 }
829 
830 /*
831  * This is the analogue to the rt_newaddrmsg which performs the same
832  * function but for multicast group memberhips.  This is easier since
833  * there is no route state to worry about.
834  */
835 void
836 rt_newmaddrmsg(cmd, ifma)
837 	int cmd;
838 	struct ifmultiaddr *ifma;
839 {
840 	struct rt_addrinfo info;
841 	struct mbuf *m = 0;
842 	struct ifnet *ifp = ifma->ifma_ifp;
843 	struct ifma_msghdr *ifmam;
844 
845 	if (route_cb.any_count == 0)
846 		return;
847 
848 	bzero((caddr_t)&info, sizeof(info));
849 	ifaaddr = ifma->ifma_addr;
850 	if (ifp && TAILQ_FIRST(&ifp->if_addrhead))
851 		ifpaddr = TAILQ_FIRST(&ifp->if_addrhead)->ifa_addr;
852 	else
853 		ifpaddr = NULL;
854 	/*
855 	 * If a link-layer address is present, present it as a ``gateway''
856 	 * (similarly to how ARP entries, e.g., are presented).
857 	 */
858 	gate = ifma->ifma_lladdr;
859 	if ((m = rt_msg1(cmd, &info)) == NULL)
860 		return;
861 	ifmam = mtod(m, struct ifma_msghdr *);
862 	ifmam->ifmam_index = ifp->if_index;
863 	ifmam->ifmam_addrs = info.rti_addrs;
864 	route_proto.sp_protocol = ifma->ifma_addr->sa_family;
865 	raw_input(m, &route_proto, &route_src, &route_dst);
866 }
867 
868 /*
869  * This is called to generate routing socket messages indicating
870  * network interface arrival and departure.
871  */
872 void
873 rt_ifannouncemsg(ifp, what)
874 	struct ifnet *ifp;
875 	int what;
876 {
877 	struct if_announcemsghdr *ifan;
878 	struct mbuf *m;
879 	struct rt_addrinfo info;
880 
881 	if (route_cb.any_count == 0)
882 		return;
883 	bzero((caddr_t)&info, sizeof(info));
884 	m = rt_msg1(RTM_IFANNOUNCE, &info);
885 	if (m == NULL)
886 		return;
887 	ifan = mtod(m, struct if_announcemsghdr *);
888 	ifan->ifan_index = ifp->if_index;
889 	snprintf(ifan->ifan_name, sizeof(ifan->ifan_name),
890 	    "%s%d", ifp->if_name, ifp->if_unit);
891 	ifan->ifan_what = what;
892 	route_proto.sp_protocol = 0;
893 	raw_input(m, &route_proto, &route_src, &route_dst);
894  }
895 
896 /*
897  * This is used in dumping the kernel table via sysctl().
898  */
899 int
900 sysctl_dumpentry(rn, vw)
901 	struct radix_node *rn;
902 	void *vw;
903 {
904 	register struct walkarg *w = vw;
905 	register struct rtentry *rt = (struct rtentry *)rn;
906 	int error = 0, size;
907 	struct rt_addrinfo info;
908 
909 	if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg))
910 		return 0;
911 	bzero((caddr_t)&info, sizeof(info));
912 	dst = rt_key(rt);
913 	gate = rt->rt_gateway;
914 	netmask = rt_mask(rt);
915 	genmask = rt->rt_genmask;
916 	if (rt->rt_ifp) {
917 		ifpaddr = TAILQ_FIRST(&rt->rt_ifp->if_addrhead)->ifa_addr;
918 		ifaaddr = rt->rt_ifa->ifa_addr;
919 		if (rt->rt_ifp->if_flags & IFF_POINTOPOINT)
920 			brdaddr = rt->rt_ifa->ifa_dstaddr;
921 	}
922 	size = rt_msg2(RTM_GET, &info, 0, w);
923 	if (w->w_req && w->w_tmem) {
924 		register struct rt_msghdr *rtm = (struct rt_msghdr *)w->w_tmem;
925 
926 		rtm->rtm_flags = rt->rt_flags;
927 		rtm->rtm_use = rt->rt_use;
928 		rtm->rtm_rmx = rt->rt_rmx;
929 		rtm->rtm_index = rt->rt_ifp->if_index;
930 		rtm->rtm_errno = rtm->rtm_pid = rtm->rtm_seq = 0;
931 		rtm->rtm_addrs = info.rti_addrs;
932 		error = SYSCTL_OUT(w->w_req, (caddr_t)rtm, size);
933 		return (error);
934 	}
935 	return (error);
936 }
937 
938 int
939 sysctl_iflist(af, w)
940 	int	af;
941 	register struct	walkarg *w;
942 {
943 	register struct ifnet *ifp;
944 	register struct ifaddr *ifa;
945 	struct	rt_addrinfo info;
946 	int	len, error = 0;
947 
948 	bzero((caddr_t)&info, sizeof(info));
949 	TAILQ_FOREACH(ifp, &ifnet, if_link) {
950 		if (w->w_arg && w->w_arg != ifp->if_index)
951 			continue;
952 		ifa = TAILQ_FIRST(&ifp->if_addrhead);
953 		ifpaddr = ifa->ifa_addr;
954 		len = rt_msg2(RTM_IFINFO, &info, (caddr_t)0, w);
955 		ifpaddr = 0;
956 		if (w->w_req && w->w_tmem) {
957 			register struct if_msghdr *ifm;
958 
959 			ifm = (struct if_msghdr *)w->w_tmem;
960 			ifm->ifm_index = ifp->if_index;
961 			ifm->ifm_flags = (u_short)ifp->if_flags;
962 			ifm->ifm_data = ifp->if_data;
963 			ifm->ifm_addrs = info.rti_addrs;
964 			error = SYSCTL_OUT(w->w_req,(caddr_t)ifm, len);
965 			if (error)
966 				goto done;
967 		}
968 		while ((ifa = TAILQ_NEXT(ifa, ifa_link)) != 0) {
969 			if (af && af != ifa->ifa_addr->sa_family)
970 				continue;
971 			if (jailed(curthread->td_ucred) &&
972 			    prison_if(curthread->td_ucred, ifa->ifa_addr))
973 				continue;
974 			ifaaddr = ifa->ifa_addr;
975 			netmask = ifa->ifa_netmask;
976 			brdaddr = ifa->ifa_dstaddr;
977 			len = rt_msg2(RTM_NEWADDR, &info, 0, w);
978 			if (w->w_req && w->w_tmem) {
979 				register struct ifa_msghdr *ifam;
980 
981 				ifam = (struct ifa_msghdr *)w->w_tmem;
982 				ifam->ifam_index = ifa->ifa_ifp->if_index;
983 				ifam->ifam_flags = ifa->ifa_flags;
984 				ifam->ifam_metric = ifa->ifa_metric;
985 				ifam->ifam_addrs = info.rti_addrs;
986 				error = SYSCTL_OUT(w->w_req, w->w_tmem, len);
987 				if (error)
988 					goto done;
989 			}
990 		}
991 		ifaaddr = netmask = brdaddr = 0;
992 	}
993 done:
994 	return (error);
995 }
996 
997 static int
998 sysctl_rtsock(SYSCTL_HANDLER_ARGS)
999 {
1000 	int	*name = (int *)arg1;
1001 	u_int	namelen = arg2;
1002 	register struct radix_node_head *rnh;
1003 	int	i, s, error = EINVAL;
1004 	u_char  af;
1005 	struct	walkarg w;
1006 
1007 	name ++;
1008 	namelen--;
1009 	if (req->newptr)
1010 		return (EPERM);
1011 	if (namelen != 3)
1012 		return ((namelen < 3) ? EISDIR : ENOTDIR);
1013 	af = name[0];
1014 	Bzero(&w, sizeof(w));
1015 	w.w_op = name[1];
1016 	w.w_arg = name[2];
1017 	w.w_req = req;
1018 
1019 	s = splnet();
1020 	switch (w.w_op) {
1021 
1022 	case NET_RT_DUMP:
1023 	case NET_RT_FLAGS:
1024 		for (i = 1; i <= AF_MAX; i++)
1025 			if ((rnh = rt_tables[i]) && (af == 0 || af == i) &&
1026 			    (error = rnh->rnh_walktree(rnh,
1027 							sysctl_dumpentry, &w)))
1028 				break;
1029 		break;
1030 
1031 	case NET_RT_IFLIST:
1032 		error = sysctl_iflist(af, &w);
1033 	}
1034 	splx(s);
1035 	if (w.w_tmem)
1036 		free(w.w_tmem, M_RTABLE);
1037 	return (error);
1038 }
1039 
1040 SYSCTL_NODE(_net, PF_ROUTE, routetable, CTLFLAG_RD, sysctl_rtsock, "");
1041 
1042 /*
1043  * Definitions of protocols supported in the ROUTE domain.
1044  */
1045 
1046 extern struct domain routedomain;		/* or at least forward */
1047 
1048 static struct protosw routesw[] = {
1049 { SOCK_RAW,	&routedomain,	0,		PR_ATOMIC|PR_ADDR,
1050   0,		route_output,	raw_ctlinput,	0,
1051   0,
1052   raw_init,	0,		0,		0,
1053   &route_usrreqs
1054 }
1055 };
1056 
1057 static struct domain routedomain =
1058     { PF_ROUTE, "route", 0, 0, 0,
1059       routesw, &routesw[sizeof(routesw)/sizeof(routesw[0])] };
1060 
1061 DOMAIN_SET(route);
1062