xref: /netbsd-src/sys/net/rtsock.c (revision da5f4674a3fc214be3572d358b66af40ab9401e7)
1 /*	$NetBSD: rtsock.c,v 1.64 2003/08/07 16:32:58 agc Exp $	*/
2 
3 /*
4  * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
5  * All rights reserved.
6  *
7  * Redistribution and use in source and binary forms, with or without
8  * modification, are permitted provided that the following conditions
9  * are met:
10  * 1. Redistributions of source code must retain the above copyright
11  *    notice, this list of conditions and the following disclaimer.
12  * 2. Redistributions in binary form must reproduce the above copyright
13  *    notice, this list of conditions and the following disclaimer in the
14  *    documentation and/or other materials provided with the distribution.
15  * 3. Neither the name of the project nor the names of its contributors
16  *    may be used to endorse or promote products derived from this software
17  *    without specific prior written permission.
18  *
19  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
20  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
23  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29  * SUCH DAMAGE.
30  */
31 
32 /*
33  * Copyright (c) 1988, 1991, 1993
34  *	The Regents of the University of California.  All rights reserved.
35  *
36  * Redistribution and use in source and binary forms, with or without
37  * modification, are permitted provided that the following conditions
38  * are met:
39  * 1. Redistributions of source code must retain the above copyright
40  *    notice, this list of conditions and the following disclaimer.
41  * 2. Redistributions in binary form must reproduce the above copyright
42  *    notice, this list of conditions and the following disclaimer in the
43  *    documentation and/or other materials provided with the distribution.
44  * 3. Neither the name of the University nor the names of its contributors
45  *    may be used to endorse or promote products derived from this software
46  *    without specific prior written permission.
47  *
48  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
49  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
50  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
51  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
52  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
53  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
54  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
55  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
56  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
57  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
58  * SUCH DAMAGE.
59  *
60  *	@(#)rtsock.c	8.7 (Berkeley) 10/12/95
61  */
62 
63 #include <sys/cdefs.h>
64 __KERNEL_RCSID(0, "$NetBSD: rtsock.c,v 1.64 2003/08/07 16:32:58 agc Exp $");
65 
66 #include "opt_inet.h"
67 
68 #include <sys/param.h>
69 #include <sys/systm.h>
70 #include <sys/proc.h>
71 #include <sys/mbuf.h>
72 #include <sys/socket.h>
73 #include <sys/socketvar.h>
74 #include <sys/domain.h>
75 #include <sys/protosw.h>
76 #include <sys/sysctl.h>
77 
78 #include <net/if.h>
79 #include <net/route.h>
80 #include <net/raw_cb.h>
81 
82 #include <machine/stdarg.h>
83 
84 extern	struct domain routedomain;		/* or at least forward */
85 
86 struct	sockaddr route_dst = { 2, PF_ROUTE, };
87 struct	sockaddr route_src = { 2, PF_ROUTE, };
88 struct	sockproto route_proto = { PF_ROUTE, };
89 
90 struct walkarg {
91 	int	w_op;
92 	int	w_arg;
93 	int	w_given;
94 	int	w_needed;
95 	caddr_t	w_where;
96 	int	w_tmemsize;
97 	int	w_tmemneeded;
98 	caddr_t	w_tmem;
99 };
100 
101 static struct mbuf *rt_msg1 __P((int, struct rt_addrinfo *, caddr_t, int));
102 static int rt_msg2 __P((int, struct rt_addrinfo *, caddr_t, struct walkarg *,
103     int *));
104 static int rt_xaddrs __P((caddr_t, caddr_t, struct rt_addrinfo *));
105 static int sysctl_dumpentry __P((struct radix_node *, void *));
106 static int sysctl_iflist __P((int, struct walkarg *, int));
107 static int sysctl_rtable __P((int *, u_int, void *, size_t *, void *, size_t));
108 static __inline void rt_adjustcount __P((int, int));
109 
110 /* Sleazy use of local variables throughout file, warning!!!! */
111 #define dst	info.rti_info[RTAX_DST]
112 #define gate	info.rti_info[RTAX_GATEWAY]
113 #define netmask	info.rti_info[RTAX_NETMASK]
114 #define genmask	info.rti_info[RTAX_GENMASK]
115 #define ifpaddr	info.rti_info[RTAX_IFP]
116 #define ifaaddr	info.rti_info[RTAX_IFA]
117 #define brdaddr	info.rti_info[RTAX_BRD]
118 
119 static __inline void
120 rt_adjustcount(af, cnt)
121 	int af, cnt;
122 {
123 	route_cb.any_count += cnt;
124 	switch (af) {
125 	case AF_INET:
126 		route_cb.ip_count += cnt;
127 		return;
128 #ifdef INET6
129 	case AF_INET6:
130 		route_cb.ip6_count += cnt;
131 		return;
132 #endif
133 	case AF_IPX:
134 		route_cb.ipx_count += cnt;
135 		return;
136 	case AF_NS:
137 		route_cb.ns_count += cnt;
138 		return;
139 	case AF_ISO:
140 		route_cb.iso_count += cnt;
141 		return;
142 	}
143 }
144 
145 /*ARGSUSED*/
146 int
147 route_usrreq(so, req, m, nam, control, p)
148 	struct socket *so;
149 	int req;
150 	struct mbuf *m, *nam, *control;
151 	struct proc *p;
152 {
153 	int error = 0;
154 	struct rawcb *rp = sotorawcb(so);
155 	int s;
156 
157 	if (req == PRU_ATTACH) {
158 		MALLOC(rp, struct rawcb *, sizeof(*rp), M_PCB, M_WAITOK);
159 		if ((so->so_pcb = rp) != NULL)
160 			memset(so->so_pcb, 0, sizeof(*rp));
161 
162 	}
163 	if (req == PRU_DETACH && rp)
164 		rt_adjustcount(rp->rcb_proto.sp_protocol, -1);
165 	s = splsoftnet();
166 
167 	/*
168 	 * Don't call raw_usrreq() in the attach case, because
169 	 * we want to allow non-privileged processes to listen on
170 	 * and send "safe" commands to the routing socket.
171 	 */
172 	if (req == PRU_ATTACH) {
173 		if (p == 0)
174 			error = EACCES;
175 		else
176 			error = raw_attach(so, (int)(long)nam);
177 	} else
178 		error = raw_usrreq(so, req, m, nam, control, p);
179 
180 	rp = sotorawcb(so);
181 	if (req == PRU_ATTACH && rp) {
182 		if (error) {
183 			free((caddr_t)rp, M_PCB);
184 			splx(s);
185 			return (error);
186 		}
187 		rt_adjustcount(rp->rcb_proto.sp_protocol, 1);
188 		rp->rcb_laddr = &route_src;
189 		rp->rcb_faddr = &route_dst;
190 		soisconnected(so);
191 		so->so_options |= SO_USELOOPBACK;
192 	}
193 	splx(s);
194 	return (error);
195 }
196 
197 /*ARGSUSED*/
198 int
199 #if __STDC__
200 route_output(struct mbuf *m, ...)
201 #else
202 route_output(m, va_alist)
203 	struct mbuf *m;
204 	va_dcl
205 #endif
206 {
207 	struct rt_msghdr *rtm = 0;
208 	struct radix_node *rn = 0;
209 	struct rtentry *rt = 0;
210 	struct rtentry *saved_nrt = 0;
211 	struct radix_node_head *rnh;
212 	struct rt_addrinfo info;
213 	int len, error = 0;
214 	struct ifnet *ifp = 0;
215 	struct ifaddr *ifa = 0;
216 	struct socket *so;
217 	va_list ap;
218 	sa_family_t family;
219 
220 	va_start(ap, m);
221 	so = va_arg(ap, struct socket *);
222 	va_end(ap);
223 
224 #define senderr(e) do { error = e; goto flush;} while (/*CONSTCOND*/ 0)
225 	if (m == 0 || ((m->m_len < sizeof(int32_t)) &&
226 	   (m = m_pullup(m, sizeof(int32_t))) == 0))
227 		return (ENOBUFS);
228 	if ((m->m_flags & M_PKTHDR) == 0)
229 		panic("route_output");
230 	len = m->m_pkthdr.len;
231 	if (len < sizeof(*rtm) ||
232 	    len != mtod(m, struct rt_msghdr *)->rtm_msglen) {
233 		dst = 0;
234 		senderr(EINVAL);
235 	}
236 	R_Malloc(rtm, struct rt_msghdr *, len);
237 	if (rtm == 0) {
238 		dst = 0;
239 		senderr(ENOBUFS);
240 	}
241 	m_copydata(m, 0, len, (caddr_t)rtm);
242 	if (rtm->rtm_version != RTM_VERSION) {
243 		dst = 0;
244 		senderr(EPROTONOSUPPORT);
245 	}
246 	rtm->rtm_pid = curproc->p_pid;
247 	memset(&info, 0, sizeof(info));
248 	info.rti_addrs = rtm->rtm_addrs;
249 	if (rt_xaddrs((caddr_t)(rtm + 1), len + (caddr_t)rtm, &info))
250 		senderr(EINVAL);
251 	info.rti_flags = rtm->rtm_flags;
252 	if (dst == 0 || (dst->sa_family >= AF_MAX))
253 		senderr(EINVAL);
254 	if (gate != 0 && (gate->sa_family >= AF_MAX))
255 		senderr(EINVAL);
256 	if (genmask) {
257 		struct radix_node *t;
258 		t = rn_addmask((caddr_t)genmask, 0, 1);
259 		if (t && genmask->sa_len >= ((struct sockaddr *)t->rn_key)->sa_len &&
260 		    Bcmp((caddr_t *)genmask + 1, (caddr_t *)t->rn_key + 1,
261 		    ((struct sockaddr *)t->rn_key)->sa_len) - 1)
262 			genmask = (struct sockaddr *)(t->rn_key);
263 		else
264 			senderr(ENOBUFS);
265 	}
266 
267 	/*
268 	 * Verify that the caller has the appropriate privilege; RTM_GET
269 	 * is the only operation the non-superuser is allowed.
270 	 */
271 	if (rtm->rtm_type != RTM_GET &&
272 	    suser(curproc->p_ucred, &curproc->p_acflag) != 0)
273 		senderr(EACCES);
274 
275 	switch (rtm->rtm_type) {
276 
277 	case RTM_ADD:
278 		if (gate == 0)
279 			senderr(EINVAL);
280 		error = rtrequest1(rtm->rtm_type, &info, &saved_nrt);
281 		if (error == 0 && saved_nrt) {
282 			rt_setmetrics(rtm->rtm_inits,
283 			    &rtm->rtm_rmx, &saved_nrt->rt_rmx);
284 			saved_nrt->rt_refcnt--;
285 			saved_nrt->rt_genmask = genmask;
286 		}
287 		break;
288 
289 	case RTM_DELETE:
290 		error = rtrequest1(rtm->rtm_type, &info, &saved_nrt);
291 		if (error == 0) {
292 			(rt = saved_nrt)->rt_refcnt++;
293 			goto report;
294 		}
295 		break;
296 
297 	case RTM_GET:
298 	case RTM_CHANGE:
299 	case RTM_LOCK:
300 		if ((rnh = rt_tables[dst->sa_family]) == 0) {
301 			senderr(EAFNOSUPPORT);
302 		}
303 		rn = rnh->rnh_lookup(dst, netmask, rnh);
304 		if (rn == NULL || (rn->rn_flags & RNF_ROOT) != 0) {
305 			senderr(ESRCH);
306 		}
307 		rt = (struct rtentry *)rn;
308 		rt->rt_refcnt++;
309 		if (rtm->rtm_type != RTM_GET) {/* XXX: too grotty */
310 			struct radix_node *rn;
311 			extern struct radix_node_head *mask_rnhead;
312 
313 			if (Bcmp(dst, rt_key(rt), dst->sa_len) != 0)
314 				senderr(ESRCH);
315 			if (netmask && (rn = rn_search(netmask,
316 					    mask_rnhead->rnh_treetop)))
317 				netmask = (struct sockaddr *)rn->rn_key;
318 			for (rn = rt->rt_nodes; rn; rn = rn->rn_dupedkey)
319 				if (netmask == (struct sockaddr *)rn->rn_mask)
320 					break;
321 			if (rn == 0)
322 				senderr(ETOOMANYREFS);
323 			rt = (struct rtentry *)rn;
324 		}
325 
326 		switch (rtm->rtm_type) {
327 		case RTM_GET:
328 		report:
329 			dst = rt_key(rt);
330 			gate = rt->rt_gateway;
331 			netmask = rt_mask(rt);
332 			genmask = rt->rt_genmask;
333 			if (rtm->rtm_addrs & (RTA_IFP | RTA_IFA)) {
334 				if ((ifp = rt->rt_ifp) != NULL) {
335 					ifpaddr = TAILQ_FIRST(&ifp->if_addrlist)->ifa_addr;
336 					ifaaddr = rt->rt_ifa->ifa_addr;
337 					if (ifp->if_flags & IFF_POINTOPOINT)
338 						brdaddr = rt->rt_ifa->ifa_dstaddr;
339 					else
340 						brdaddr = 0;
341 					rtm->rtm_index = ifp->if_index;
342 				} else {
343 					ifpaddr = 0;
344 					ifaaddr = 0;
345 				}
346 			}
347 			(void)rt_msg2(rtm->rtm_type, &info, (caddr_t)0,
348 			    (struct walkarg *)0, &len);
349 			if (len > rtm->rtm_msglen) {
350 				struct rt_msghdr *new_rtm;
351 				R_Malloc(new_rtm, struct rt_msghdr *, len);
352 				if (new_rtm == 0)
353 					senderr(ENOBUFS);
354 				Bcopy(rtm, new_rtm, rtm->rtm_msglen);
355 				Free(rtm); rtm = new_rtm;
356 			}
357 			(void)rt_msg2(rtm->rtm_type, &info, (caddr_t)rtm,
358 			    (struct walkarg *)0, 0);
359 			rtm->rtm_flags = rt->rt_flags;
360 			rtm->rtm_rmx = rt->rt_rmx;
361 			rtm->rtm_addrs = info.rti_addrs;
362 			break;
363 
364 		case RTM_CHANGE:
365 			/*
366 			 * new gateway could require new ifaddr, ifp;
367 			 * flags may also be different; ifp may be specified
368 			 * by ll sockaddr when protocol address is ambiguous
369 			 */
370 			if ((error = rt_getifa(&info)) != 0)
371 				senderr(error);
372 			if (gate && rt_setgate(rt, rt_key(rt), gate))
373 				senderr(EDQUOT);
374 			/* new gateway could require new ifaddr, ifp;
375 			   flags may also be different; ifp may be specified
376 			   by ll sockaddr when protocol address is ambiguous */
377 			if (ifpaddr && (ifa = ifa_ifwithnet(ifpaddr)) &&
378 			    (ifp = ifa->ifa_ifp) && (ifaaddr || gate))
379 				ifa = ifaof_ifpforaddr(ifaaddr ? ifaaddr : gate,
380 				    ifp);
381 			else if ((ifaaddr && (ifa = ifa_ifwithaddr(ifaaddr))) ||
382 			    (gate && (ifa = ifa_ifwithroute(rt->rt_flags,
383 			    rt_key(rt), gate))))
384 				ifp = ifa->ifa_ifp;
385 			if (ifa) {
386 				struct ifaddr *oifa = rt->rt_ifa;
387 				if (oifa != ifa) {
388 				    if (oifa && oifa->ifa_rtrequest)
389 					oifa->ifa_rtrequest(RTM_DELETE, rt,
390 					    &info);
391 				    IFAFREE(rt->rt_ifa);
392 				    rt->rt_ifa = ifa;
393 				    IFAREF(rt->rt_ifa);
394 				    rt->rt_ifp = ifp;
395 				}
396 			}
397 			rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx,
398 			    &rt->rt_rmx);
399 			if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest)
400 				rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, &info);
401 			if (genmask)
402 				rt->rt_genmask = genmask;
403 			/*
404 			 * Fall into
405 			 */
406 		case RTM_LOCK:
407 			rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits);
408 			rt->rt_rmx.rmx_locks |=
409 			    (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks);
410 			break;
411 		}
412 		break;
413 
414 	default:
415 		senderr(EOPNOTSUPP);
416 	}
417 
418 flush:
419 	if (rtm) {
420 		if (error)
421 			rtm->rtm_errno = error;
422 		else
423 			rtm->rtm_flags |= RTF_DONE;
424 	}
425 	family = dst ? dst->sa_family : 0;
426 	if (rt)
427 		rtfree(rt);
428     {
429 	struct rawcb *rp = 0;
430 	/*
431 	 * Check to see if we don't want our own messages.
432 	 */
433 	if ((so->so_options & SO_USELOOPBACK) == 0) {
434 		if (route_cb.any_count <= 1) {
435 			if (rtm)
436 				Free(rtm);
437 			m_freem(m);
438 			return (error);
439 		}
440 		/* There is another listener, so construct message */
441 		rp = sotorawcb(so);
442 	}
443 	if (rtm) {
444 		m_copyback(m, 0, rtm->rtm_msglen, (caddr_t)rtm);
445 		if (m->m_pkthdr.len < rtm->rtm_msglen) {
446 			m_freem(m);
447 			m = NULL;
448 		} else if (m->m_pkthdr.len > rtm->rtm_msglen)
449 			m_adj(m, rtm->rtm_msglen - m->m_pkthdr.len);
450 		Free(rtm);
451 	}
452 	if (rp)
453 		rp->rcb_proto.sp_family = 0; /* Avoid us */
454 	if (family)
455 		route_proto.sp_protocol = family;
456 	if (m)
457 		raw_input(m, &route_proto, &route_src, &route_dst);
458 	if (rp)
459 		rp->rcb_proto.sp_family = PF_ROUTE;
460     }
461 	return (error);
462 }
463 
464 void
465 rt_setmetrics(which, in, out)
466 	u_long which;
467 	struct rt_metrics *in, *out;
468 {
469 #define metric(f, e) if (which & (f)) out->e = in->e;
470 	metric(RTV_RPIPE, rmx_recvpipe);
471 	metric(RTV_SPIPE, rmx_sendpipe);
472 	metric(RTV_SSTHRESH, rmx_ssthresh);
473 	metric(RTV_RTT, rmx_rtt);
474 	metric(RTV_RTTVAR, rmx_rttvar);
475 	metric(RTV_HOPCOUNT, rmx_hopcount);
476 	metric(RTV_MTU, rmx_mtu);
477 	metric(RTV_EXPIRE, rmx_expire);
478 #undef metric
479 }
480 
481 #define ROUNDUP(a) \
482 	((a) > 0 ? (1 + (((a) - 1) | (sizeof(long) - 1))) : sizeof(long))
483 #define ADVANCE(x, n) (x += ROUNDUP((n)->sa_len))
484 
485 static int
486 rt_xaddrs(cp, cplim, rtinfo)
487 	caddr_t cp, cplim;
488 	struct rt_addrinfo *rtinfo;
489 {
490 	struct sockaddr *sa = NULL;	/* Quell compiler warning */
491 	int i;
492 
493 	for (i = 0; (i < RTAX_MAX) && (cp < cplim); i++) {
494 		if ((rtinfo->rti_addrs & (1 << i)) == 0)
495 			continue;
496 		rtinfo->rti_info[i] = sa = (struct sockaddr *)cp;
497 		ADVANCE(cp, sa);
498 	}
499 
500 	/* Check for extra addresses specified.  */
501 	if ((rtinfo->rti_addrs & (~0 << i)) != 0)
502 		return (1);
503 	/* Check for bad data length.  */
504 	if (cp != cplim) {
505 		if (i == RTAX_NETMASK + 1 &&
506 		    cp - ROUNDUP(sa->sa_len) + sa->sa_len == cplim)
507 			/*
508 			 * The last sockaddr was netmask.
509 			 * We accept this for now for the sake of old
510 			 * binaries or third party softwares.
511 			 */
512 			;
513 		else
514 			return (1);
515 	}
516 	return (0);
517 }
518 
519 static struct mbuf *
520 rt_msg1(type, rtinfo, data, datalen)
521 	int type;
522 	struct rt_addrinfo *rtinfo;
523 	caddr_t data;
524 	int datalen;
525 {
526 	struct rt_msghdr *rtm;
527 	struct mbuf *m;
528 	int i;
529 	struct sockaddr *sa;
530 	int len, dlen;
531 
532 	m = m_gethdr(M_DONTWAIT, MT_DATA);
533 	if (m == 0)
534 		return (m);
535 	MCLAIM(m, &routedomain.dom_mowner);
536 	switch (type) {
537 
538 	case RTM_DELADDR:
539 	case RTM_NEWADDR:
540 		len = sizeof(struct ifa_msghdr);
541 		break;
542 
543 #ifdef COMPAT_14
544 	case RTM_OIFINFO:
545 		len = sizeof(struct if_msghdr14);
546 		break;
547 #endif
548 
549 	case RTM_IFINFO:
550 		len = sizeof(struct if_msghdr);
551 		break;
552 
553 	case RTM_IFANNOUNCE:
554 		len = sizeof(struct if_announcemsghdr);
555 		break;
556 
557 	default:
558 		len = sizeof(struct rt_msghdr);
559 	}
560 	if (len > MHLEN + MLEN)
561 		panic("rt_msg1: message too long");
562 	else if (len > MHLEN) {
563 		m->m_next = m_get(M_DONTWAIT, MT_DATA);
564 		if (m->m_next == NULL) {
565 			m_freem(m);
566 			return (NULL);
567 		}
568 		MCLAIM(m->m_next, m->m_owner);
569 		m->m_pkthdr.len = len;
570 		m->m_len = MHLEN;
571 		m->m_next->m_len = len - MHLEN;
572 	} else {
573 		m->m_pkthdr.len = m->m_len = len;
574 	}
575 	m->m_pkthdr.rcvif = 0;
576 	m_copyback(m, 0, datalen, data);
577 	rtm = mtod(m, struct rt_msghdr *);
578 	for (i = 0; i < RTAX_MAX; i++) {
579 		if ((sa = rtinfo->rti_info[i]) == NULL)
580 			continue;
581 		rtinfo->rti_addrs |= (1 << i);
582 		dlen = ROUNDUP(sa->sa_len);
583 		m_copyback(m, len, dlen, (caddr_t)sa);
584 		len += dlen;
585 	}
586 	if (m->m_pkthdr.len != len) {
587 		m_freem(m);
588 		return (NULL);
589 	}
590 	rtm->rtm_msglen = len;
591 	rtm->rtm_version = RTM_VERSION;
592 	rtm->rtm_type = type;
593 	return (m);
594 }
595 
596 /*
597  * rt_msg2
598  *
599  *	 fills 'cp' or 'w'.w_tmem with the routing socket message and
600  *		returns the length of the message in 'lenp'.
601  *
602  * if walkarg is 0, cp is expected to be 0 or a buffer large enough to hold
603  *	the message
604  * otherwise walkarg's w_needed is updated and if the user buffer is
605  *	specified and w_needed indicates space exists the information is copied
606  *	into the temp space (w_tmem). w_tmem is [re]allocated if necessary,
607  *	if the allocation fails ENOBUFS is returned.
608  */
609 static int
610 rt_msg2(type, rtinfo, cp, w, lenp)
611 	int type;
612 	struct rt_addrinfo *rtinfo;
613 	caddr_t cp;
614 	struct walkarg *w;
615 	int *lenp;
616 {
617 	int i;
618 	int len, dlen, second_time = 0;
619 	caddr_t cp0;
620 
621 	rtinfo->rti_addrs = 0;
622 again:
623 	switch (type) {
624 
625 	case RTM_DELADDR:
626 	case RTM_NEWADDR:
627 		len = sizeof(struct ifa_msghdr);
628 		break;
629 #ifdef COMPAT_14
630 	case RTM_OIFINFO:
631 		len = sizeof(struct if_msghdr14);
632 		break;
633 #endif
634 
635 	case RTM_IFINFO:
636 		len = sizeof(struct if_msghdr);
637 		break;
638 
639 	default:
640 		len = sizeof(struct rt_msghdr);
641 	}
642 	if ((cp0 = cp) != NULL)
643 		cp += len;
644 	for (i = 0; i < RTAX_MAX; i++) {
645 		struct sockaddr *sa;
646 
647 		if ((sa = rtinfo->rti_info[i]) == 0)
648 			continue;
649 		rtinfo->rti_addrs |= (1 << i);
650 		dlen = ROUNDUP(sa->sa_len);
651 		if (cp) {
652 			bcopy(sa, cp, (unsigned)dlen);
653 			cp += dlen;
654 		}
655 		len += dlen;
656 	}
657 	if (cp == 0 && w != NULL && !second_time) {
658 		struct walkarg *rw = w;
659 
660 		rw->w_needed += len;
661 		if (rw->w_needed <= 0 && rw->w_where) {
662 			if (rw->w_tmemsize < len) {
663 				if (rw->w_tmem)
664 					free(rw->w_tmem, M_RTABLE);
665 				rw->w_tmem = (caddr_t) malloc(len, M_RTABLE,
666 				    M_NOWAIT);
667 				if (rw->w_tmem)
668 					rw->w_tmemsize = len;
669 			}
670 			if (rw->w_tmem) {
671 				cp = rw->w_tmem;
672 				second_time = 1;
673 				goto again;
674 			} else {
675 				rw->w_tmemneeded = len;
676 				return (ENOBUFS);
677 			}
678 		}
679 	}
680 	if (cp) {
681 		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 	if (lenp)
688 		*lenp = len;
689 	return (0);
690 }
691 
692 /*
693  * This routine is called to generate a message from the routing
694  * socket indicating that a redirect has occurred, a routing lookup
695  * has failed, or that a protocol has detected timeouts to a particular
696  * destination.
697  */
698 void
699 rt_missmsg(type, rtinfo, flags, error)
700 	int type, flags, error;
701 	struct rt_addrinfo *rtinfo;
702 {
703 	struct rt_msghdr rtm;
704 	struct mbuf *m;
705 	struct sockaddr *sa = rtinfo->rti_info[RTAX_DST];
706 
707 	if (route_cb.any_count == 0)
708 		return;
709 	memset(&rtm, 0, sizeof(rtm));
710 	rtm.rtm_flags = RTF_DONE | flags;
711 	rtm.rtm_errno = error;
712 	m = rt_msg1(type, rtinfo, (caddr_t)&rtm, sizeof(rtm));
713 	if (m == 0)
714 		return;
715 	mtod(m, struct rt_msghdr *)->rtm_addrs = rtinfo->rti_addrs;
716 	route_proto.sp_protocol = sa ? sa->sa_family : 0;
717 	raw_input(m, &route_proto, &route_src, &route_dst);
718 }
719 
720 /*
721  * This routine is called to generate a message from the routing
722  * socket indicating that the status of a network interface has changed.
723  */
724 void
725 rt_ifmsg(ifp)
726 	struct ifnet *ifp;
727 {
728 	struct if_msghdr ifm;
729 #ifdef COMPAT_14
730 	struct if_msghdr14 oifm;
731 #endif
732 	struct mbuf *m;
733 	struct rt_addrinfo info;
734 
735 	if (route_cb.any_count == 0)
736 		return;
737 	memset(&info, 0, sizeof(info));
738 	memset(&ifm, 0, sizeof(ifm));
739 	ifm.ifm_index = ifp->if_index;
740 	ifm.ifm_flags = ifp->if_flags;
741 	ifm.ifm_data = ifp->if_data;
742 	ifm.ifm_addrs = 0;
743 	m = rt_msg1(RTM_IFINFO, &info, (caddr_t)&ifm, sizeof(ifm));
744 	if (m == 0)
745 		return;
746 	route_proto.sp_protocol = 0;
747 	raw_input(m, &route_proto, &route_src, &route_dst);
748 #ifdef COMPAT_14
749 	memset(&info, 0, sizeof(info));
750 	memset(&oifm, 0, sizeof(oifm));
751 	oifm.ifm_index = ifp->if_index;
752 	oifm.ifm_flags = ifp->if_flags;
753 	oifm.ifm_data.ifi_type = ifp->if_data.ifi_type;
754 	oifm.ifm_data.ifi_addrlen = ifp->if_data.ifi_addrlen;
755 	oifm.ifm_data.ifi_hdrlen = ifp->if_data.ifi_hdrlen;
756 	oifm.ifm_data.ifi_mtu = ifp->if_data.ifi_mtu;
757 	oifm.ifm_data.ifi_metric = ifp->if_data.ifi_metric;
758 	oifm.ifm_data.ifi_baudrate = ifp->if_data.ifi_baudrate;
759 	oifm.ifm_data.ifi_ipackets = ifp->if_data.ifi_ipackets;
760 	oifm.ifm_data.ifi_ierrors = ifp->if_data.ifi_ierrors;
761 	oifm.ifm_data.ifi_opackets = ifp->if_data.ifi_opackets;
762 	oifm.ifm_data.ifi_oerrors = ifp->if_data.ifi_oerrors;
763 	oifm.ifm_data.ifi_collisions = ifp->if_data.ifi_collisions;
764 	oifm.ifm_data.ifi_ibytes = ifp->if_data.ifi_ibytes;
765 	oifm.ifm_data.ifi_obytes = ifp->if_data.ifi_obytes;
766 	oifm.ifm_data.ifi_imcasts = ifp->if_data.ifi_imcasts;
767 	oifm.ifm_data.ifi_omcasts = ifp->if_data.ifi_omcasts;
768 	oifm.ifm_data.ifi_iqdrops = ifp->if_data.ifi_iqdrops;
769 	oifm.ifm_data.ifi_noproto = ifp->if_data.ifi_noproto;
770 	oifm.ifm_data.ifi_lastchange = ifp->if_data.ifi_lastchange;
771 	oifm.ifm_addrs = 0;
772 	m = rt_msg1(RTM_OIFINFO, &info, (caddr_t)&oifm, sizeof(oifm));
773 	if (m == 0)
774 		return;
775 	route_proto.sp_protocol = 0;
776 	raw_input(m, &route_proto, &route_src, &route_dst);
777 #endif
778 }
779 
780 /*
781  * This is called to generate messages from the routing socket
782  * indicating a network interface has had addresses associated with it.
783  * if we ever reverse the logic and replace messages TO the routing
784  * socket indicate a request to configure interfaces, then it will
785  * be unnecessary as the routing socket will automatically generate
786  * copies of it.
787  */
788 void
789 rt_newaddrmsg(cmd, ifa, error, rt)
790 	int cmd, error;
791 	struct ifaddr *ifa;
792 	struct rtentry *rt;
793 {
794 	struct rt_addrinfo info;
795 	struct sockaddr *sa = NULL;
796 	int pass;
797 	struct mbuf *m = NULL;
798 	struct ifnet *ifp = ifa->ifa_ifp;
799 
800 	if (route_cb.any_count == 0)
801 		return;
802 	for (pass = 1; pass < 3; pass++) {
803 		memset(&info, 0, sizeof(info));
804 		if ((cmd == RTM_ADD && pass == 1) ||
805 		    (cmd == RTM_DELETE && pass == 2)) {
806 			struct ifa_msghdr ifam;
807 			int ncmd = cmd == RTM_ADD ? RTM_NEWADDR : RTM_DELADDR;
808 
809 			ifaaddr = sa = ifa->ifa_addr;
810 			ifpaddr = TAILQ_FIRST(&ifp->if_addrlist)->ifa_addr;
811 			netmask = ifa->ifa_netmask;
812 			brdaddr = ifa->ifa_dstaddr;
813 			memset(&ifam, 0, sizeof(ifam));
814 			ifam.ifam_index = ifp->if_index;
815 			ifam.ifam_metric = ifa->ifa_metric;
816 			ifam.ifam_flags = ifa->ifa_flags;
817 			m = rt_msg1(ncmd, &info, (caddr_t)&ifam, sizeof(ifam));
818 			if (m == NULL)
819 				continue;
820 			mtod(m, struct ifa_msghdr *)->ifam_addrs =
821 			    info.rti_addrs;
822 		}
823 		if ((cmd == RTM_ADD && pass == 2) ||
824 		    (cmd == RTM_DELETE && pass == 1)) {
825 			struct rt_msghdr rtm;
826 
827 			if (rt == 0)
828 				continue;
829 			netmask = rt_mask(rt);
830 			dst = sa = rt_key(rt);
831 			gate = rt->rt_gateway;
832 			memset(&rtm, 0, sizeof(rtm));
833 			rtm.rtm_index = ifp->if_index;
834 			rtm.rtm_flags |= rt->rt_flags;
835 			rtm.rtm_errno = error;
836 			m = rt_msg1(cmd, &info, (caddr_t)&rtm, sizeof(rtm));
837 			if (m == NULL)
838 				continue;
839 			mtod(m, struct rt_msghdr *)->rtm_addrs = info.rti_addrs;
840 		}
841 		route_proto.sp_protocol = sa ? sa->sa_family : 0;
842 		raw_input(m, &route_proto, &route_src, &route_dst);
843 	}
844 }
845 
846 /*
847  * This is called to generate routing socket messages indicating
848  * network interface arrival and departure.
849  */
850 void
851 rt_ifannouncemsg(ifp, what)
852 	struct ifnet *ifp;
853 	int what;
854 {
855 	struct if_announcemsghdr ifan;
856 	struct mbuf *m;
857 	struct rt_addrinfo info;
858 
859 	if (route_cb.any_count == 0)
860 		return;
861 	memset(&info, 0, sizeof(info));
862 	memset(&ifan, 0, sizeof(ifan));
863 	ifan.ifan_index = ifp->if_index;
864 	strlcpy(ifan.ifan_name, ifp->if_xname, sizeof(ifan.ifan_name));
865 	ifan.ifan_what = what;
866 	m = rt_msg1(RTM_IFANNOUNCE, &info, (caddr_t)&ifan, sizeof(ifan));
867 	if (m == 0)
868 		return;
869 	route_proto.sp_protocol = 0;
870 	raw_input(m, &route_proto, &route_src, &route_dst);
871 }
872 
873 /*
874  * This is used in dumping the kernel table via sysctl().
875  */
876 static int
877 sysctl_dumpentry(rn, v)
878 	struct radix_node *rn;
879 	void *v;
880 {
881 	struct walkarg *w = v;
882 	struct rtentry *rt = (struct rtentry *)rn;
883 	int error = 0, size;
884 	struct rt_addrinfo info;
885 
886 	if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg))
887 		return 0;
888 	memset(&info, 0, sizeof(info));
889 	dst = rt_key(rt);
890 	gate = rt->rt_gateway;
891 	netmask = rt_mask(rt);
892 	genmask = rt->rt_genmask;
893 	if (rt->rt_ifp) {
894 		ifpaddr = TAILQ_FIRST(&rt->rt_ifp->if_addrlist)->ifa_addr;
895 		ifaaddr = rt->rt_ifa->ifa_addr;
896 		if (rt->rt_ifp->if_flags & IFF_POINTOPOINT)
897 			brdaddr = rt->rt_ifa->ifa_dstaddr;
898 	}
899 	if ((error = rt_msg2(RTM_GET, &info, 0, w, &size)))
900 		return (error);
901 	if (w->w_where && w->w_tmem && w->w_needed <= 0) {
902 		struct rt_msghdr *rtm = (struct rt_msghdr *)w->w_tmem;
903 
904 		rtm->rtm_flags = rt->rt_flags;
905 		rtm->rtm_use = rt->rt_use;
906 		rtm->rtm_rmx = rt->rt_rmx;
907 		rtm->rtm_index = rt->rt_ifp->if_index;
908 		rtm->rtm_errno = rtm->rtm_pid = rtm->rtm_seq = 0;
909 		rtm->rtm_addrs = info.rti_addrs;
910 		if ((error = copyout(rtm, w->w_where, size)) != 0)
911 			w->w_where = NULL;
912 		else
913 			w->w_where += size;
914 	}
915 	return (error);
916 }
917 
918 static int
919 sysctl_iflist(af, w, type)
920 	int	af;
921 	struct	walkarg *w;
922 	int type;
923 {
924 	struct ifnet *ifp;
925 	struct ifaddr *ifa;
926 	struct	rt_addrinfo info;
927 	int	len, error = 0;
928 
929 	memset(&info, 0, sizeof(info));
930 	TAILQ_FOREACH(ifp, &ifnet, if_list) {
931 		if (w->w_arg && w->w_arg != ifp->if_index)
932 			continue;
933 		ifa = TAILQ_FIRST(&ifp->if_addrlist);
934 		ifpaddr = ifa->ifa_addr;
935 		switch (type) {
936 		case NET_RT_IFLIST:
937 			error =
938 			    rt_msg2(RTM_IFINFO, &info, (caddr_t)0, w, &len);
939 			break;
940 #ifdef COMPAT_14
941 		case NET_RT_OIFLIST:
942 			error =
943 			    rt_msg2(RTM_OIFINFO, &info, (caddr_t)0, w, &len);
944 			break;
945 #endif
946 		default:
947 			panic("sysctl_iflist(1)");
948 		}
949 		if (error)
950 			return (error);
951 		ifpaddr = 0;
952 		if (w->w_where && w->w_tmem && w->w_needed <= 0) {
953 			switch (type) {
954 			case NET_RT_IFLIST: {
955 				struct if_msghdr *ifm;
956 
957 				ifm = (struct if_msghdr *)w->w_tmem;
958 				ifm->ifm_index = ifp->if_index;
959 				ifm->ifm_flags = ifp->if_flags;
960 				ifm->ifm_data = ifp->if_data;
961 				ifm->ifm_addrs = info.rti_addrs;
962 				error = copyout(ifm, w->w_where, len);
963 				if (error)
964 					return (error);
965 				w->w_where += len;
966 				break;
967 			}
968 
969 #ifdef COMPAT_14
970 			case NET_RT_OIFLIST: {
971 				struct if_msghdr14 *ifm;
972 
973 				ifm = (struct if_msghdr14 *)w->w_tmem;
974 				ifm->ifm_index = ifp->if_index;
975 				ifm->ifm_flags = ifp->if_flags;
976 				ifm->ifm_data.ifi_type = ifp->if_data.ifi_type;
977 				ifm->ifm_data.ifi_addrlen =
978 				    ifp->if_data.ifi_addrlen;
979 				ifm->ifm_data.ifi_hdrlen =
980 				    ifp->if_data.ifi_hdrlen;
981 				ifm->ifm_data.ifi_mtu = ifp->if_data.ifi_mtu;
982 				ifm->ifm_data.ifi_metric =
983 				    ifp->if_data.ifi_metric;
984 				ifm->ifm_data.ifi_baudrate =
985 				    ifp->if_data.ifi_baudrate;
986 				ifm->ifm_data.ifi_ipackets =
987 				    ifp->if_data.ifi_ipackets;
988 				ifm->ifm_data.ifi_ierrors =
989 				    ifp->if_data.ifi_ierrors;
990 				ifm->ifm_data.ifi_opackets =
991 				    ifp->if_data.ifi_opackets;
992 				ifm->ifm_data.ifi_oerrors =
993 				    ifp->if_data.ifi_oerrors;
994 				ifm->ifm_data.ifi_collisions =
995 				    ifp->if_data.ifi_collisions;
996 				ifm->ifm_data.ifi_ibytes =
997 				    ifp->if_data.ifi_ibytes;
998 				ifm->ifm_data.ifi_obytes =
999 				    ifp->if_data.ifi_obytes;
1000 				ifm->ifm_data.ifi_imcasts =
1001 				    ifp->if_data.ifi_imcasts;
1002 				ifm->ifm_data.ifi_omcasts =
1003 				    ifp->if_data.ifi_omcasts;
1004 				ifm->ifm_data.ifi_iqdrops =
1005 				    ifp->if_data.ifi_iqdrops;
1006 				ifm->ifm_data.ifi_noproto =
1007 				    ifp->if_data.ifi_noproto;
1008 				ifm->ifm_data.ifi_lastchange =
1009 				    ifp->if_data.ifi_lastchange;
1010 				ifm->ifm_addrs = info.rti_addrs;
1011 				error = copyout(ifm, w->w_where, len);
1012 				if (error)
1013 					return (error);
1014 				w->w_where += len;
1015 				break;
1016 			}
1017 #endif
1018 			default:
1019 				panic("sysctl_iflist(2)");
1020 			}
1021 		}
1022 		while ((ifa = TAILQ_NEXT(ifa, ifa_list)) != NULL) {
1023 			if (af && af != ifa->ifa_addr->sa_family)
1024 				continue;
1025 			ifaaddr = ifa->ifa_addr;
1026 			netmask = ifa->ifa_netmask;
1027 			brdaddr = ifa->ifa_dstaddr;
1028 			if ((error = rt_msg2(RTM_NEWADDR, &info, 0, w, &len)))
1029 				return (error);
1030 			if (w->w_where && w->w_tmem && w->w_needed <= 0) {
1031 				struct ifa_msghdr *ifam;
1032 
1033 				ifam = (struct ifa_msghdr *)w->w_tmem;
1034 				ifam->ifam_index = ifa->ifa_ifp->if_index;
1035 				ifam->ifam_flags = ifa->ifa_flags;
1036 				ifam->ifam_metric = ifa->ifa_metric;
1037 				ifam->ifam_addrs = info.rti_addrs;
1038 				error = copyout(w->w_tmem, w->w_where, len);
1039 				if (error)
1040 					return (error);
1041 				w->w_where += len;
1042 			}
1043 		}
1044 		ifaaddr = netmask = brdaddr = 0;
1045 	}
1046 	return (0);
1047 }
1048 
1049 static int
1050 sysctl_rtable(name, namelen, where, given, new, newlen)
1051 	int	*name;
1052 	u_int	namelen;
1053 	void 	*where;
1054 	size_t	*given;
1055 	void	*new;
1056 	size_t	newlen;
1057 {
1058 	struct radix_node_head *rnh;
1059 	int	i, s, error = EINVAL;
1060 	u_char  af;
1061 	struct	walkarg w;
1062 
1063 	if (new)
1064 		return (EPERM);
1065 	if (namelen != 3)
1066 		return (EINVAL);
1067 	af = name[0];
1068 	w.w_tmemneeded = 0;
1069 	w.w_tmemsize = 0;
1070 	w.w_tmem = NULL;
1071 again:
1072 	/* we may return here if a later [re]alloc of the t_mem buffer fails */
1073 	if (w.w_tmemneeded) {
1074 		w.w_tmem = (caddr_t) malloc(w.w_tmemneeded, M_RTABLE, M_WAITOK);
1075 		w.w_tmemsize = w.w_tmemneeded;
1076 		w.w_tmemneeded = 0;
1077 	}
1078 	w.w_op = name[1];
1079 	w.w_arg = name[2];
1080 	w.w_given = *given;
1081 	w.w_needed = 0 - w.w_given;
1082 	w.w_where = where;
1083 
1084 	s = splsoftnet();
1085 	switch (w.w_op) {
1086 
1087 	case NET_RT_DUMP:
1088 	case NET_RT_FLAGS:
1089 		for (i = 1; i <= AF_MAX; i++)
1090 			if ((rnh = rt_tables[i]) && (af == 0 || af == i) &&
1091 			    (error = (*rnh->rnh_walktree)(rnh,
1092 			    sysctl_dumpentry, &w)))
1093 				break;
1094 		break;
1095 
1096 #ifdef COMPAT_14
1097 	case NET_RT_OIFLIST:
1098 		error = sysctl_iflist(af, &w, w.w_op);
1099 		break;
1100 #endif
1101 
1102 	case NET_RT_IFLIST:
1103 		error = sysctl_iflist(af, &w, w.w_op);
1104 	}
1105 	splx(s);
1106 
1107 	/* check to see if we couldn't allocate memory with NOWAIT */
1108 	if (error == ENOBUFS && w.w_tmem == 0 && w.w_tmemneeded)
1109 		goto again;
1110 
1111 	if (w.w_tmem)
1112 		free(w.w_tmem, M_RTABLE);
1113 	w.w_needed += w.w_given;
1114 	if (where) {
1115 		*given = w.w_where - (caddr_t) where;
1116 		if (*given < w.w_needed)
1117 			return (ENOMEM);
1118 	} else {
1119 		*given = (11 * w.w_needed) / 10;
1120 	}
1121 	return (error);
1122 }
1123 
1124 /*
1125  * Definitions of protocols supported in the ROUTE domain.
1126  */
1127 
1128 struct protosw routesw[] = {
1129 { SOCK_RAW,	&routedomain,	0,		PR_ATOMIC|PR_ADDR,
1130   raw_input,	route_output,	raw_ctlinput,	0,
1131   route_usrreq,
1132   raw_init,	0,		0,		0,
1133   sysctl_rtable,
1134 }
1135 };
1136 
1137 struct domain routedomain =
1138     { PF_ROUTE, "route", route_init, 0, 0,
1139       routesw, &routesw[sizeof(routesw)/sizeof(routesw[0])] };
1140