xref: /netbsd-src/sys/net/rtsock.c (revision 001c68bd94f75ce9270b69227c4199fbf34ee396)
1 /*	$NetBSD: rtsock.c,v 1.63 2003/06/29 22:31:53 fvdl 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. All advertising materials mentioning features or use of this software
45  *    must display the following acknowledgement:
46  *	This product includes software developed by the University of
47  *	California, Berkeley and its contributors.
48  * 4. Neither the name of the University nor the names of its contributors
49  *    may be used to endorse or promote products derived from this software
50  *    without specific prior written permission.
51  *
52  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
53  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
54  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
55  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
56  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
57  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
58  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
59  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
60  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
61  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
62  * SUCH DAMAGE.
63  *
64  *	@(#)rtsock.c	8.7 (Berkeley) 10/12/95
65  */
66 
67 #include <sys/cdefs.h>
68 __KERNEL_RCSID(0, "$NetBSD: rtsock.c,v 1.63 2003/06/29 22:31:53 fvdl Exp $");
69 
70 #include "opt_inet.h"
71 
72 #include <sys/param.h>
73 #include <sys/systm.h>
74 #include <sys/proc.h>
75 #include <sys/mbuf.h>
76 #include <sys/socket.h>
77 #include <sys/socketvar.h>
78 #include <sys/domain.h>
79 #include <sys/protosw.h>
80 #include <sys/sysctl.h>
81 
82 #include <net/if.h>
83 #include <net/route.h>
84 #include <net/raw_cb.h>
85 
86 #include <machine/stdarg.h>
87 
88 extern	struct domain routedomain;		/* or at least forward */
89 
90 struct	sockaddr route_dst = { 2, PF_ROUTE, };
91 struct	sockaddr route_src = { 2, PF_ROUTE, };
92 struct	sockproto route_proto = { PF_ROUTE, };
93 
94 struct walkarg {
95 	int	w_op;
96 	int	w_arg;
97 	int	w_given;
98 	int	w_needed;
99 	caddr_t	w_where;
100 	int	w_tmemsize;
101 	int	w_tmemneeded;
102 	caddr_t	w_tmem;
103 };
104 
105 static struct mbuf *rt_msg1 __P((int, struct rt_addrinfo *, caddr_t, int));
106 static int rt_msg2 __P((int, struct rt_addrinfo *, caddr_t, struct walkarg *,
107     int *));
108 static int rt_xaddrs __P((caddr_t, caddr_t, struct rt_addrinfo *));
109 static int sysctl_dumpentry __P((struct radix_node *, void *));
110 static int sysctl_iflist __P((int, struct walkarg *, int));
111 static int sysctl_rtable __P((int *, u_int, void *, size_t *, void *, size_t));
112 static __inline void rt_adjustcount __P((int, int));
113 
114 /* Sleazy use of local variables throughout file, warning!!!! */
115 #define dst	info.rti_info[RTAX_DST]
116 #define gate	info.rti_info[RTAX_GATEWAY]
117 #define netmask	info.rti_info[RTAX_NETMASK]
118 #define genmask	info.rti_info[RTAX_GENMASK]
119 #define ifpaddr	info.rti_info[RTAX_IFP]
120 #define ifaaddr	info.rti_info[RTAX_IFA]
121 #define brdaddr	info.rti_info[RTAX_BRD]
122 
123 static __inline void
124 rt_adjustcount(af, cnt)
125 	int af, cnt;
126 {
127 	route_cb.any_count += cnt;
128 	switch (af) {
129 	case AF_INET:
130 		route_cb.ip_count += cnt;
131 		return;
132 #ifdef INET6
133 	case AF_INET6:
134 		route_cb.ip6_count += cnt;
135 		return;
136 #endif
137 	case AF_IPX:
138 		route_cb.ipx_count += cnt;
139 		return;
140 	case AF_NS:
141 		route_cb.ns_count += cnt;
142 		return;
143 	case AF_ISO:
144 		route_cb.iso_count += cnt;
145 		return;
146 	}
147 }
148 
149 /*ARGSUSED*/
150 int
151 route_usrreq(so, req, m, nam, control, p)
152 	struct socket *so;
153 	int req;
154 	struct mbuf *m, *nam, *control;
155 	struct proc *p;
156 {
157 	int error = 0;
158 	struct rawcb *rp = sotorawcb(so);
159 	int s;
160 
161 	if (req == PRU_ATTACH) {
162 		MALLOC(rp, struct rawcb *, sizeof(*rp), M_PCB, M_WAITOK);
163 		if ((so->so_pcb = rp) != NULL)
164 			memset(so->so_pcb, 0, sizeof(*rp));
165 
166 	}
167 	if (req == PRU_DETACH && rp)
168 		rt_adjustcount(rp->rcb_proto.sp_protocol, -1);
169 	s = splsoftnet();
170 
171 	/*
172 	 * Don't call raw_usrreq() in the attach case, because
173 	 * we want to allow non-privileged processes to listen on
174 	 * and send "safe" commands to the routing socket.
175 	 */
176 	if (req == PRU_ATTACH) {
177 		if (p == 0)
178 			error = EACCES;
179 		else
180 			error = raw_attach(so, (int)(long)nam);
181 	} else
182 		error = raw_usrreq(so, req, m, nam, control, p);
183 
184 	rp = sotorawcb(so);
185 	if (req == PRU_ATTACH && rp) {
186 		if (error) {
187 			free((caddr_t)rp, M_PCB);
188 			splx(s);
189 			return (error);
190 		}
191 		rt_adjustcount(rp->rcb_proto.sp_protocol, 1);
192 		rp->rcb_laddr = &route_src;
193 		rp->rcb_faddr = &route_dst;
194 		soisconnected(so);
195 		so->so_options |= SO_USELOOPBACK;
196 	}
197 	splx(s);
198 	return (error);
199 }
200 
201 /*ARGSUSED*/
202 int
203 #if __STDC__
204 route_output(struct mbuf *m, ...)
205 #else
206 route_output(m, va_alist)
207 	struct mbuf *m;
208 	va_dcl
209 #endif
210 {
211 	struct rt_msghdr *rtm = 0;
212 	struct radix_node *rn = 0;
213 	struct rtentry *rt = 0;
214 	struct rtentry *saved_nrt = 0;
215 	struct radix_node_head *rnh;
216 	struct rt_addrinfo info;
217 	int len, error = 0;
218 	struct ifnet *ifp = 0;
219 	struct ifaddr *ifa = 0;
220 	struct socket *so;
221 	va_list ap;
222 	sa_family_t family;
223 
224 	va_start(ap, m);
225 	so = va_arg(ap, struct socket *);
226 	va_end(ap);
227 
228 #define senderr(e) do { error = e; goto flush;} while (/*CONSTCOND*/ 0)
229 	if (m == 0 || ((m->m_len < sizeof(int32_t)) &&
230 	   (m = m_pullup(m, sizeof(int32_t))) == 0))
231 		return (ENOBUFS);
232 	if ((m->m_flags & M_PKTHDR) == 0)
233 		panic("route_output");
234 	len = m->m_pkthdr.len;
235 	if (len < sizeof(*rtm) ||
236 	    len != mtod(m, struct rt_msghdr *)->rtm_msglen) {
237 		dst = 0;
238 		senderr(EINVAL);
239 	}
240 	R_Malloc(rtm, struct rt_msghdr *, len);
241 	if (rtm == 0) {
242 		dst = 0;
243 		senderr(ENOBUFS);
244 	}
245 	m_copydata(m, 0, len, (caddr_t)rtm);
246 	if (rtm->rtm_version != RTM_VERSION) {
247 		dst = 0;
248 		senderr(EPROTONOSUPPORT);
249 	}
250 	rtm->rtm_pid = curproc->p_pid;
251 	memset(&info, 0, sizeof(info));
252 	info.rti_addrs = rtm->rtm_addrs;
253 	if (rt_xaddrs((caddr_t)(rtm + 1), len + (caddr_t)rtm, &info))
254 		senderr(EINVAL);
255 	info.rti_flags = rtm->rtm_flags;
256 	if (dst == 0 || (dst->sa_family >= AF_MAX))
257 		senderr(EINVAL);
258 	if (gate != 0 && (gate->sa_family >= AF_MAX))
259 		senderr(EINVAL);
260 	if (genmask) {
261 		struct radix_node *t;
262 		t = rn_addmask((caddr_t)genmask, 0, 1);
263 		if (t && genmask->sa_len >= ((struct sockaddr *)t->rn_key)->sa_len &&
264 		    Bcmp((caddr_t *)genmask + 1, (caddr_t *)t->rn_key + 1,
265 		    ((struct sockaddr *)t->rn_key)->sa_len) - 1)
266 			genmask = (struct sockaddr *)(t->rn_key);
267 		else
268 			senderr(ENOBUFS);
269 	}
270 
271 	/*
272 	 * Verify that the caller has the appropriate privilege; RTM_GET
273 	 * is the only operation the non-superuser is allowed.
274 	 */
275 	if (rtm->rtm_type != RTM_GET &&
276 	    suser(curproc->p_ucred, &curproc->p_acflag) != 0)
277 		senderr(EACCES);
278 
279 	switch (rtm->rtm_type) {
280 
281 	case RTM_ADD:
282 		if (gate == 0)
283 			senderr(EINVAL);
284 		error = rtrequest1(rtm->rtm_type, &info, &saved_nrt);
285 		if (error == 0 && saved_nrt) {
286 			rt_setmetrics(rtm->rtm_inits,
287 			    &rtm->rtm_rmx, &saved_nrt->rt_rmx);
288 			saved_nrt->rt_refcnt--;
289 			saved_nrt->rt_genmask = genmask;
290 		}
291 		break;
292 
293 	case RTM_DELETE:
294 		error = rtrequest1(rtm->rtm_type, &info, &saved_nrt);
295 		if (error == 0) {
296 			(rt = saved_nrt)->rt_refcnt++;
297 			goto report;
298 		}
299 		break;
300 
301 	case RTM_GET:
302 	case RTM_CHANGE:
303 	case RTM_LOCK:
304 		if ((rnh = rt_tables[dst->sa_family]) == 0) {
305 			senderr(EAFNOSUPPORT);
306 		}
307 		rn = rnh->rnh_lookup(dst, netmask, rnh);
308 		if (rn == NULL || (rn->rn_flags & RNF_ROOT) != 0) {
309 			senderr(ESRCH);
310 		}
311 		rt = (struct rtentry *)rn;
312 		rt->rt_refcnt++;
313 		if (rtm->rtm_type != RTM_GET) {/* XXX: too grotty */
314 			struct radix_node *rn;
315 			extern struct radix_node_head *mask_rnhead;
316 
317 			if (Bcmp(dst, rt_key(rt), dst->sa_len) != 0)
318 				senderr(ESRCH);
319 			if (netmask && (rn = rn_search(netmask,
320 					    mask_rnhead->rnh_treetop)))
321 				netmask = (struct sockaddr *)rn->rn_key;
322 			for (rn = rt->rt_nodes; rn; rn = rn->rn_dupedkey)
323 				if (netmask == (struct sockaddr *)rn->rn_mask)
324 					break;
325 			if (rn == 0)
326 				senderr(ETOOMANYREFS);
327 			rt = (struct rtentry *)rn;
328 		}
329 
330 		switch (rtm->rtm_type) {
331 		case RTM_GET:
332 		report:
333 			dst = rt_key(rt);
334 			gate = rt->rt_gateway;
335 			netmask = rt_mask(rt);
336 			genmask = rt->rt_genmask;
337 			if (rtm->rtm_addrs & (RTA_IFP | RTA_IFA)) {
338 				if ((ifp = rt->rt_ifp) != NULL) {
339 					ifpaddr = TAILQ_FIRST(&ifp->if_addrlist)->ifa_addr;
340 					ifaaddr = rt->rt_ifa->ifa_addr;
341 					if (ifp->if_flags & IFF_POINTOPOINT)
342 						brdaddr = rt->rt_ifa->ifa_dstaddr;
343 					else
344 						brdaddr = 0;
345 					rtm->rtm_index = ifp->if_index;
346 				} else {
347 					ifpaddr = 0;
348 					ifaaddr = 0;
349 				}
350 			}
351 			(void)rt_msg2(rtm->rtm_type, &info, (caddr_t)0,
352 			    (struct walkarg *)0, &len);
353 			if (len > rtm->rtm_msglen) {
354 				struct rt_msghdr *new_rtm;
355 				R_Malloc(new_rtm, struct rt_msghdr *, len);
356 				if (new_rtm == 0)
357 					senderr(ENOBUFS);
358 				Bcopy(rtm, new_rtm, rtm->rtm_msglen);
359 				Free(rtm); rtm = new_rtm;
360 			}
361 			(void)rt_msg2(rtm->rtm_type, &info, (caddr_t)rtm,
362 			    (struct walkarg *)0, 0);
363 			rtm->rtm_flags = rt->rt_flags;
364 			rtm->rtm_rmx = rt->rt_rmx;
365 			rtm->rtm_addrs = info.rti_addrs;
366 			break;
367 
368 		case RTM_CHANGE:
369 			/*
370 			 * new gateway could require new ifaddr, ifp;
371 			 * flags may also be different; ifp may be specified
372 			 * by ll sockaddr when protocol address is ambiguous
373 			 */
374 			if ((error = rt_getifa(&info)) != 0)
375 				senderr(error);
376 			if (gate && rt_setgate(rt, rt_key(rt), gate))
377 				senderr(EDQUOT);
378 			/* new gateway could require new ifaddr, ifp;
379 			   flags may also be different; ifp may be specified
380 			   by ll sockaddr when protocol address is ambiguous */
381 			if (ifpaddr && (ifa = ifa_ifwithnet(ifpaddr)) &&
382 			    (ifp = ifa->ifa_ifp) && (ifaaddr || gate))
383 				ifa = ifaof_ifpforaddr(ifaaddr ? ifaaddr : gate,
384 				    ifp);
385 			else if ((ifaaddr && (ifa = ifa_ifwithaddr(ifaaddr))) ||
386 			    (gate && (ifa = ifa_ifwithroute(rt->rt_flags,
387 			    rt_key(rt), gate))))
388 				ifp = ifa->ifa_ifp;
389 			if (ifa) {
390 				struct ifaddr *oifa = rt->rt_ifa;
391 				if (oifa != ifa) {
392 				    if (oifa && oifa->ifa_rtrequest)
393 					oifa->ifa_rtrequest(RTM_DELETE, rt,
394 					    &info);
395 				    IFAFREE(rt->rt_ifa);
396 				    rt->rt_ifa = ifa;
397 				    IFAREF(rt->rt_ifa);
398 				    rt->rt_ifp = ifp;
399 				}
400 			}
401 			rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx,
402 			    &rt->rt_rmx);
403 			if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest)
404 				rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, &info);
405 			if (genmask)
406 				rt->rt_genmask = genmask;
407 			/*
408 			 * Fall into
409 			 */
410 		case RTM_LOCK:
411 			rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits);
412 			rt->rt_rmx.rmx_locks |=
413 			    (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks);
414 			break;
415 		}
416 		break;
417 
418 	default:
419 		senderr(EOPNOTSUPP);
420 	}
421 
422 flush:
423 	if (rtm) {
424 		if (error)
425 			rtm->rtm_errno = error;
426 		else
427 			rtm->rtm_flags |= RTF_DONE;
428 	}
429 	family = dst ? dst->sa_family : 0;
430 	if (rt)
431 		rtfree(rt);
432     {
433 	struct rawcb *rp = 0;
434 	/*
435 	 * Check to see if we don't want our own messages.
436 	 */
437 	if ((so->so_options & SO_USELOOPBACK) == 0) {
438 		if (route_cb.any_count <= 1) {
439 			if (rtm)
440 				Free(rtm);
441 			m_freem(m);
442 			return (error);
443 		}
444 		/* There is another listener, so construct message */
445 		rp = sotorawcb(so);
446 	}
447 	if (rtm) {
448 		m_copyback(m, 0, rtm->rtm_msglen, (caddr_t)rtm);
449 		if (m->m_pkthdr.len < rtm->rtm_msglen) {
450 			m_freem(m);
451 			m = NULL;
452 		} else if (m->m_pkthdr.len > rtm->rtm_msglen)
453 			m_adj(m, rtm->rtm_msglen - m->m_pkthdr.len);
454 		Free(rtm);
455 	}
456 	if (rp)
457 		rp->rcb_proto.sp_family = 0; /* Avoid us */
458 	if (family)
459 		route_proto.sp_protocol = family;
460 	if (m)
461 		raw_input(m, &route_proto, &route_src, &route_dst);
462 	if (rp)
463 		rp->rcb_proto.sp_family = PF_ROUTE;
464     }
465 	return (error);
466 }
467 
468 void
469 rt_setmetrics(which, in, out)
470 	u_long which;
471 	struct rt_metrics *in, *out;
472 {
473 #define metric(f, e) if (which & (f)) out->e = in->e;
474 	metric(RTV_RPIPE, rmx_recvpipe);
475 	metric(RTV_SPIPE, rmx_sendpipe);
476 	metric(RTV_SSTHRESH, rmx_ssthresh);
477 	metric(RTV_RTT, rmx_rtt);
478 	metric(RTV_RTTVAR, rmx_rttvar);
479 	metric(RTV_HOPCOUNT, rmx_hopcount);
480 	metric(RTV_MTU, rmx_mtu);
481 	metric(RTV_EXPIRE, rmx_expire);
482 #undef metric
483 }
484 
485 #define ROUNDUP(a) \
486 	((a) > 0 ? (1 + (((a) - 1) | (sizeof(long) - 1))) : sizeof(long))
487 #define ADVANCE(x, n) (x += ROUNDUP((n)->sa_len))
488 
489 static int
490 rt_xaddrs(cp, cplim, rtinfo)
491 	caddr_t cp, cplim;
492 	struct rt_addrinfo *rtinfo;
493 {
494 	struct sockaddr *sa = NULL;	/* Quell compiler warning */
495 	int i;
496 
497 	for (i = 0; (i < RTAX_MAX) && (cp < cplim); i++) {
498 		if ((rtinfo->rti_addrs & (1 << i)) == 0)
499 			continue;
500 		rtinfo->rti_info[i] = sa = (struct sockaddr *)cp;
501 		ADVANCE(cp, sa);
502 	}
503 
504 	/* Check for extra addresses specified.  */
505 	if ((rtinfo->rti_addrs & (~0 << i)) != 0)
506 		return (1);
507 	/* Check for bad data length.  */
508 	if (cp != cplim) {
509 		if (i == RTAX_NETMASK + 1 &&
510 		    cp - ROUNDUP(sa->sa_len) + sa->sa_len == cplim)
511 			/*
512 			 * The last sockaddr was netmask.
513 			 * We accept this for now for the sake of old
514 			 * binaries or third party softwares.
515 			 */
516 			;
517 		else
518 			return (1);
519 	}
520 	return (0);
521 }
522 
523 static struct mbuf *
524 rt_msg1(type, rtinfo, data, datalen)
525 	int type;
526 	struct rt_addrinfo *rtinfo;
527 	caddr_t data;
528 	int datalen;
529 {
530 	struct rt_msghdr *rtm;
531 	struct mbuf *m;
532 	int i;
533 	struct sockaddr *sa;
534 	int len, dlen;
535 
536 	m = m_gethdr(M_DONTWAIT, MT_DATA);
537 	if (m == 0)
538 		return (m);
539 	MCLAIM(m, &routedomain.dom_mowner);
540 	switch (type) {
541 
542 	case RTM_DELADDR:
543 	case RTM_NEWADDR:
544 		len = sizeof(struct ifa_msghdr);
545 		break;
546 
547 #ifdef COMPAT_14
548 	case RTM_OIFINFO:
549 		len = sizeof(struct if_msghdr14);
550 		break;
551 #endif
552 
553 	case RTM_IFINFO:
554 		len = sizeof(struct if_msghdr);
555 		break;
556 
557 	case RTM_IFANNOUNCE:
558 		len = sizeof(struct if_announcemsghdr);
559 		break;
560 
561 	default:
562 		len = sizeof(struct rt_msghdr);
563 	}
564 	if (len > MHLEN + MLEN)
565 		panic("rt_msg1: message too long");
566 	else if (len > MHLEN) {
567 		m->m_next = m_get(M_DONTWAIT, MT_DATA);
568 		if (m->m_next == NULL) {
569 			m_freem(m);
570 			return (NULL);
571 		}
572 		MCLAIM(m->m_next, m->m_owner);
573 		m->m_pkthdr.len = len;
574 		m->m_len = MHLEN;
575 		m->m_next->m_len = len - MHLEN;
576 	} else {
577 		m->m_pkthdr.len = m->m_len = len;
578 	}
579 	m->m_pkthdr.rcvif = 0;
580 	m_copyback(m, 0, datalen, data);
581 	rtm = mtod(m, struct rt_msghdr *);
582 	for (i = 0; i < RTAX_MAX; i++) {
583 		if ((sa = rtinfo->rti_info[i]) == NULL)
584 			continue;
585 		rtinfo->rti_addrs |= (1 << i);
586 		dlen = ROUNDUP(sa->sa_len);
587 		m_copyback(m, len, dlen, (caddr_t)sa);
588 		len += dlen;
589 	}
590 	if (m->m_pkthdr.len != len) {
591 		m_freem(m);
592 		return (NULL);
593 	}
594 	rtm->rtm_msglen = len;
595 	rtm->rtm_version = RTM_VERSION;
596 	rtm->rtm_type = type;
597 	return (m);
598 }
599 
600 /*
601  * rt_msg2
602  *
603  *	 fills 'cp' or 'w'.w_tmem with the routing socket message and
604  *		returns the length of the message in 'lenp'.
605  *
606  * if walkarg is 0, cp is expected to be 0 or a buffer large enough to hold
607  *	the message
608  * otherwise walkarg's w_needed is updated and if the user buffer is
609  *	specified and w_needed indicates space exists the information is copied
610  *	into the temp space (w_tmem). w_tmem is [re]allocated if necessary,
611  *	if the allocation fails ENOBUFS is returned.
612  */
613 static int
614 rt_msg2(type, rtinfo, cp, w, lenp)
615 	int type;
616 	struct rt_addrinfo *rtinfo;
617 	caddr_t cp;
618 	struct walkarg *w;
619 	int *lenp;
620 {
621 	int i;
622 	int len, dlen, second_time = 0;
623 	caddr_t cp0;
624 
625 	rtinfo->rti_addrs = 0;
626 again:
627 	switch (type) {
628 
629 	case RTM_DELADDR:
630 	case RTM_NEWADDR:
631 		len = sizeof(struct ifa_msghdr);
632 		break;
633 #ifdef COMPAT_14
634 	case RTM_OIFINFO:
635 		len = sizeof(struct if_msghdr14);
636 		break;
637 #endif
638 
639 	case RTM_IFINFO:
640 		len = sizeof(struct if_msghdr);
641 		break;
642 
643 	default:
644 		len = sizeof(struct rt_msghdr);
645 	}
646 	if ((cp0 = cp) != NULL)
647 		cp += len;
648 	for (i = 0; i < RTAX_MAX; i++) {
649 		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(sa, cp, (unsigned)dlen);
657 			cp += dlen;
658 		}
659 		len += dlen;
660 	}
661 	if (cp == 0 && w != NULL && !second_time) {
662 		struct walkarg *rw = w;
663 
664 		rw->w_needed += len;
665 		if (rw->w_needed <= 0 && rw->w_where) {
666 			if (rw->w_tmemsize < len) {
667 				if (rw->w_tmem)
668 					free(rw->w_tmem, M_RTABLE);
669 				rw->w_tmem = (caddr_t) malloc(len, M_RTABLE,
670 				    M_NOWAIT);
671 				if (rw->w_tmem)
672 					rw->w_tmemsize = len;
673 			}
674 			if (rw->w_tmem) {
675 				cp = rw->w_tmem;
676 				second_time = 1;
677 				goto again;
678 			} else {
679 				rw->w_tmemneeded = len;
680 				return (ENOBUFS);
681 			}
682 		}
683 	}
684 	if (cp) {
685 		struct rt_msghdr *rtm = (struct rt_msghdr *)cp0;
686 
687 		rtm->rtm_version = RTM_VERSION;
688 		rtm->rtm_type = type;
689 		rtm->rtm_msglen = len;
690 	}
691 	if (lenp)
692 		*lenp = len;
693 	return (0);
694 }
695 
696 /*
697  * This routine is called to generate a message from the routing
698  * socket indicating that a redirect has occurred, a routing lookup
699  * has failed, or that a protocol has detected timeouts to a particular
700  * destination.
701  */
702 void
703 rt_missmsg(type, rtinfo, flags, error)
704 	int type, flags, error;
705 	struct rt_addrinfo *rtinfo;
706 {
707 	struct rt_msghdr rtm;
708 	struct mbuf *m;
709 	struct sockaddr *sa = rtinfo->rti_info[RTAX_DST];
710 
711 	if (route_cb.any_count == 0)
712 		return;
713 	memset(&rtm, 0, sizeof(rtm));
714 	rtm.rtm_flags = RTF_DONE | flags;
715 	rtm.rtm_errno = error;
716 	m = rt_msg1(type, rtinfo, (caddr_t)&rtm, sizeof(rtm));
717 	if (m == 0)
718 		return;
719 	mtod(m, struct rt_msghdr *)->rtm_addrs = rtinfo->rti_addrs;
720 	route_proto.sp_protocol = sa ? sa->sa_family : 0;
721 	raw_input(m, &route_proto, &route_src, &route_dst);
722 }
723 
724 /*
725  * This routine is called to generate a message from the routing
726  * socket indicating that the status of a network interface has changed.
727  */
728 void
729 rt_ifmsg(ifp)
730 	struct ifnet *ifp;
731 {
732 	struct if_msghdr ifm;
733 #ifdef COMPAT_14
734 	struct if_msghdr14 oifm;
735 #endif
736 	struct mbuf *m;
737 	struct rt_addrinfo info;
738 
739 	if (route_cb.any_count == 0)
740 		return;
741 	memset(&info, 0, sizeof(info));
742 	memset(&ifm, 0, sizeof(ifm));
743 	ifm.ifm_index = ifp->if_index;
744 	ifm.ifm_flags = ifp->if_flags;
745 	ifm.ifm_data = ifp->if_data;
746 	ifm.ifm_addrs = 0;
747 	m = rt_msg1(RTM_IFINFO, &info, (caddr_t)&ifm, sizeof(ifm));
748 	if (m == 0)
749 		return;
750 	route_proto.sp_protocol = 0;
751 	raw_input(m, &route_proto, &route_src, &route_dst);
752 #ifdef COMPAT_14
753 	memset(&info, 0, sizeof(info));
754 	memset(&oifm, 0, sizeof(oifm));
755 	oifm.ifm_index = ifp->if_index;
756 	oifm.ifm_flags = ifp->if_flags;
757 	oifm.ifm_data.ifi_type = ifp->if_data.ifi_type;
758 	oifm.ifm_data.ifi_addrlen = ifp->if_data.ifi_addrlen;
759 	oifm.ifm_data.ifi_hdrlen = ifp->if_data.ifi_hdrlen;
760 	oifm.ifm_data.ifi_mtu = ifp->if_data.ifi_mtu;
761 	oifm.ifm_data.ifi_metric = ifp->if_data.ifi_metric;
762 	oifm.ifm_data.ifi_baudrate = ifp->if_data.ifi_baudrate;
763 	oifm.ifm_data.ifi_ipackets = ifp->if_data.ifi_ipackets;
764 	oifm.ifm_data.ifi_ierrors = ifp->if_data.ifi_ierrors;
765 	oifm.ifm_data.ifi_opackets = ifp->if_data.ifi_opackets;
766 	oifm.ifm_data.ifi_oerrors = ifp->if_data.ifi_oerrors;
767 	oifm.ifm_data.ifi_collisions = ifp->if_data.ifi_collisions;
768 	oifm.ifm_data.ifi_ibytes = ifp->if_data.ifi_ibytes;
769 	oifm.ifm_data.ifi_obytes = ifp->if_data.ifi_obytes;
770 	oifm.ifm_data.ifi_imcasts = ifp->if_data.ifi_imcasts;
771 	oifm.ifm_data.ifi_omcasts = ifp->if_data.ifi_omcasts;
772 	oifm.ifm_data.ifi_iqdrops = ifp->if_data.ifi_iqdrops;
773 	oifm.ifm_data.ifi_noproto = ifp->if_data.ifi_noproto;
774 	oifm.ifm_data.ifi_lastchange = ifp->if_data.ifi_lastchange;
775 	oifm.ifm_addrs = 0;
776 	m = rt_msg1(RTM_OIFINFO, &info, (caddr_t)&oifm, sizeof(oifm));
777 	if (m == 0)
778 		return;
779 	route_proto.sp_protocol = 0;
780 	raw_input(m, &route_proto, &route_src, &route_dst);
781 #endif
782 }
783 
784 /*
785  * This is called to generate messages from the routing socket
786  * indicating a network interface has had addresses associated with it.
787  * if we ever reverse the logic and replace messages TO the routing
788  * socket indicate a request to configure interfaces, then it will
789  * be unnecessary as the routing socket will automatically generate
790  * copies of it.
791  */
792 void
793 rt_newaddrmsg(cmd, ifa, error, rt)
794 	int cmd, error;
795 	struct ifaddr *ifa;
796 	struct rtentry *rt;
797 {
798 	struct rt_addrinfo info;
799 	struct sockaddr *sa = NULL;
800 	int pass;
801 	struct mbuf *m = NULL;
802 	struct ifnet *ifp = ifa->ifa_ifp;
803 
804 	if (route_cb.any_count == 0)
805 		return;
806 	for (pass = 1; pass < 3; pass++) {
807 		memset(&info, 0, sizeof(info));
808 		if ((cmd == RTM_ADD && pass == 1) ||
809 		    (cmd == RTM_DELETE && pass == 2)) {
810 			struct ifa_msghdr ifam;
811 			int ncmd = cmd == RTM_ADD ? RTM_NEWADDR : RTM_DELADDR;
812 
813 			ifaaddr = sa = ifa->ifa_addr;
814 			ifpaddr = TAILQ_FIRST(&ifp->if_addrlist)->ifa_addr;
815 			netmask = ifa->ifa_netmask;
816 			brdaddr = ifa->ifa_dstaddr;
817 			memset(&ifam, 0, sizeof(ifam));
818 			ifam.ifam_index = ifp->if_index;
819 			ifam.ifam_metric = ifa->ifa_metric;
820 			ifam.ifam_flags = ifa->ifa_flags;
821 			m = rt_msg1(ncmd, &info, (caddr_t)&ifam, sizeof(ifam));
822 			if (m == NULL)
823 				continue;
824 			mtod(m, struct ifa_msghdr *)->ifam_addrs =
825 			    info.rti_addrs;
826 		}
827 		if ((cmd == RTM_ADD && pass == 2) ||
828 		    (cmd == RTM_DELETE && pass == 1)) {
829 			struct rt_msghdr rtm;
830 
831 			if (rt == 0)
832 				continue;
833 			netmask = rt_mask(rt);
834 			dst = sa = rt_key(rt);
835 			gate = rt->rt_gateway;
836 			memset(&rtm, 0, sizeof(rtm));
837 			rtm.rtm_index = ifp->if_index;
838 			rtm.rtm_flags |= rt->rt_flags;
839 			rtm.rtm_errno = error;
840 			m = rt_msg1(cmd, &info, (caddr_t)&rtm, sizeof(rtm));
841 			if (m == NULL)
842 				continue;
843 			mtod(m, struct rt_msghdr *)->rtm_addrs = info.rti_addrs;
844 		}
845 		route_proto.sp_protocol = sa ? sa->sa_family : 0;
846 		raw_input(m, &route_proto, &route_src, &route_dst);
847 	}
848 }
849 
850 /*
851  * This is called to generate routing socket messages indicating
852  * network interface arrival and departure.
853  */
854 void
855 rt_ifannouncemsg(ifp, what)
856 	struct ifnet *ifp;
857 	int what;
858 {
859 	struct if_announcemsghdr ifan;
860 	struct mbuf *m;
861 	struct rt_addrinfo info;
862 
863 	if (route_cb.any_count == 0)
864 		return;
865 	memset(&info, 0, sizeof(info));
866 	memset(&ifan, 0, sizeof(ifan));
867 	ifan.ifan_index = ifp->if_index;
868 	strlcpy(ifan.ifan_name, ifp->if_xname, sizeof(ifan.ifan_name));
869 	ifan.ifan_what = what;
870 	m = rt_msg1(RTM_IFANNOUNCE, &info, (caddr_t)&ifan, sizeof(ifan));
871 	if (m == 0)
872 		return;
873 	route_proto.sp_protocol = 0;
874 	raw_input(m, &route_proto, &route_src, &route_dst);
875 }
876 
877 /*
878  * This is used in dumping the kernel table via sysctl().
879  */
880 static int
881 sysctl_dumpentry(rn, v)
882 	struct radix_node *rn;
883 	void *v;
884 {
885 	struct walkarg *w = v;
886 	struct rtentry *rt = (struct rtentry *)rn;
887 	int error = 0, size;
888 	struct rt_addrinfo info;
889 
890 	if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg))
891 		return 0;
892 	memset(&info, 0, sizeof(info));
893 	dst = rt_key(rt);
894 	gate = rt->rt_gateway;
895 	netmask = rt_mask(rt);
896 	genmask = rt->rt_genmask;
897 	if (rt->rt_ifp) {
898 		ifpaddr = TAILQ_FIRST(&rt->rt_ifp->if_addrlist)->ifa_addr;
899 		ifaaddr = rt->rt_ifa->ifa_addr;
900 		if (rt->rt_ifp->if_flags & IFF_POINTOPOINT)
901 			brdaddr = rt->rt_ifa->ifa_dstaddr;
902 	}
903 	if ((error = rt_msg2(RTM_GET, &info, 0, w, &size)))
904 		return (error);
905 	if (w->w_where && w->w_tmem && w->w_needed <= 0) {
906 		struct rt_msghdr *rtm = (struct rt_msghdr *)w->w_tmem;
907 
908 		rtm->rtm_flags = rt->rt_flags;
909 		rtm->rtm_use = rt->rt_use;
910 		rtm->rtm_rmx = rt->rt_rmx;
911 		rtm->rtm_index = rt->rt_ifp->if_index;
912 		rtm->rtm_errno = rtm->rtm_pid = rtm->rtm_seq = 0;
913 		rtm->rtm_addrs = info.rti_addrs;
914 		if ((error = copyout(rtm, w->w_where, size)) != 0)
915 			w->w_where = NULL;
916 		else
917 			w->w_where += size;
918 	}
919 	return (error);
920 }
921 
922 static int
923 sysctl_iflist(af, w, type)
924 	int	af;
925 	struct	walkarg *w;
926 	int type;
927 {
928 	struct ifnet *ifp;
929 	struct ifaddr *ifa;
930 	struct	rt_addrinfo info;
931 	int	len, error = 0;
932 
933 	memset(&info, 0, sizeof(info));
934 	TAILQ_FOREACH(ifp, &ifnet, if_list) {
935 		if (w->w_arg && w->w_arg != ifp->if_index)
936 			continue;
937 		ifa = TAILQ_FIRST(&ifp->if_addrlist);
938 		ifpaddr = ifa->ifa_addr;
939 		switch (type) {
940 		case NET_RT_IFLIST:
941 			error =
942 			    rt_msg2(RTM_IFINFO, &info, (caddr_t)0, w, &len);
943 			break;
944 #ifdef COMPAT_14
945 		case NET_RT_OIFLIST:
946 			error =
947 			    rt_msg2(RTM_OIFINFO, &info, (caddr_t)0, w, &len);
948 			break;
949 #endif
950 		default:
951 			panic("sysctl_iflist(1)");
952 		}
953 		if (error)
954 			return (error);
955 		ifpaddr = 0;
956 		if (w->w_where && w->w_tmem && w->w_needed <= 0) {
957 			switch (type) {
958 			case NET_RT_IFLIST: {
959 				struct if_msghdr *ifm;
960 
961 				ifm = (struct if_msghdr *)w->w_tmem;
962 				ifm->ifm_index = ifp->if_index;
963 				ifm->ifm_flags = ifp->if_flags;
964 				ifm->ifm_data = ifp->if_data;
965 				ifm->ifm_addrs = info.rti_addrs;
966 				error = copyout(ifm, w->w_where, len);
967 				if (error)
968 					return (error);
969 				w->w_where += len;
970 				break;
971 			}
972 
973 #ifdef COMPAT_14
974 			case NET_RT_OIFLIST: {
975 				struct if_msghdr14 *ifm;
976 
977 				ifm = (struct if_msghdr14 *)w->w_tmem;
978 				ifm->ifm_index = ifp->if_index;
979 				ifm->ifm_flags = ifp->if_flags;
980 				ifm->ifm_data.ifi_type = ifp->if_data.ifi_type;
981 				ifm->ifm_data.ifi_addrlen =
982 				    ifp->if_data.ifi_addrlen;
983 				ifm->ifm_data.ifi_hdrlen =
984 				    ifp->if_data.ifi_hdrlen;
985 				ifm->ifm_data.ifi_mtu = ifp->if_data.ifi_mtu;
986 				ifm->ifm_data.ifi_metric =
987 				    ifp->if_data.ifi_metric;
988 				ifm->ifm_data.ifi_baudrate =
989 				    ifp->if_data.ifi_baudrate;
990 				ifm->ifm_data.ifi_ipackets =
991 				    ifp->if_data.ifi_ipackets;
992 				ifm->ifm_data.ifi_ierrors =
993 				    ifp->if_data.ifi_ierrors;
994 				ifm->ifm_data.ifi_opackets =
995 				    ifp->if_data.ifi_opackets;
996 				ifm->ifm_data.ifi_oerrors =
997 				    ifp->if_data.ifi_oerrors;
998 				ifm->ifm_data.ifi_collisions =
999 				    ifp->if_data.ifi_collisions;
1000 				ifm->ifm_data.ifi_ibytes =
1001 				    ifp->if_data.ifi_ibytes;
1002 				ifm->ifm_data.ifi_obytes =
1003 				    ifp->if_data.ifi_obytes;
1004 				ifm->ifm_data.ifi_imcasts =
1005 				    ifp->if_data.ifi_imcasts;
1006 				ifm->ifm_data.ifi_omcasts =
1007 				    ifp->if_data.ifi_omcasts;
1008 				ifm->ifm_data.ifi_iqdrops =
1009 				    ifp->if_data.ifi_iqdrops;
1010 				ifm->ifm_data.ifi_noproto =
1011 				    ifp->if_data.ifi_noproto;
1012 				ifm->ifm_data.ifi_lastchange =
1013 				    ifp->if_data.ifi_lastchange;
1014 				ifm->ifm_addrs = info.rti_addrs;
1015 				error = copyout(ifm, w->w_where, len);
1016 				if (error)
1017 					return (error);
1018 				w->w_where += len;
1019 				break;
1020 			}
1021 #endif
1022 			default:
1023 				panic("sysctl_iflist(2)");
1024 			}
1025 		}
1026 		while ((ifa = TAILQ_NEXT(ifa, ifa_list)) != NULL) {
1027 			if (af && af != ifa->ifa_addr->sa_family)
1028 				continue;
1029 			ifaaddr = ifa->ifa_addr;
1030 			netmask = ifa->ifa_netmask;
1031 			brdaddr = ifa->ifa_dstaddr;
1032 			if ((error = rt_msg2(RTM_NEWADDR, &info, 0, w, &len)))
1033 				return (error);
1034 			if (w->w_where && w->w_tmem && w->w_needed <= 0) {
1035 				struct ifa_msghdr *ifam;
1036 
1037 				ifam = (struct ifa_msghdr *)w->w_tmem;
1038 				ifam->ifam_index = ifa->ifa_ifp->if_index;
1039 				ifam->ifam_flags = ifa->ifa_flags;
1040 				ifam->ifam_metric = ifa->ifa_metric;
1041 				ifam->ifam_addrs = info.rti_addrs;
1042 				error = copyout(w->w_tmem, w->w_where, len);
1043 				if (error)
1044 					return (error);
1045 				w->w_where += len;
1046 			}
1047 		}
1048 		ifaaddr = netmask = brdaddr = 0;
1049 	}
1050 	return (0);
1051 }
1052 
1053 static int
1054 sysctl_rtable(name, namelen, where, given, new, newlen)
1055 	int	*name;
1056 	u_int	namelen;
1057 	void 	*where;
1058 	size_t	*given;
1059 	void	*new;
1060 	size_t	newlen;
1061 {
1062 	struct radix_node_head *rnh;
1063 	int	i, s, error = EINVAL;
1064 	u_char  af;
1065 	struct	walkarg w;
1066 
1067 	if (new)
1068 		return (EPERM);
1069 	if (namelen != 3)
1070 		return (EINVAL);
1071 	af = name[0];
1072 	w.w_tmemneeded = 0;
1073 	w.w_tmemsize = 0;
1074 	w.w_tmem = NULL;
1075 again:
1076 	/* we may return here if a later [re]alloc of the t_mem buffer fails */
1077 	if (w.w_tmemneeded) {
1078 		w.w_tmem = (caddr_t) malloc(w.w_tmemneeded, M_RTABLE, M_WAITOK);
1079 		w.w_tmemsize = w.w_tmemneeded;
1080 		w.w_tmemneeded = 0;
1081 	}
1082 	w.w_op = name[1];
1083 	w.w_arg = name[2];
1084 	w.w_given = *given;
1085 	w.w_needed = 0 - w.w_given;
1086 	w.w_where = where;
1087 
1088 	s = splsoftnet();
1089 	switch (w.w_op) {
1090 
1091 	case NET_RT_DUMP:
1092 	case NET_RT_FLAGS:
1093 		for (i = 1; i <= AF_MAX; i++)
1094 			if ((rnh = rt_tables[i]) && (af == 0 || af == i) &&
1095 			    (error = (*rnh->rnh_walktree)(rnh,
1096 			    sysctl_dumpentry, &w)))
1097 				break;
1098 		break;
1099 
1100 #ifdef COMPAT_14
1101 	case NET_RT_OIFLIST:
1102 		error = sysctl_iflist(af, &w, w.w_op);
1103 		break;
1104 #endif
1105 
1106 	case NET_RT_IFLIST:
1107 		error = sysctl_iflist(af, &w, w.w_op);
1108 	}
1109 	splx(s);
1110 
1111 	/* check to see if we couldn't allocate memory with NOWAIT */
1112 	if (error == ENOBUFS && w.w_tmem == 0 && w.w_tmemneeded)
1113 		goto again;
1114 
1115 	if (w.w_tmem)
1116 		free(w.w_tmem, M_RTABLE);
1117 	w.w_needed += w.w_given;
1118 	if (where) {
1119 		*given = w.w_where - (caddr_t) where;
1120 		if (*given < w.w_needed)
1121 			return (ENOMEM);
1122 	} else {
1123 		*given = (11 * w.w_needed) / 10;
1124 	}
1125 	return (error);
1126 }
1127 
1128 /*
1129  * Definitions of protocols supported in the ROUTE domain.
1130  */
1131 
1132 struct protosw routesw[] = {
1133 { SOCK_RAW,	&routedomain,	0,		PR_ATOMIC|PR_ADDR,
1134   raw_input,	route_output,	raw_ctlinput,	0,
1135   route_usrreq,
1136   raw_init,	0,		0,		0,
1137   sysctl_rtable,
1138 }
1139 };
1140 
1141 struct domain routedomain =
1142     { PF_ROUTE, "route", route_init, 0, 0,
1143       routesw, &routesw[sizeof(routesw)/sizeof(routesw[0])] };
1144