xref: /netbsd-src/sys/net/rtsock.c (revision 89c5a767f8fc7a4633b2d409966e2becbb98ff92)
1 /*	$NetBSD: rtsock.c,v 1.38 2000/03/12 11:58:15 itojun 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 "opt_inet.h"
68 
69 #include <sys/param.h>
70 #include <sys/systm.h>
71 #include <sys/proc.h>
72 #include <sys/mbuf.h>
73 #include <sys/socket.h>
74 #include <sys/socketvar.h>
75 #include <sys/domain.h>
76 #include <sys/protosw.h>
77 
78 #include <vm/vm.h>
79 #include <sys/sysctl.h>
80 
81 #include <net/if.h>
82 #include <net/route.h>
83 #include <net/raw_cb.h>
84 
85 #include <machine/stdarg.h>
86 
87 struct	sockaddr route_dst = { 2, PF_ROUTE, };
88 struct	sockaddr route_src = { 2, PF_ROUTE, };
89 struct	sockproto route_proto = { PF_ROUTE, };
90 
91 struct walkarg {
92 	int	w_op;
93 	int	w_arg;
94 	int	w_given;
95 	int	w_needed;
96 	caddr_t	w_where;
97 	int	w_tmemsize;
98 	int	w_tmemneeded;
99 	caddr_t	w_tmem;
100 };
101 
102 static struct mbuf *rt_msg1 __P((int, struct rt_addrinfo *, caddr_t, int));
103 static int rt_msg2 __P((int, struct rt_addrinfo *, caddr_t, struct walkarg *,
104     int *));
105 static void rt_xaddrs __P((caddr_t, caddr_t, struct rt_addrinfo *));
106 static __inline void rt_adjustcount __P((int, int));
107 
108 /* Sleazy use of local variables throughout file, warning!!!! */
109 #define dst	info.rti_info[RTAX_DST]
110 #define gate	info.rti_info[RTAX_GATEWAY]
111 #define netmask	info.rti_info[RTAX_NETMASK]
112 #define genmask	info.rti_info[RTAX_GENMASK]
113 #define ifpaddr	info.rti_info[RTAX_IFP]
114 #define ifaaddr	info.rti_info[RTAX_IFA]
115 #define brdaddr	info.rti_info[RTAX_BRD]
116 
117 static __inline void
118 rt_adjustcount(af, cnt)
119 	int af, cnt;
120 {
121 	route_cb.any_count += cnt;
122 	switch (af) {
123 	case AF_INET:
124 		route_cb.ip_count += cnt;
125 		return;
126 #ifdef INET6
127 	case AF_INET6:
128 		route_cb.ip6_count += cnt;
129 		return;
130 #endif
131 	case AF_IPX:
132 		route_cb.ipx_count += cnt;
133 		return;
134 	case AF_NS:
135 		route_cb.ns_count += cnt;
136 		return;
137 	case AF_ISO:
138 		route_cb.iso_count += cnt;
139 		return;
140 	}
141 }
142 
143 /*ARGSUSED*/
144 int
145 route_usrreq(so, req, m, nam, control, p)
146 	register struct socket *so;
147 	int req;
148 	struct mbuf *m, *nam, *control;
149 	struct proc *p;
150 {
151 	register int error = 0;
152 	register struct rawcb *rp = sotorawcb(so);
153 	int s;
154 
155 	if (req == PRU_ATTACH) {
156 		MALLOC(rp, struct rawcb *, sizeof(*rp), M_PCB, M_WAITOK);
157 		if ((so->so_pcb = rp) != NULL)
158 			bzero(so->so_pcb, sizeof(*rp));
159 
160 	}
161 	if (req == PRU_DETACH && rp)
162 		rt_adjustcount(rp->rcb_proto.sp_protocol, -1);
163 	s = splsoftnet();
164 
165 	/*
166 	 * Don't call raw_usrreq() in the attach case, because
167 	 * we want to allow non-privileged processes to listen on
168 	 * and send "safe" commands to the routing socket.
169 	 */
170 	if (req == PRU_ATTACH) {
171 		if (p == 0)
172 			error = EACCES;
173 		else
174 			error = raw_attach(so, (int)(long)nam);
175 	} else
176 		error = raw_usrreq(so, req, m, nam, control, p);
177 
178 	rp = sotorawcb(so);
179 	if (req == PRU_ATTACH && rp) {
180 		if (error) {
181 			free((caddr_t)rp, M_PCB);
182 			splx(s);
183 			return (error);
184 		}
185 		rt_adjustcount(rp->rcb_proto.sp_protocol, 1);
186 		rp->rcb_laddr = &route_src;
187 		rp->rcb_faddr = &route_dst;
188 		soisconnected(so);
189 		so->so_options |= SO_USELOOPBACK;
190 	}
191 	splx(s);
192 	return (error);
193 }
194 
195 /*ARGSUSED*/
196 int
197 #if __STDC__
198 route_output(struct mbuf *m, ...)
199 #else
200 route_output(m, va_alist)
201 	struct mbuf *m;
202 	va_dcl
203 #endif
204 {
205 	register struct rt_msghdr *rtm = 0;
206 	register struct radix_node *rn = 0;
207 	register struct rtentry *rt = 0;
208 	struct rtentry *saved_nrt = 0;
209 	struct radix_node_head *rnh;
210 	struct rt_addrinfo info;
211 	int len, error = 0;
212 	struct ifnet *ifp = 0;
213 	struct ifaddr *ifa = 0;
214 	struct socket *so;
215 	va_list ap;
216 
217 	va_start(ap, m);
218 	so = va_arg(ap, struct socket *);
219 	va_end(ap);
220 
221 	bzero(&info, sizeof(info));
222 #define senderr(e) do { error = e; goto flush;} while (0)
223 	if (m == 0 || ((m->m_len < sizeof(int32_t)) &&
224 	   (m = m_pullup(m, sizeof(int32_t))) == 0))
225 		return (ENOBUFS);
226 	if ((m->m_flags & M_PKTHDR) == 0)
227 		panic("route_output");
228 	len = m->m_pkthdr.len;
229 	if (len < sizeof(*rtm) ||
230 	    len != mtod(m, struct rt_msghdr *)->rtm_msglen) {
231 		dst = 0;
232 		senderr(EINVAL);
233 	}
234 	R_Malloc(rtm, struct rt_msghdr *, len);
235 	if (rtm == 0) {
236 		dst = 0;
237 		senderr(ENOBUFS);
238 	}
239 	m_copydata(m, 0, len, (caddr_t)rtm);
240 	if (rtm->rtm_version != RTM_VERSION) {
241 		dst = 0;
242 		senderr(EPROTONOSUPPORT);
243 	}
244 	rtm->rtm_pid = curproc->p_pid;
245 	info.rti_addrs = rtm->rtm_addrs;
246 	rt_xaddrs((caddr_t)(rtm + 1), len + (caddr_t)rtm, &info);
247 	if (dst == 0 || (dst->sa_family >= AF_MAX))
248 		senderr(EINVAL);
249 	if (gate != 0 && (gate->sa_family >= AF_MAX))
250 		senderr(EINVAL);
251 	if (genmask) {
252 		struct radix_node *t;
253 		t = rn_addmask((caddr_t)genmask, 0, 1);
254 		if (t && Bcmp(genmask, t->rn_key, *(u_char *)genmask) == 0)
255 			genmask = (struct sockaddr *)(t->rn_key);
256 		else
257 			senderr(ENOBUFS);
258 	}
259 
260 	/*
261 	 * Verify that the caller has the appropriate privilege; RTM_GET
262 	 * is the only operation the non-superuser is allowed.
263 	 */
264 	if (rtm->rtm_type != RTM_GET &&
265 	    suser(curproc->p_ucred, &curproc->p_acflag) != 0)
266 		senderr(EACCES);
267 
268 	switch (rtm->rtm_type) {
269 
270 	case RTM_ADD:
271 		if (gate == 0)
272 			senderr(EINVAL);
273 		error = rtrequest(RTM_ADD, dst, gate, netmask,
274 		    rtm->rtm_flags, &saved_nrt);
275 		if (error == 0 && saved_nrt) {
276 			rt_setmetrics(rtm->rtm_inits,
277 			    &rtm->rtm_rmx, &saved_nrt->rt_rmx);
278 			saved_nrt->rt_refcnt--;
279 			saved_nrt->rt_genmask = genmask;
280 		}
281 		break;
282 
283 	case RTM_DELETE:
284 		error = rtrequest(RTM_DELETE, dst, gate, netmask,
285 		    rtm->rtm_flags, &saved_nrt);
286 		if (error == 0) {
287 			(rt = saved_nrt)->rt_refcnt++;
288 			goto report;
289 		}
290 		break;
291 
292 	case RTM_GET:
293 	case RTM_CHANGE:
294 	case RTM_LOCK:
295 		if ((rnh = rt_tables[dst->sa_family]) == 0) {
296 			senderr(EAFNOSUPPORT);
297 		}
298 		rn = rnh->rnh_lookup(dst, netmask, rnh);
299 		if (rn == NULL || (rn->rn_flags & RNF_ROOT) != 0) {
300 			senderr(ESRCH);
301 		}
302 		rt = (struct rtentry *)rn;
303 		rt->rt_refcnt++;
304 
305 		switch(rtm->rtm_type) {
306 
307 		case RTM_GET:
308 		report:
309 			dst = rt_key(rt);
310 			gate = rt->rt_gateway;
311 			netmask = rt_mask(rt);
312 			genmask = rt->rt_genmask;
313 			if (rtm->rtm_addrs & (RTA_IFP | RTA_IFA)) {
314 				if ((ifp = rt->rt_ifp) != NULL) {
315 					ifpaddr = ifp->if_addrlist.tqh_first->ifa_addr;
316 					ifaaddr = rt->rt_ifa->ifa_addr;
317 					if (ifp->if_flags & IFF_POINTOPOINT)
318 						brdaddr = rt->rt_ifa->ifa_dstaddr;
319 					else
320 						brdaddr = 0;
321 					rtm->rtm_index = ifp->if_index;
322 				} else {
323 					ifpaddr = 0;
324 					ifaaddr = 0;
325 				}
326 			}
327 			(void)rt_msg2(rtm->rtm_type, &info, (caddr_t)0,
328 			    (struct walkarg *)0, &len);
329 			if (len > rtm->rtm_msglen) {
330 				struct rt_msghdr *new_rtm;
331 				R_Malloc(new_rtm, struct rt_msghdr *, len);
332 				if (new_rtm == 0)
333 					senderr(ENOBUFS);
334 				Bcopy(rtm, new_rtm, rtm->rtm_msglen);
335 				Free(rtm); rtm = new_rtm;
336 			}
337 			(void)rt_msg2(rtm->rtm_type, &info, (caddr_t)rtm,
338 			    (struct walkarg *)0, 0);
339 			rtm->rtm_flags = rt->rt_flags;
340 			rtm->rtm_rmx = rt->rt_rmx;
341 			rtm->rtm_addrs = info.rti_addrs;
342 			break;
343 
344 		case RTM_CHANGE:
345 			if (gate && rt_setgate(rt, rt_key(rt), gate))
346 				senderr(EDQUOT);
347 			/* new gateway could require new ifaddr, ifp;
348 			   flags may also be different; ifp may be specified
349 			   by ll sockaddr when protocol address is ambiguous */
350 			if (ifpaddr && (ifa = ifa_ifwithnet(ifpaddr)) &&
351 			    (ifp = ifa->ifa_ifp) && (ifaaddr || gate))
352 				ifa = ifaof_ifpforaddr(ifaaddr ? ifaaddr : gate,
353 				    ifp);
354 			else if ((ifaaddr && (ifa = ifa_ifwithaddr(ifaaddr))) ||
355 			    (gate && (ifa = ifa_ifwithroute(rt->rt_flags,
356 			    rt_key(rt), gate))))
357 				ifp = ifa->ifa_ifp;
358 			if (ifa) {
359 				register struct ifaddr *oifa = rt->rt_ifa;
360 				if (oifa != ifa) {
361 				    if (oifa && oifa->ifa_rtrequest)
362 					oifa->ifa_rtrequest(RTM_DELETE,
363 					rt, gate);
364 				    IFAFREE(rt->rt_ifa);
365 				    rt->rt_ifa = ifa;
366 				    IFAREF(rt->rt_ifa);
367 				    rt->rt_ifp = ifp;
368 				}
369 			}
370 			rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx,
371 			    &rt->rt_rmx);
372 			if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest)
373 				rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, gate);
374 			if (genmask)
375 				rt->rt_genmask = genmask;
376 			/*
377 			 * Fall into
378 			 */
379 		case RTM_LOCK:
380 			rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits);
381 			rt->rt_rmx.rmx_locks |=
382 			    (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks);
383 			break;
384 		}
385 		break;
386 
387 	default:
388 		senderr(EOPNOTSUPP);
389 	}
390 
391 flush:
392 	if (rtm) {
393 		if (error)
394 			rtm->rtm_errno = error;
395 		else
396 			rtm->rtm_flags |= RTF_DONE;
397 	}
398 	if (rt)
399 		rtfree(rt);
400     {
401 	register struct rawcb *rp = 0;
402 	/*
403 	 * Check to see if we don't want our own messages.
404 	 */
405 	if ((so->so_options & SO_USELOOPBACK) == 0) {
406 		if (route_cb.any_count <= 1) {
407 			if (rtm)
408 				Free(rtm);
409 			m_freem(m);
410 			return (error);
411 		}
412 		/* There is another listener, so construct message */
413 		rp = sotorawcb(so);
414 	}
415 	if (rtm) {
416 		m_copyback(m, 0, rtm->rtm_msglen, (caddr_t)rtm);
417 		Free(rtm);
418 	}
419 	if (rp)
420 		rp->rcb_proto.sp_family = 0; /* Avoid us */
421 	if (dst)
422 		route_proto.sp_protocol = dst->sa_family;
423 	raw_input(m, &route_proto, &route_src, &route_dst);
424 	if (rp)
425 		rp->rcb_proto.sp_family = PF_ROUTE;
426     }
427 	return (error);
428 }
429 
430 void
431 rt_setmetrics(which, in, out)
432 	u_long which;
433 	register struct rt_metrics *in, *out;
434 {
435 #define metric(f, e) if (which & (f)) out->e = in->e;
436 	metric(RTV_RPIPE, rmx_recvpipe);
437 	metric(RTV_SPIPE, rmx_sendpipe);
438 	metric(RTV_SSTHRESH, rmx_ssthresh);
439 	metric(RTV_RTT, rmx_rtt);
440 	metric(RTV_RTTVAR, rmx_rttvar);
441 	metric(RTV_HOPCOUNT, rmx_hopcount);
442 	metric(RTV_MTU, rmx_mtu);
443 	metric(RTV_EXPIRE, rmx_expire);
444 #undef metric
445 }
446 
447 #define ROUNDUP(a) \
448 	((a) > 0 ? (1 + (((a) - 1) | (sizeof(long) - 1))) : sizeof(long))
449 #define ADVANCE(x, n) (x += ROUNDUP((n)->sa_len))
450 
451 static void
452 rt_xaddrs(cp, cplim, rtinfo)
453 	register caddr_t cp, cplim;
454 	register struct rt_addrinfo *rtinfo;
455 {
456 	register struct sockaddr *sa;
457 	register int i;
458 
459 	bzero(rtinfo->rti_info, sizeof(rtinfo->rti_info));
460 	for (i = 0; (i < RTAX_MAX) && (cp < cplim); i++) {
461 		if ((rtinfo->rti_addrs & (1 << i)) == 0)
462 			continue;
463 		rtinfo->rti_info[i] = sa = (struct sockaddr *)cp;
464 		ADVANCE(cp, sa);
465 	}
466 }
467 
468 static struct mbuf *
469 rt_msg1(type, rtinfo, data, datalen)
470 	int type;
471 	register struct rt_addrinfo *rtinfo;
472 	caddr_t data;
473 	int datalen;
474 {
475 	register struct rt_msghdr *rtm;
476 	register struct mbuf *m;
477 	register int i;
478 	register struct sockaddr *sa;
479 	int len, dlen;
480 
481 	m = m_gethdr(M_DONTWAIT, MT_DATA);
482 	if (m == 0)
483 		return (m);
484 	switch (type) {
485 
486 	case RTM_DELADDR:
487 	case RTM_NEWADDR:
488 		len = sizeof(struct ifa_msghdr);
489 		break;
490 
491 #ifdef COMPAT_14
492 	case RTM_OIFINFO:
493 		len = sizeof(struct if_msghdr14);
494 		break;
495 #endif
496 
497 	case RTM_IFINFO:
498 		len = sizeof(struct if_msghdr);
499 		break;
500 
501 	case RTM_IFANNOUNCE:
502 		len = sizeof(struct if_announcemsghdr);
503 		break;
504 
505 	default:
506 		len = sizeof(struct rt_msghdr);
507 	}
508 	if (len > MHLEN + MLEN)
509 		panic("rt_msg1: message too long");
510 	else if (len > MHLEN) {
511 		m->m_next = m_get(M_DONTWAIT, MT_DATA);
512 		if (m->m_next == NULL) {
513 			m_freem(m);
514 			return (NULL);
515 		}
516 		m->m_pkthdr.len = len;
517 		m->m_len = MHLEN;
518 		m->m_next->m_len = len - MHLEN;
519 	} else {
520 		m->m_pkthdr.len = m->m_len = len;
521 	}
522 	m->m_pkthdr.rcvif = 0;
523 	m_copyback(m, 0, datalen, data);
524 	rtm = mtod(m, struct rt_msghdr *);
525 	for (i = 0; i < RTAX_MAX; i++) {
526 		if ((sa = rtinfo->rti_info[i]) == NULL)
527 			continue;
528 		rtinfo->rti_addrs |= (1 << i);
529 		dlen = ROUNDUP(sa->sa_len);
530 		m_copyback(m, len, dlen, (caddr_t)sa);
531 		len += dlen;
532 	}
533 	rtm->rtm_msglen = len;
534 	rtm->rtm_version = RTM_VERSION;
535 	rtm->rtm_type = type;
536 	return (m);
537 }
538 
539 /*
540  * rt_msg2
541  *
542  *	 fills 'cp' or 'w'.w_tmem with the routing socket message and
543  *		returns the length of the message in 'lenp'.
544  *
545  * if walkarg is 0, cp is expected to be 0 or a buffer large enough to hold
546  *	the message
547  * otherwise walkarg's w_needed is updated and if the user buffer is
548  *	specified and w_needed indicates space exists the information is copied
549  *	into the temp space (w_tmem). w_tmem is [re]allocated if necessary,
550  *	if the allocation fails ENOBUFS is returned.
551  */
552 static int
553 rt_msg2(type, rtinfo, cp, w, lenp)
554 	int type;
555 	register struct rt_addrinfo *rtinfo;
556 	caddr_t cp;
557 	struct walkarg *w;
558 	int *lenp;
559 {
560 	register int i;
561 	int len, dlen, second_time = 0;
562 	caddr_t cp0;
563 
564 	rtinfo->rti_addrs = 0;
565 again:
566 	switch (type) {
567 
568 	case RTM_DELADDR:
569 	case RTM_NEWADDR:
570 		len = sizeof(struct ifa_msghdr);
571 		break;
572 #ifdef COMPAT_14
573 	case RTM_OIFINFO:
574 		len = sizeof(struct if_msghdr14);
575 		break;
576 #endif
577 
578 	case RTM_IFINFO:
579 		len = sizeof(struct if_msghdr);
580 		break;
581 
582 	default:
583 		len = sizeof(struct rt_msghdr);
584 	}
585 	if ((cp0 = cp) != NULL)
586 		cp += len;
587 	for (i = 0; i < RTAX_MAX; i++) {
588 		register struct sockaddr *sa;
589 
590 		if ((sa = rtinfo->rti_info[i]) == 0)
591 			continue;
592 		rtinfo->rti_addrs |= (1 << i);
593 		dlen = ROUNDUP(sa->sa_len);
594 		if (cp) {
595 			bcopy(sa, cp, (unsigned)dlen);
596 			cp += dlen;
597 		}
598 		len += dlen;
599 	}
600 	if (cp == 0 && w != NULL && !second_time) {
601 		register struct walkarg *rw = w;
602 
603 		rw->w_needed += len;
604 		if (rw->w_needed <= 0 && rw->w_where) {
605 			if (rw->w_tmemsize < len) {
606 				if (rw->w_tmem)
607 					free(rw->w_tmem, M_RTABLE);
608 				rw->w_tmem = (caddr_t) malloc(len, M_RTABLE,
609 				    M_NOWAIT);
610 				if (rw->w_tmem)
611 					rw->w_tmemsize = len;
612 			}
613 			if (rw->w_tmem) {
614 				cp = rw->w_tmem;
615 				second_time = 1;
616 				goto again;
617 			} else {
618 				rw->w_tmemneeded = len;
619 				return (ENOBUFS);
620 			}
621 		}
622 	}
623 	if (cp) {
624 		register struct rt_msghdr *rtm = (struct rt_msghdr *)cp0;
625 
626 		rtm->rtm_version = RTM_VERSION;
627 		rtm->rtm_type = type;
628 		rtm->rtm_msglen = len;
629 	}
630 	if (lenp)
631 		*lenp = len;
632 	return (0);
633 }
634 
635 /*
636  * This routine is called to generate a message from the routing
637  * socket indicating that a redirect has occured, a routing lookup
638  * has failed, or that a protocol has detected timeouts to a particular
639  * destination.
640  */
641 void
642 rt_missmsg(type, rtinfo, flags, error)
643 	int type, flags, error;
644 	register struct rt_addrinfo *rtinfo;
645 {
646 	struct rt_msghdr rtm;
647 	register struct mbuf *m;
648 	struct sockaddr *sa = rtinfo->rti_info[RTAX_DST];
649 
650 	if (route_cb.any_count == 0)
651 		return;
652 	bzero(&rtm, sizeof(rtm));
653 	rtm.rtm_flags = RTF_DONE | flags;
654 	rtm.rtm_errno = error;
655 	m = rt_msg1(type, rtinfo, (caddr_t)&rtm, sizeof(rtm));
656 	if (m == 0)
657 		return;
658 	mtod(m, struct rt_msghdr *)->rtm_addrs = rtinfo->rti_addrs;
659 	route_proto.sp_protocol = sa ? sa->sa_family : 0;
660 	raw_input(m, &route_proto, &route_src, &route_dst);
661 }
662 
663 /*
664  * This routine is called to generate a message from the routing
665  * socket indicating that the status of a network interface has changed.
666  */
667 void
668 rt_ifmsg(ifp)
669 	register struct ifnet *ifp;
670 {
671 	struct if_msghdr ifm;
672 #ifdef COMPAT_14
673 	struct if_msghdr14 oifm;
674 #endif
675 	struct mbuf *m;
676 	struct rt_addrinfo info;
677 
678 	if (route_cb.any_count == 0)
679 		return;
680 	bzero(&info, sizeof(info));
681 	bzero(&ifm, sizeof(ifm));
682 	ifm.ifm_index = ifp->if_index;
683 	ifm.ifm_flags = ifp->if_flags;
684 	ifm.ifm_data = ifp->if_data;
685 	ifm.ifm_addrs = 0;
686 	m = rt_msg1(RTM_IFINFO, &info, (caddr_t)&ifm, sizeof(ifm));
687 	if (m == 0)
688 		return;
689 	route_proto.sp_protocol = 0;
690 	raw_input(m, &route_proto, &route_src, &route_dst);
691 #ifdef COMPAT_14
692 	bzero(&info, sizeof(info));
693 	bzero(&oifm, sizeof(ifm));
694 	oifm.ifm_index = ifp->if_index;
695 	oifm.ifm_flags = ifp->if_flags;
696 	oifm.ifm_data.ifi_type = ifp->if_data.ifi_type;
697 	oifm.ifm_data.ifi_addrlen = ifp->if_data.ifi_addrlen;
698 	oifm.ifm_data.ifi_hdrlen = ifp->if_data.ifi_hdrlen;
699 	oifm.ifm_data.ifi_mtu = ifp->if_data.ifi_mtu;
700 	oifm.ifm_data.ifi_metric = ifp->if_data.ifi_metric;
701 	oifm.ifm_data.ifi_baudrate = ifp->if_data.ifi_baudrate;
702 	oifm.ifm_data.ifi_ipackets = ifp->if_data.ifi_ipackets;
703 	oifm.ifm_data.ifi_ierrors = ifp->if_data.ifi_ierrors;
704 	oifm.ifm_data.ifi_opackets = ifp->if_data.ifi_opackets;
705 	oifm.ifm_data.ifi_oerrors = ifp->if_data.ifi_oerrors;
706 	oifm.ifm_data.ifi_collisions = ifp->if_data.ifi_collisions;
707 	oifm.ifm_data.ifi_ibytes = ifp->if_data.ifi_ibytes;
708 	oifm.ifm_data.ifi_obytes = ifp->if_data.ifi_obytes;
709 	oifm.ifm_data.ifi_imcasts = ifp->if_data.ifi_imcasts;
710 	oifm.ifm_data.ifi_omcasts = ifp->if_data.ifi_omcasts;
711 	oifm.ifm_data.ifi_iqdrops = ifp->if_data.ifi_iqdrops;
712 	oifm.ifm_data.ifi_noproto = ifp->if_data.ifi_noproto;
713 	oifm.ifm_data.ifi_lastchange = ifp->if_data.ifi_lastchange;
714 	oifm.ifm_addrs = 0;
715 	m = rt_msg1(RTM_OIFINFO, &info, (caddr_t)&oifm, sizeof(oifm));
716 	if (m == 0)
717 		return;
718 	route_proto.sp_protocol = 0;
719 	raw_input(m, &route_proto, &route_src, &route_dst);
720 #endif
721 }
722 
723 /*
724  * This is called to generate messages from the routing socket
725  * indicating a network interface has had addresses associated with it.
726  * if we ever reverse the logic and replace messages TO the routing
727  * socket indicate a request to configure interfaces, then it will
728  * be unnecessary as the routing socket will automatically generate
729  * copies of it.
730  */
731 void
732 rt_newaddrmsg(cmd, ifa, error, rt)
733 	int cmd, error;
734 	register struct ifaddr *ifa;
735 	register struct rtentry *rt;
736 {
737 	struct rt_addrinfo info;
738 	struct sockaddr *sa = NULL;
739 	int pass;
740 	struct mbuf *m = NULL;
741 	struct ifnet *ifp = ifa->ifa_ifp;
742 
743 	if (route_cb.any_count == 0)
744 		return;
745 	for (pass = 1; pass < 3; pass++) {
746 		bzero(&info, sizeof(info));
747 		if ((cmd == RTM_ADD && pass == 1) ||
748 		    (cmd == RTM_DELETE && pass == 2)) {
749 			struct ifa_msghdr ifam;
750 			int ncmd = cmd == RTM_ADD ? RTM_NEWADDR : RTM_DELADDR;
751 
752 			ifaaddr = sa = ifa->ifa_addr;
753 			ifpaddr = ifp->if_addrlist.tqh_first->ifa_addr;
754 			netmask = ifa->ifa_netmask;
755 			brdaddr = ifa->ifa_dstaddr;
756 			bzero(&ifam, sizeof(ifam));
757 			ifam.ifam_index = ifp->if_index;
758 			ifam.ifam_metric = ifa->ifa_metric;
759 			ifam.ifam_flags = ifa->ifa_flags;
760 			m = rt_msg1(ncmd, &info, (caddr_t)&ifam, sizeof(ifam));
761 			if (m == NULL)
762 				continue;
763 			mtod(m, struct ifa_msghdr *)->ifam_addrs =
764 			    info.rti_addrs;
765 		}
766 		if ((cmd == RTM_ADD && pass == 2) ||
767 		    (cmd == RTM_DELETE && pass == 1)) {
768 			struct rt_msghdr rtm;
769 
770 			if (rt == 0)
771 				continue;
772 			netmask = rt_mask(rt);
773 			dst = sa = rt_key(rt);
774 			gate = rt->rt_gateway;
775 			bzero(&rtm, sizeof(rtm));
776 			rtm.rtm_index = ifp->if_index;
777 			rtm.rtm_flags |= rt->rt_flags;
778 			rtm.rtm_errno = error;
779 			m = rt_msg1(cmd, &info, (caddr_t)&rtm, sizeof(rtm));
780 			if (m == NULL)
781 				continue;
782 			mtod(m, struct rt_msghdr *)->rtm_addrs = info.rti_addrs;
783 		}
784 		route_proto.sp_protocol = sa ? sa->sa_family : 0;
785 		raw_input(m, &route_proto, &route_src, &route_dst);
786 	}
787 }
788 
789 /*
790  * This is called to generate routing socket messages indicating
791  * network interface arrival and departure.
792  */
793 void
794 rt_ifannouncemsg(ifp, what)
795 	struct ifnet *ifp;
796 	int what;
797 {
798 	struct if_announcemsghdr ifan;
799 	struct mbuf *m;
800 	struct rt_addrinfo info;
801 
802 	if (route_cb.any_count == 0)
803 		return;
804 	bzero(&info, sizeof(info));
805 	bzero(&ifan, sizeof(ifan));
806 	ifan.ifan_index = ifp->if_index;
807 	strcpy(ifan.ifan_name, ifp->if_xname);
808 	ifan.ifan_what = what;
809 	m = rt_msg1(RTM_IFANNOUNCE, &info, (caddr_t)&ifan, sizeof(ifan));
810 	if (m == 0)
811 		return;
812 	route_proto.sp_protocol = 0;
813 	raw_input(m, &route_proto, &route_src, &route_dst);
814 }
815 
816 /*
817  * This is used in dumping the kernel table via sysctl().
818  */
819 int
820 sysctl_dumpentry(rn, v)
821 	struct radix_node *rn;
822 	register void *v;
823 {
824 	register struct walkarg *w = v;
825 	register struct rtentry *rt = (struct rtentry *)rn;
826 	int error = 0, size;
827 	struct rt_addrinfo info;
828 
829 	if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg))
830 		return 0;
831 	bzero(&info, sizeof(info));
832 	dst = rt_key(rt);
833 	gate = rt->rt_gateway;
834 	netmask = rt_mask(rt);
835 	genmask = rt->rt_genmask;
836 	if (rt->rt_ifp) {
837 		ifpaddr = rt->rt_ifp->if_addrlist.tqh_first->ifa_addr;
838 		ifaaddr = rt->rt_ifa->ifa_addr;
839 		if (rt->rt_ifp->if_flags & IFF_POINTOPOINT)
840 			brdaddr = rt->rt_ifa->ifa_dstaddr;
841 	}
842 	if ((error = rt_msg2(RTM_GET, &info, 0, w, &size)))
843 		return (error);
844 	if (w->w_where && w->w_tmem && w->w_needed <= 0) {
845 		register struct rt_msghdr *rtm = (struct rt_msghdr *)w->w_tmem;
846 
847 		rtm->rtm_flags = rt->rt_flags;
848 		rtm->rtm_use = rt->rt_use;
849 		rtm->rtm_rmx = rt->rt_rmx;
850 		rtm->rtm_index = rt->rt_ifp->if_index;
851 		rtm->rtm_errno = rtm->rtm_pid = rtm->rtm_seq = 0;
852 		rtm->rtm_addrs = info.rti_addrs;
853 		if ((error = copyout(rtm, w->w_where, size)) != 0)
854 			w->w_where = NULL;
855 		else
856 			w->w_where += size;
857 	}
858 	return (error);
859 }
860 
861 int
862 sysctl_iflist(af, w, type)
863 	int	af;
864 	register struct	walkarg *w;
865 	int type;
866 {
867 	register struct ifnet *ifp;
868 	register struct ifaddr *ifa;
869 	struct	rt_addrinfo info;
870 	int	len, error = 0;
871 
872 	bzero(&info, sizeof(info));
873 	for (ifp = ifnet.tqh_first; ifp != 0; ifp = ifp->if_list.tqe_next) {
874 		if (w->w_arg && w->w_arg != ifp->if_index)
875 			continue;
876 		ifa = ifp->if_addrlist.tqh_first;
877 		ifpaddr = ifa->ifa_addr;
878 		switch(type) {
879 		case NET_RT_IFLIST:
880 			error =
881 			    rt_msg2(RTM_IFINFO, &info, (caddr_t)0, w, &len);
882 			break;
883 #ifdef COMPAT_14
884 		case NET_RT_OIFLIST:
885 			error =
886 			    rt_msg2(RTM_OIFINFO, &info, (caddr_t)0, w, &len);
887 			break;
888 #endif
889 		default:
890 			panic("sysctl_iflist(1)");
891 		}
892 		if (error)
893 			return (error);
894 		ifpaddr = 0;
895 		if (w->w_where && w->w_tmem && w->w_needed <= 0) {
896 			switch(type) {
897 			case NET_RT_IFLIST: {
898 				register struct if_msghdr *ifm;
899 
900 				ifm = (struct if_msghdr *)w->w_tmem;
901 				ifm->ifm_index = ifp->if_index;
902 				ifm->ifm_flags = ifp->if_flags;
903 				ifm->ifm_data = ifp->if_data;
904 				ifm->ifm_addrs = info.rti_addrs;
905 				error = copyout(ifm, w->w_where, len);
906 				if (error)
907 					return (error);
908 				w->w_where += len;
909 				break;
910 			}
911 
912 #ifdef COMPAT_14
913 			case NET_RT_OIFLIST: {
914 				register struct if_msghdr14 *ifm;
915 
916 				ifm = (struct if_msghdr14 *)w->w_tmem;
917 				ifm->ifm_index = ifp->if_index;
918 				ifm->ifm_flags = ifp->if_flags;
919 				ifm->ifm_data.ifi_type = ifp->if_data.ifi_type;
920 				ifm->ifm_data.ifi_addrlen =
921 				    ifp->if_data.ifi_addrlen;
922 				ifm->ifm_data.ifi_hdrlen =
923 				    ifp->if_data.ifi_hdrlen;
924 				ifm->ifm_data.ifi_mtu = ifp->if_data.ifi_mtu;
925 				ifm->ifm_data.ifi_metric =
926 				    ifp->if_data.ifi_metric;
927 				ifm->ifm_data.ifi_baudrate =
928 				    ifp->if_data.ifi_baudrate;
929 				ifm->ifm_data.ifi_ipackets =
930 				    ifp->if_data.ifi_ipackets;
931 				ifm->ifm_data.ifi_ierrors =
932 				    ifp->if_data.ifi_ierrors;
933 				ifm->ifm_data.ifi_opackets =
934 				    ifp->if_data.ifi_opackets;
935 				ifm->ifm_data.ifi_oerrors =
936 				    ifp->if_data.ifi_oerrors;
937 				ifm->ifm_data.ifi_collisions =
938 				    ifp->if_data.ifi_collisions;
939 				ifm->ifm_data.ifi_ibytes =
940 				    ifp->if_data.ifi_ibytes;
941 				ifm->ifm_data.ifi_obytes =
942 				    ifp->if_data.ifi_obytes;
943 				ifm->ifm_data.ifi_imcasts =
944 				    ifp->if_data.ifi_imcasts;
945 				ifm->ifm_data.ifi_omcasts =
946 				    ifp->if_data.ifi_omcasts;
947 				ifm->ifm_data.ifi_iqdrops =
948 				    ifp->if_data.ifi_iqdrops;
949 				ifm->ifm_data.ifi_noproto =
950 				    ifp->if_data.ifi_noproto;
951 				ifm->ifm_data.ifi_lastchange =
952 				    ifp->if_data.ifi_lastchange;
953 				ifm->ifm_addrs = info.rti_addrs;
954 				error = copyout(ifm, w->w_where, len);
955 				if (error)
956 					return (error);
957 				w->w_where += len;
958 				break;
959 			}
960 #endif
961 			default:
962 				panic("sysctl_iflist(2)");
963 			}
964 		}
965 		while ((ifa = ifa->ifa_list.tqe_next) != NULL) {
966 			if (af && af != ifa->ifa_addr->sa_family)
967 				continue;
968 			ifaaddr = ifa->ifa_addr;
969 			netmask = ifa->ifa_netmask;
970 			brdaddr = ifa->ifa_dstaddr;
971 			if ((error = rt_msg2(RTM_NEWADDR, &info, 0, w, &len)))
972 				return (error);
973 			if (w->w_where && w->w_tmem && w->w_needed <= 0) {
974 				register struct ifa_msghdr *ifam;
975 
976 				ifam = (struct ifa_msghdr *)w->w_tmem;
977 				ifam->ifam_index = ifa->ifa_ifp->if_index;
978 				ifam->ifam_flags = ifa->ifa_flags;
979 				ifam->ifam_metric = ifa->ifa_metric;
980 				ifam->ifam_addrs = info.rti_addrs;
981 				error = copyout(w->w_tmem, w->w_where, len);
982 				if (error)
983 					return (error);
984 				w->w_where += len;
985 			}
986 		}
987 		ifaaddr = netmask = brdaddr = 0;
988 	}
989 	return (0);
990 }
991 
992 int
993 sysctl_rtable(name, namelen, where, given, new, newlen)
994 	int	*name;
995 	u_int	namelen;
996 	void 	*where;
997 	size_t	*given;
998 	void	*new;
999 	size_t	newlen;
1000 {
1001 	register struct radix_node_head *rnh;
1002 	int	i, s, error = EINVAL;
1003 	u_char  af;
1004 	struct	walkarg w;
1005 
1006 	if (new)
1007 		return (EPERM);
1008 	if (namelen != 3)
1009 		return (EINVAL);
1010 	af = name[0];
1011 	w.w_tmemneeded = 0;
1012 	w.w_tmemsize = 0;
1013 	w.w_tmem = NULL;
1014 again:
1015 	/* we may return here if a later [re]alloc of the t_mem buffer fails */
1016 	if (w.w_tmemneeded) {
1017 		w.w_tmem = (caddr_t) malloc(w.w_tmemneeded, M_RTABLE, M_WAITOK);
1018 		w.w_tmemsize = w.w_tmemneeded;
1019 		w.w_tmemneeded = 0;
1020 	}
1021 	w.w_op = name[1];
1022 	w.w_arg = name[2];
1023 	w.w_given = *given;
1024 	w.w_needed = 0 - w.w_given;
1025 	w.w_where = where;
1026 
1027 	s = splsoftnet();
1028 	switch (w.w_op) {
1029 
1030 	case NET_RT_DUMP:
1031 	case NET_RT_FLAGS:
1032 		for (i = 1; i <= AF_MAX; i++)
1033 			if ((rnh = rt_tables[i]) && (af == 0 || af == i) &&
1034 			    (error = (*rnh->rnh_walktree)(rnh,
1035 			    sysctl_dumpentry, &w)))
1036 				break;
1037 		break;
1038 
1039 #ifdef COMPAT_14
1040 	case NET_RT_OIFLIST:
1041 		error = sysctl_iflist(af, &w, w.w_op);
1042 		break;
1043 #endif
1044 
1045 	case NET_RT_IFLIST:
1046 		error = sysctl_iflist(af, &w, w.w_op);
1047 	}
1048 	splx(s);
1049 
1050 	/* check to see if we couldn't allocate memory with NOWAIT */
1051 	if (error == ENOBUFS && w.w_tmem == 0 && w.w_tmemneeded)
1052 		goto again;
1053 
1054 	if (w.w_tmem)
1055 		free(w.w_tmem, M_RTABLE);
1056 	w.w_needed += w.w_given;
1057 	if (where) {
1058 		*given = w.w_where - (caddr_t) where;
1059 		if (*given < w.w_needed)
1060 			return (ENOMEM);
1061 	} else {
1062 		*given = (11 * w.w_needed) / 10;
1063 	}
1064 	return (error);
1065 }
1066 
1067 /*
1068  * Definitions of protocols supported in the ROUTE domain.
1069  */
1070 
1071 extern	struct domain routedomain;		/* or at least forward */
1072 
1073 struct protosw routesw[] = {
1074 { SOCK_RAW,	&routedomain,	0,		PR_ATOMIC|PR_ADDR,
1075   raw_input,	route_output,	raw_ctlinput,	0,
1076   route_usrreq,
1077   raw_init,	0,		0,		0,
1078   sysctl_rtable,
1079 }
1080 };
1081 
1082 struct domain routedomain =
1083     { PF_ROUTE, "route", route_init, 0, 0,
1084       routesw, &routesw[sizeof(routesw)/sizeof(routesw[0])] };
1085