xref: /netbsd-src/sys/net/route.c (revision 2a399c6883d870daece976daec6ffa7bb7f934ce)
1 /*	$NetBSD: route.c,v 1.17 1997/04/02 21:17:28 christos Exp $	*/
2 
3 /*
4  * Copyright (c) 1980, 1986, 1991, 1993
5  *	The Regents of the University of California.  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. All advertising materials mentioning features or use of this software
16  *    must display the following acknowledgement:
17  *	This product includes software developed by the University of
18  *	California, Berkeley and its contributors.
19  * 4. Neither the name of the University nor the names of its contributors
20  *    may be used to endorse or promote products derived from this software
21  *    without specific prior written permission.
22  *
23  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
24  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
25  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
26  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
27  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
28  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
29  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
30  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
31  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
32  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
33  * SUCH DAMAGE.
34  *
35  *	@(#)route.c	8.3 (Berkeley) 1/9/95
36  */
37 
38 #include <sys/param.h>
39 #include <sys/systm.h>
40 #include <sys/proc.h>
41 #include <sys/mbuf.h>
42 #include <sys/socket.h>
43 #include <sys/socketvar.h>
44 #include <sys/domain.h>
45 #include <sys/protosw.h>
46 #include <sys/ioctl.h>
47 
48 #include <net/if.h>
49 #include <net/route.h>
50 #include <net/raw_cb.h>
51 
52 #include <netinet/in.h>
53 #include <netinet/in_var.h>
54 
55 #ifdef NS
56 #include <netns/ns.h>
57 #endif
58 
59 #define	SA(p) ((struct sockaddr *)(p))
60 
61 int	rttrash;		/* routes not in table but not freed */
62 struct	sockaddr wildcard;	/* zero valued cookie for wildcard searches */
63 
64 void
65 rtable_init(table)
66 	void **table;
67 {
68 	struct domain *dom;
69 	for (dom = domains; dom; dom = dom->dom_next)
70 		if (dom->dom_rtattach)
71 			dom->dom_rtattach(&table[dom->dom_family],
72 			    dom->dom_rtoffset);
73 }
74 
75 void
76 route_init()
77 {
78 	rn_init();	/* initialize all zeroes, all ones, mask table */
79 	rtable_init((void **)rt_tables);
80 }
81 
82 /*
83  * Packet routing routines.
84  */
85 void
86 rtalloc(ro)
87 	register struct route *ro;
88 {
89 	if (ro->ro_rt && ro->ro_rt->rt_ifp && (ro->ro_rt->rt_flags & RTF_UP))
90 		return;				 /* XXX */
91 	ro->ro_rt = rtalloc1(&ro->ro_dst, 1);
92 }
93 
94 struct rtentry *
95 rtalloc1(dst, report)
96 	register struct sockaddr *dst;
97 	int report;
98 {
99 	register struct radix_node_head *rnh = rt_tables[dst->sa_family];
100 	register struct rtentry *rt;
101 	register struct radix_node *rn;
102 	struct rtentry *newrt = 0;
103 	struct rt_addrinfo info;
104 	int  s = splsoftnet(), err = 0, msgtype = RTM_MISS;
105 
106 	if (rnh && (rn = rnh->rnh_matchaddr((caddr_t)dst, rnh)) &&
107 	    ((rn->rn_flags & RNF_ROOT) == 0)) {
108 		newrt = rt = (struct rtentry *)rn;
109 		if (report && (rt->rt_flags & RTF_CLONING)) {
110 			err = rtrequest(RTM_RESOLVE, dst, SA(0),
111 					      SA(0), 0, &newrt);
112 			if (err) {
113 				newrt = rt;
114 				rt->rt_refcnt++;
115 				goto miss;
116 			}
117 			if ((rt = newrt) && (rt->rt_flags & RTF_XRESOLVE)) {
118 				msgtype = RTM_RESOLVE;
119 				goto miss;
120 			}
121 		} else
122 			rt->rt_refcnt++;
123 	} else {
124 		rtstat.rts_unreach++;
125 	miss:	if (report) {
126 			bzero((caddr_t)&info, sizeof(info));
127 			info.rti_info[RTAX_DST] = dst;
128 			rt_missmsg(msgtype, &info, 0, err);
129 		}
130 	}
131 	splx(s);
132 	return (newrt);
133 }
134 
135 void
136 rtfree(rt)
137 	register struct rtentry *rt;
138 {
139 	register struct ifaddr *ifa;
140 
141 	if (rt == 0)
142 		panic("rtfree");
143 	rt->rt_refcnt--;
144 	if (rt->rt_refcnt <= 0 && (rt->rt_flags & RTF_UP) == 0) {
145 		if (rt->rt_nodes->rn_flags & (RNF_ACTIVE | RNF_ROOT))
146 			panic ("rtfree 2");
147 		rttrash--;
148 		if (rt->rt_refcnt < 0) {
149 			printf("rtfree: %p not freed (neg refs)\n", rt);
150 			return;
151 		}
152 		ifa = rt->rt_ifa;
153 		IFAFREE(ifa);
154 		Free(rt_key(rt));
155 		Free(rt);
156 	}
157 }
158 
159 void
160 ifafree(ifa)
161 	register struct ifaddr *ifa;
162 {
163 	if (ifa == NULL)
164 		panic("ifafree");
165 	if (ifa->ifa_refcnt == 0)
166 		free(ifa, M_IFADDR);
167 	else
168 		ifa->ifa_refcnt--;
169 }
170 
171 /*
172  * Force a routing table entry to the specified
173  * destination to go through the given gateway.
174  * Normally called as a result of a routing redirect
175  * message from the network layer.
176  *
177  * N.B.: must be called at splsoftnet
178  *
179  */
180 void
181 rtredirect(dst, gateway, netmask, flags, src, rtp)
182 	struct sockaddr *dst, *gateway, *netmask, *src;
183 	int flags;
184 	struct rtentry **rtp;
185 {
186 	register struct rtentry *rt;
187 	int error = 0;
188 	short *stat = 0;
189 	struct rt_addrinfo info;
190 	struct ifaddr *ifa;
191 
192 	/* verify the gateway is directly reachable */
193 	if ((ifa = ifa_ifwithnet(gateway)) == 0) {
194 		error = ENETUNREACH;
195 		goto out;
196 	}
197 	rt = rtalloc1(dst, 0);
198 	/*
199 	 * If the redirect isn't from our current router for this dst,
200 	 * it's either old or wrong.  If it redirects us to ourselves,
201 	 * we have a routing loop, perhaps as a result of an interface
202 	 * going down recently.
203 	 */
204 #define	equal(a1, a2) (bcmp((caddr_t)(a1), (caddr_t)(a2), (a1)->sa_len) == 0)
205 	if (!(flags & RTF_DONE) && rt &&
206 	     (!equal(src, rt->rt_gateway) || rt->rt_ifa != ifa))
207 		error = EINVAL;
208 	else if (ifa_ifwithaddr(gateway))
209 		error = EHOSTUNREACH;
210 	if (error)
211 		goto done;
212 	/*
213 	 * Create a new entry if we just got back a wildcard entry
214 	 * or the the lookup failed.  This is necessary for hosts
215 	 * which use routing redirects generated by smart gateways
216 	 * to dynamically build the routing tables.
217 	 */
218 	if ((rt == 0) || (rt_mask(rt) && rt_mask(rt)->sa_len < 2))
219 		goto create;
220 	/*
221 	 * Don't listen to the redirect if it's
222 	 * for a route to an interface.
223 	 */
224 	if (rt->rt_flags & RTF_GATEWAY) {
225 		if (((rt->rt_flags & RTF_HOST) == 0) && (flags & RTF_HOST)) {
226 			/*
227 			 * Changing from route to net => route to host.
228 			 * Create new route, rather than smashing route to net.
229 			 */
230 		create:
231 			flags |=  RTF_GATEWAY | RTF_DYNAMIC;
232 			error = rtrequest((int)RTM_ADD, dst, gateway,
233 				    netmask, flags,
234 				    (struct rtentry **)0);
235 			stat = &rtstat.rts_dynamic;
236 		} else {
237 			/*
238 			 * Smash the current notion of the gateway to
239 			 * this destination.  Should check about netmask!!!
240 			 */
241 			rt->rt_flags |= RTF_MODIFIED;
242 			flags |= RTF_MODIFIED;
243 			stat = &rtstat.rts_newgateway;
244 			rt_setgate(rt, rt_key(rt), gateway);
245 		}
246 	} else
247 		error = EHOSTUNREACH;
248 done:
249 	if (rt) {
250 		if (rtp && !error)
251 			*rtp = rt;
252 		else
253 			rtfree(rt);
254 	}
255 out:
256 	if (error)
257 		rtstat.rts_badredirect++;
258 	else if (stat != NULL)
259 		(*stat)++;
260 	bzero((caddr_t)&info, sizeof(info));
261 	info.rti_info[RTAX_DST] = dst;
262 	info.rti_info[RTAX_GATEWAY] = gateway;
263 	info.rti_info[RTAX_NETMASK] = netmask;
264 	info.rti_info[RTAX_AUTHOR] = src;
265 	rt_missmsg(RTM_REDIRECT, &info, flags, error);
266 }
267 
268 /*
269 * Routing table ioctl interface.
270 */
271 int
272 rtioctl(req, data, p)
273 	u_long req;
274 	caddr_t data;
275 	struct proc *p;
276 {
277 	return (EOPNOTSUPP);
278 }
279 
280 struct ifaddr *
281 ifa_ifwithroute(flags, dst, gateway)
282 	int flags;
283 	struct sockaddr	*dst, *gateway;
284 {
285 	register struct ifaddr *ifa;
286 	if ((flags & RTF_GATEWAY) == 0) {
287 		/*
288 		 * If we are adding a route to an interface,
289 		 * and the interface is a pt to pt link
290 		 * we should search for the destination
291 		 * as our clue to the interface.  Otherwise
292 		 * we can use the local address.
293 		 */
294 		ifa = 0;
295 		if (flags & RTF_HOST)
296 			ifa = ifa_ifwithdstaddr(dst);
297 		if (ifa == 0)
298 			ifa = ifa_ifwithaddr(gateway);
299 	} else {
300 		/*
301 		 * If we are adding a route to a remote net
302 		 * or host, the gateway may still be on the
303 		 * other end of a pt to pt link.
304 		 */
305 		ifa = ifa_ifwithdstaddr(gateway);
306 	}
307 	if (ifa == 0)
308 		ifa = ifa_ifwithnet(gateway);
309 	if (ifa == 0) {
310 		struct rtentry *rt = rtalloc1(dst, 0);
311 		if (rt == 0)
312 			return (0);
313 		rt->rt_refcnt--;
314 		if ((ifa = rt->rt_ifa) == 0)
315 			return (0);
316 	}
317 	if (ifa->ifa_addr->sa_family != dst->sa_family) {
318 		struct ifaddr *oifa = ifa;
319 		ifa = ifaof_ifpforaddr(dst, ifa->ifa_ifp);
320 		if (ifa == 0)
321 			ifa = oifa;
322 	}
323 	return (ifa);
324 }
325 
326 #define ROUNDUP(a) (a>0 ? (1 + (((a) - 1) | (sizeof(long) - 1))) : sizeof(long))
327 
328 int
329 rtrequest(req, dst, gateway, netmask, flags, ret_nrt)
330 	int req, flags;
331 	struct sockaddr *dst, *gateway, *netmask;
332 	struct rtentry **ret_nrt;
333 {
334 	int s = splsoftnet(); int error = 0;
335 	register struct rtentry *rt;
336 	register struct radix_node *rn;
337 	register struct radix_node_head *rnh;
338 	struct ifaddr *ifa;
339 	struct sockaddr *ndst;
340 #define senderr(x) { error = x ; goto bad; }
341 
342 	if ((rnh = rt_tables[dst->sa_family]) == 0)
343 		senderr(ESRCH);
344 	if (flags & RTF_HOST)
345 		netmask = 0;
346 	switch (req) {
347 	case RTM_DELETE:
348 		if ((rn = rnh->rnh_deladdr(dst, netmask, rnh)) == 0)
349 			senderr(ESRCH);
350 		if (rn->rn_flags & (RNF_ACTIVE | RNF_ROOT))
351 			panic ("rtrequest delete");
352 		rt = (struct rtentry *)rn;
353 		rt->rt_flags &= ~RTF_UP;
354 		if (rt->rt_gwroute) {
355 			rt = rt->rt_gwroute; RTFREE(rt);
356 			(rt = (struct rtentry *)rn)->rt_gwroute = 0;
357 		}
358 		if ((ifa = rt->rt_ifa) && ifa->ifa_rtrequest)
359 			ifa->ifa_rtrequest(RTM_DELETE, rt, SA(0));
360 		rttrash++;
361 		if (ret_nrt)
362 			*ret_nrt = rt;
363 		else if (rt->rt_refcnt <= 0) {
364 			rt->rt_refcnt++;
365 			rtfree(rt);
366 		}
367 		break;
368 
369 	case RTM_RESOLVE:
370 		if (ret_nrt == 0 || (rt = *ret_nrt) == 0)
371 			senderr(EINVAL);
372 		ifa = rt->rt_ifa;
373 		flags = rt->rt_flags & ~RTF_CLONING;
374 		gateway = rt->rt_gateway;
375 		if ((netmask = rt->rt_genmask) == 0)
376 			flags |= RTF_HOST;
377 		goto makeroute;
378 
379 	case RTM_ADD:
380 		if ((ifa = ifa_ifwithroute(flags, dst, gateway)) == 0)
381 			senderr(ENETUNREACH);
382 	makeroute:
383 		R_Malloc(rt, struct rtentry *, sizeof(*rt));
384 		if (rt == 0)
385 			senderr(ENOBUFS);
386 		Bzero(rt, sizeof(*rt));
387 		rt->rt_flags = RTF_UP | flags;
388 		if (rt_setgate(rt, dst, gateway)) {
389 			Free(rt);
390 			senderr(ENOBUFS);
391 		}
392 		ndst = rt_key(rt);
393 		if (netmask) {
394 			rt_maskedcopy(dst, ndst, netmask);
395 		} else
396 			Bcopy(dst, ndst, dst->sa_len);
397 		rn = rnh->rnh_addaddr((caddr_t)ndst, (caddr_t)netmask,
398 					rnh, rt->rt_nodes);
399 		if (rn == 0) {
400 			if (rt->rt_gwroute)
401 				rtfree(rt->rt_gwroute);
402 			Free(rt_key(rt));
403 			Free(rt);
404 			senderr(EEXIST);
405 		}
406 		ifa->ifa_refcnt++;
407 		rt->rt_ifa = ifa;
408 		rt->rt_ifp = ifa->ifa_ifp;
409 		if (req == RTM_RESOLVE)
410 			rt->rt_rmx = (*ret_nrt)->rt_rmx; /* copy metrics */
411 		if (ifa->ifa_rtrequest)
412 			ifa->ifa_rtrequest(req, rt, SA(ret_nrt ? *ret_nrt : 0));
413 		if (ret_nrt) {
414 			*ret_nrt = rt;
415 			rt->rt_refcnt++;
416 		}
417 		break;
418 	}
419 bad:
420 	splx(s);
421 	return (error);
422 }
423 
424 int
425 rt_setgate(rt0, dst, gate)
426 	struct rtentry *rt0;
427 	struct sockaddr *dst, *gate;
428 {
429 	caddr_t new, old;
430 	int dlen = ROUNDUP(dst->sa_len), glen = ROUNDUP(gate->sa_len);
431 	register struct rtentry *rt = rt0;
432 
433 	if (rt->rt_gateway == 0 || glen > ROUNDUP(rt->rt_gateway->sa_len)) {
434 		old = (caddr_t)rt_key(rt);
435 		R_Malloc(new, caddr_t, dlen + glen);
436 		if (new == 0)
437 			return 1;
438 		rt->rt_nodes->rn_key = new;
439 	} else {
440 		new = rt->rt_nodes->rn_key;
441 		old = 0;
442 	}
443 	Bcopy(gate, (rt->rt_gateway = (struct sockaddr *)(new + dlen)), glen);
444 	if (old) {
445 		Bcopy(dst, new, dlen);
446 		Free(old);
447 	}
448 	if (rt->rt_gwroute) {
449 		rt = rt->rt_gwroute; RTFREE(rt);
450 		rt = rt0; rt->rt_gwroute = 0;
451 	}
452 	if (rt->rt_flags & RTF_GATEWAY) {
453 		rt->rt_gwroute = rtalloc1(gate, 1);
454 	}
455 	return 0;
456 }
457 
458 void
459 rt_maskedcopy(src, dst, netmask)
460 	struct sockaddr *src, *dst, *netmask;
461 {
462 	register u_char *cp1 = (u_char *)src;
463 	register u_char *cp2 = (u_char *)dst;
464 	register u_char *cp3 = (u_char *)netmask;
465 	u_char *cplim = cp2 + *cp3;
466 	u_char *cplim2 = cp2 + *cp1;
467 
468 	*cp2++ = *cp1++; *cp2++ = *cp1++; /* copies sa_len & sa_family */
469 	cp3 += 2;
470 	if (cplim > cplim2)
471 		cplim = cplim2;
472 	while (cp2 < cplim)
473 		*cp2++ = *cp1++ & *cp3++;
474 	if (cp2 < cplim2)
475 		bzero((caddr_t)cp2, (unsigned)(cplim2 - cp2));
476 }
477 
478 /*
479  * Set up a routing table entry, normally
480  * for an interface.
481  */
482 int
483 rtinit(ifa, cmd, flags)
484 	register struct ifaddr *ifa;
485 	int cmd, flags;
486 {
487 	register struct rtentry *rt;
488 	register struct sockaddr *dst;
489 	register struct sockaddr *deldst;
490 	struct mbuf *m = 0;
491 	struct rtentry *nrt = 0;
492 	int error;
493 
494 	dst = flags & RTF_HOST ? ifa->ifa_dstaddr : ifa->ifa_addr;
495 	if (cmd == RTM_DELETE) {
496 		if ((flags & RTF_HOST) == 0 && ifa->ifa_netmask) {
497 			m = m_get(M_WAIT, MT_SONAME);
498 			deldst = mtod(m, struct sockaddr *);
499 			rt_maskedcopy(dst, deldst, ifa->ifa_netmask);
500 			dst = deldst;
501 		}
502 		if ((rt = rtalloc1(dst, 0)) != NULL) {
503 			rt->rt_refcnt--;
504 			if (rt->rt_ifa != ifa) {
505 				if (m)
506 					(void) m_free(m);
507 				return (flags & RTF_HOST ? EHOSTUNREACH
508 							: ENETUNREACH);
509 			}
510 		}
511 	}
512 	error = rtrequest(cmd, dst, ifa->ifa_addr, ifa->ifa_netmask,
513 			flags | ifa->ifa_flags, &nrt);
514 	if (m)
515 		(void) m_free(m);
516 	if (cmd == RTM_DELETE && error == 0 && (rt = nrt)) {
517 		rt_newaddrmsg(cmd, ifa, error, nrt);
518 		if (rt->rt_refcnt <= 0) {
519 			rt->rt_refcnt++;
520 			rtfree(rt);
521 		}
522 	}
523 	if (cmd == RTM_ADD && error == 0 && (rt = nrt)) {
524 		rt->rt_refcnt--;
525 		if (rt->rt_ifa != ifa) {
526 			printf("rtinit: wrong ifa (%p) was (%p)\n", ifa,
527 				rt->rt_ifa);
528 			if (rt->rt_ifa->ifa_rtrequest)
529 			    rt->rt_ifa->ifa_rtrequest(RTM_DELETE, rt, SA(0));
530 			IFAFREE(rt->rt_ifa);
531 			rt->rt_ifa = ifa;
532 			rt->rt_ifp = ifa->ifa_ifp;
533 			ifa->ifa_refcnt++;
534 			if (ifa->ifa_rtrequest)
535 			    ifa->ifa_rtrequest(RTM_ADD, rt, SA(0));
536 		}
537 		rt_newaddrmsg(cmd, ifa, error, nrt);
538 	}
539 	return (error);
540 }
541