xref: /netbsd-src/sys/net/route.c (revision 89c5a767f8fc7a4633b2d409966e2becbb98ff92)
1 /*	$NetBSD: route.c,v 1.32 2000/03/10 14:47:12 itojun Exp $	*/
2 
3 /*-
4  * Copyright (c) 1998 The NetBSD Foundation, Inc.
5  * All rights reserved.
6  *
7  * This code is derived from software contributed to The NetBSD Foundation
8  * by Kevin M. Lahey of the Numerical Aerospace Simulation Facility,
9  * NASA Ames Research Center.
10  *
11  * Redistribution and use in source and binary forms, with or without
12  * modification, are permitted provided that the following conditions
13  * are met:
14  * 1. Redistributions of source code must retain the above copyright
15  *    notice, this list of conditions and the following disclaimer.
16  * 2. Redistributions in binary form must reproduce the above copyright
17  *    notice, this list of conditions and the following disclaimer in the
18  *    documentation and/or other materials provided with the distribution.
19  * 3. All advertising materials mentioning features or use of this software
20  *    must display the following acknowledgement:
21  *	This product includes software developed by the NetBSD
22  *	Foundation, Inc. and its contributors.
23  * 4. Neither the name of The NetBSD Foundation nor the names of its
24  *    contributors may be used to endorse or promote products derived
25  *    from this software without specific prior written permission.
26  *
27  * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
28  * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
29  * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
30  * PURPOSE ARE DISCLAIMED.  IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
31  * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
32  * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
33  * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
34  * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
35  * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
36  * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
37  * POSSIBILITY OF SUCH DAMAGE.
38  */
39 
40 /*
41  * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
42  * All rights reserved.
43  *
44  * Redistribution and use in source and binary forms, with or without
45  * modification, are permitted provided that the following conditions
46  * are met:
47  * 1. Redistributions of source code must retain the above copyright
48  *    notice, this list of conditions and the following disclaimer.
49  * 2. Redistributions in binary form must reproduce the above copyright
50  *    notice, this list of conditions and the following disclaimer in the
51  *    documentation and/or other materials provided with the distribution.
52  * 3. Neither the name of the project nor the names of its contributors
53  *    may be used to endorse or promote products derived from this software
54  *    without specific prior written permission.
55  *
56  * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
57  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
58  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
59  * ARE DISCLAIMED.  IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
60  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
61  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
62  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
63  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
64  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
65  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
66  * SUCH DAMAGE.
67  */
68 
69 /*
70  * Copyright (c) 1980, 1986, 1991, 1993
71  *	The Regents of the University of California.  All rights reserved.
72  *
73  * Redistribution and use in source and binary forms, with or without
74  * modification, are permitted provided that the following conditions
75  * are met:
76  * 1. Redistributions of source code must retain the above copyright
77  *    notice, this list of conditions and the following disclaimer.
78  * 2. Redistributions in binary form must reproduce the above copyright
79  *    notice, this list of conditions and the following disclaimer in the
80  *    documentation and/or other materials provided with the distribution.
81  * 3. All advertising materials mentioning features or use of this software
82  *    must display the following acknowledgement:
83  *	This product includes software developed by the University of
84  *	California, Berkeley and its contributors.
85  * 4. Neither the name of the University nor the names of its contributors
86  *    may be used to endorse or promote products derived from this software
87  *    without specific prior written permission.
88  *
89  * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
90  * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
91  * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
92  * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
93  * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
94  * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
95  * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
96  * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
97  * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
98  * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
99  * SUCH DAMAGE.
100  *
101  *	@(#)route.c	8.3 (Berkeley) 1/9/95
102  */
103 
104 #include "opt_ns.h"
105 
106 #include <sys/param.h>
107 #include <sys/systm.h>
108 #include <sys/proc.h>
109 #include <sys/mbuf.h>
110 #include <sys/socket.h>
111 #include <sys/socketvar.h>
112 #include <sys/domain.h>
113 #include <sys/protosw.h>
114 #include <sys/kernel.h>
115 #include <sys/ioctl.h>
116 #include <sys/pool.h>
117 
118 #include <net/if.h>
119 #include <net/route.h>
120 #include <net/raw_cb.h>
121 
122 #include <netinet/in.h>
123 #include <netinet/in_var.h>
124 
125 #ifdef NS
126 #include <netns/ns.h>
127 #endif
128 
129 #define	SA(p) ((struct sockaddr *)(p))
130 
131 int	rttrash;		/* routes not in table but not freed */
132 struct	sockaddr wildcard;	/* zero valued cookie for wildcard searches */
133 
134 struct pool rtentry_pool;	/* pool for rtentry structures */
135 struct pool rttimer_pool;	/* pool for rttimer structures */
136 
137 void
138 rtable_init(table)
139 	void **table;
140 {
141 	struct domain *dom;
142 	for (dom = domains; dom; dom = dom->dom_next)
143 		if (dom->dom_rtattach)
144 			dom->dom_rtattach(&table[dom->dom_family],
145 			    dom->dom_rtoffset);
146 }
147 
148 void
149 route_init()
150 {
151 
152 	pool_init(&rtentry_pool, sizeof(struct rtentry), 0, 0, 0, "rtentpl",
153 	    0, NULL, NULL, M_RTABLE);
154 
155 	rn_init();	/* initialize all zeroes, all ones, mask table */
156 	rtable_init((void **)rt_tables);
157 }
158 
159 /*
160  * Packet routing routines.
161  */
162 void
163 rtalloc(ro)
164 	register struct route *ro;
165 {
166 	if (ro->ro_rt && ro->ro_rt->rt_ifp && (ro->ro_rt->rt_flags & RTF_UP))
167 		return;				 /* XXX */
168 	ro->ro_rt = rtalloc1(&ro->ro_dst, 1);
169 }
170 
171 struct rtentry *
172 rtalloc1(dst, report)
173 	register struct sockaddr *dst;
174 	int report;
175 {
176 	register struct radix_node_head *rnh = rt_tables[dst->sa_family];
177 	register struct rtentry *rt;
178 	register struct radix_node *rn;
179 	struct rtentry *newrt = 0;
180 	struct rt_addrinfo info;
181 	int  s = splsoftnet(), err = 0, msgtype = RTM_MISS;
182 
183 	if (rnh && (rn = rnh->rnh_matchaddr((caddr_t)dst, rnh)) &&
184 	    ((rn->rn_flags & RNF_ROOT) == 0)) {
185 		newrt = rt = (struct rtentry *)rn;
186 		if (report && (rt->rt_flags & RTF_CLONING)) {
187 			err = rtrequest(RTM_RESOLVE, dst, SA(0),
188 					      SA(0), 0, &newrt);
189 			if (err) {
190 				newrt = rt;
191 				rt->rt_refcnt++;
192 				goto miss;
193 			}
194 			if ((rt = newrt) && (rt->rt_flags & RTF_XRESOLVE)) {
195 				msgtype = RTM_RESOLVE;
196 				goto miss;
197 			}
198 		} else
199 			rt->rt_refcnt++;
200 	} else {
201 		rtstat.rts_unreach++;
202 	miss:	if (report) {
203 			bzero((caddr_t)&info, sizeof(info));
204 			info.rti_info[RTAX_DST] = dst;
205 			rt_missmsg(msgtype, &info, 0, err);
206 		}
207 	}
208 	splx(s);
209 	return (newrt);
210 }
211 
212 void
213 rtfree(rt)
214 	register struct rtentry *rt;
215 {
216 	register struct ifaddr *ifa;
217 
218 	if (rt == 0)
219 		panic("rtfree");
220 	rt->rt_refcnt--;
221 	if (rt->rt_refcnt <= 0 && (rt->rt_flags & RTF_UP) == 0) {
222 		if (rt->rt_nodes->rn_flags & (RNF_ACTIVE | RNF_ROOT))
223 			panic ("rtfree 2");
224 		rttrash--;
225 		if (rt->rt_refcnt < 0) {
226 			printf("rtfree: %p not freed (neg refs)\n", rt);
227 			return;
228 		}
229 		rt_timer_remove_all(rt);
230 		ifa = rt->rt_ifa;
231 		IFAFREE(ifa);
232 		Free(rt_key(rt));
233 		pool_put(&rtentry_pool, rt);
234 	}
235 }
236 
237 void
238 ifafree(ifa)
239 	register struct ifaddr *ifa;
240 {
241 
242 #ifdef DIAGNOSTIC
243 	if (ifa == NULL)
244 		panic("ifafree: null ifa");
245 	if (ifa->ifa_refcnt != 0)
246 		panic("ifafree: ifa_refcnt != 0 (%d)", ifa->ifa_refcnt);
247 #endif
248 #ifdef IFAREF_DEBUG
249 	printf("ifafree: freeing ifaddr %p\n", ifa);
250 #endif
251 	free(ifa, M_IFADDR);
252 }
253 
254 /*
255  * Force a routing table entry to the specified
256  * destination to go through the given gateway.
257  * Normally called as a result of a routing redirect
258  * message from the network layer.
259  *
260  * N.B.: must be called at splsoftnet
261  */
262 void
263 rtredirect(dst, gateway, netmask, flags, src, rtp)
264 	struct sockaddr *dst, *gateway, *netmask, *src;
265 	int flags;
266 	struct rtentry **rtp;
267 {
268 	register struct rtentry *rt;
269 	int error = 0;
270 	short *stat = 0;
271 	struct rt_addrinfo info;
272 	struct ifaddr *ifa;
273 
274 	/* verify the gateway is directly reachable */
275 	if ((ifa = ifa_ifwithnet(gateway)) == 0) {
276 		error = ENETUNREACH;
277 		goto out;
278 	}
279 	rt = rtalloc1(dst, 0);
280 	/*
281 	 * If the redirect isn't from our current router for this dst,
282 	 * it's either old or wrong.  If it redirects us to ourselves,
283 	 * we have a routing loop, perhaps as a result of an interface
284 	 * going down recently.
285 	 */
286 #define	equal(a1, a2) (bcmp((caddr_t)(a1), (caddr_t)(a2), (a1)->sa_len) == 0)
287 	if (!(flags & RTF_DONE) && rt &&
288 	     (!equal(src, rt->rt_gateway) || rt->rt_ifa != ifa))
289 		error = EINVAL;
290 	else if (ifa_ifwithaddr(gateway))
291 		error = EHOSTUNREACH;
292 	if (error)
293 		goto done;
294 	/*
295 	 * Create a new entry if we just got back a wildcard entry
296 	 * or the the lookup failed.  This is necessary for hosts
297 	 * which use routing redirects generated by smart gateways
298 	 * to dynamically build the routing tables.
299 	 */
300 	if ((rt == 0) || (rt_mask(rt) && rt_mask(rt)->sa_len < 2))
301 		goto create;
302 	/*
303 	 * Don't listen to the redirect if it's
304 	 * for a route to an interface.
305 	 */
306 	if (rt->rt_flags & RTF_GATEWAY) {
307 		if (((rt->rt_flags & RTF_HOST) == 0) && (flags & RTF_HOST)) {
308 			/*
309 			 * Changing from route to net => route to host.
310 			 * Create new route, rather than smashing route to net.
311 			 */
312 		create:
313 			flags |=  RTF_GATEWAY | RTF_DYNAMIC;
314 			error = rtrequest((int)RTM_ADD, dst, gateway,
315 				    netmask, flags,
316 				    (struct rtentry **)0);
317 			stat = &rtstat.rts_dynamic;
318 		} else {
319 			/*
320 			 * Smash the current notion of the gateway to
321 			 * this destination.  Should check about netmask!!!
322 			 */
323 			rt->rt_flags |= RTF_MODIFIED;
324 			flags |= RTF_MODIFIED;
325 			stat = &rtstat.rts_newgateway;
326 			rt_setgate(rt, rt_key(rt), gateway);
327 		}
328 	} else
329 		error = EHOSTUNREACH;
330 done:
331 	if (rt) {
332 		if (rtp && !error)
333 			*rtp = rt;
334 		else
335 			rtfree(rt);
336 	}
337 out:
338 	if (error)
339 		rtstat.rts_badredirect++;
340 	else if (stat != NULL)
341 		(*stat)++;
342 	bzero((caddr_t)&info, sizeof(info));
343 	info.rti_info[RTAX_DST] = dst;
344 	info.rti_info[RTAX_GATEWAY] = gateway;
345 	info.rti_info[RTAX_NETMASK] = netmask;
346 	info.rti_info[RTAX_AUTHOR] = src;
347 	rt_missmsg(RTM_REDIRECT, &info, flags, error);
348 }
349 
350 /*
351  * Routing table ioctl interface.
352  */
353 int
354 rtioctl(req, data, p)
355 	u_long req;
356 	caddr_t data;
357 	struct proc *p;
358 {
359 	return (EOPNOTSUPP);
360 }
361 
362 struct ifaddr *
363 ifa_ifwithroute(flags, dst, gateway)
364 	int flags;
365 	struct sockaddr	*dst, *gateway;
366 {
367 	register struct ifaddr *ifa;
368 	if ((flags & RTF_GATEWAY) == 0) {
369 		/*
370 		 * If we are adding a route to an interface,
371 		 * and the interface is a pt to pt link
372 		 * we should search for the destination
373 		 * as our clue to the interface.  Otherwise
374 		 * we can use the local address.
375 		 */
376 		ifa = 0;
377 		if (flags & RTF_HOST)
378 			ifa = ifa_ifwithdstaddr(dst);
379 		if (ifa == 0)
380 			ifa = ifa_ifwithaddr(gateway);
381 	} else {
382 		/*
383 		 * If we are adding a route to a remote net
384 		 * or host, the gateway may still be on the
385 		 * other end of a pt to pt link.
386 		 */
387 		ifa = ifa_ifwithdstaddr(gateway);
388 	}
389 	if (ifa == 0)
390 		ifa = ifa_ifwithnet(gateway);
391 	if (ifa == 0) {
392 		struct rtentry *rt = rtalloc1(dst, 0);
393 		if (rt == 0)
394 			return (0);
395 		rt->rt_refcnt--;
396 		if ((ifa = rt->rt_ifa) == 0)
397 			return (0);
398 	}
399 	if (ifa->ifa_addr->sa_family != dst->sa_family) {
400 		struct ifaddr *oifa = ifa;
401 		ifa = ifaof_ifpforaddr(dst, ifa->ifa_ifp);
402 		if (ifa == 0)
403 			ifa = oifa;
404 	}
405 	return (ifa);
406 }
407 
408 #define ROUNDUP(a) (a>0 ? (1 + (((a) - 1) | (sizeof(long) - 1))) : sizeof(long))
409 
410 int
411 rtrequest(req, dst, gateway, netmask, flags, ret_nrt)
412 	int req, flags;
413 	struct sockaddr *dst, *gateway, *netmask;
414 	struct rtentry **ret_nrt;
415 {
416 	int s = splsoftnet(); int error = 0;
417 	register struct rtentry *rt;
418 	register struct radix_node *rn;
419 	register struct radix_node_head *rnh;
420 	struct ifaddr *ifa;
421 	struct sockaddr *ndst;
422 #define senderr(x) { error = x ; goto bad; }
423 
424 	if ((rnh = rt_tables[dst->sa_family]) == 0)
425 		senderr(ESRCH);
426 	if (flags & RTF_HOST)
427 		netmask = 0;
428 	switch (req) {
429 	case RTM_DELETE:
430 		if ((rn = rnh->rnh_deladdr(dst, netmask, rnh)) == 0)
431 			senderr(ESRCH);
432 		if (rn->rn_flags & (RNF_ACTIVE | RNF_ROOT))
433 			panic ("rtrequest delete");
434 		rt = (struct rtentry *)rn;
435 		if (rt->rt_gwroute) {
436 			rt = rt->rt_gwroute; RTFREE(rt);
437 			(rt = (struct rtentry *)rn)->rt_gwroute = 0;
438 		}
439 		rt->rt_flags &= ~RTF_UP;
440 		if ((ifa = rt->rt_ifa) && ifa->ifa_rtrequest)
441 			ifa->ifa_rtrequest(RTM_DELETE, rt, SA(0));
442 		rttrash++;
443 		if (ret_nrt)
444 			*ret_nrt = rt;
445 		else if (rt->rt_refcnt <= 0) {
446 			rt->rt_refcnt++;
447 			rtfree(rt);
448 		}
449 		break;
450 
451 	case RTM_RESOLVE:
452 		if (ret_nrt == 0 || (rt = *ret_nrt) == 0)
453 			senderr(EINVAL);
454 		ifa = rt->rt_ifa;
455 		flags = rt->rt_flags & ~RTF_CLONING;
456 		gateway = rt->rt_gateway;
457 		if ((netmask = rt->rt_genmask) == 0)
458 			flags |= RTF_HOST;
459 		goto makeroute;
460 
461 	case RTM_ADD:
462 		if ((ifa = ifa_ifwithroute(flags, dst, gateway)) == 0)
463 			senderr(ENETUNREACH);
464 
465 		/* The interface found in the previous statement may
466 		 * be overridden later by rt_setif.  See the code
467 		 * for case RTM_ADD in rtsock.c:route_output.
468 		 */
469 	makeroute:
470 		rt = pool_get(&rtentry_pool, PR_NOWAIT);
471 		if (rt == 0)
472 			senderr(ENOBUFS);
473 		Bzero(rt, sizeof(*rt));
474 		rt->rt_flags = RTF_UP | flags;
475 		LIST_INIT(&rt->rt_timer);
476 		if (rt_setgate(rt, dst, gateway)) {
477 			pool_put(&rtentry_pool, rt);
478 			senderr(ENOBUFS);
479 		}
480 		ndst = rt_key(rt);
481 		if (netmask) {
482 			rt_maskedcopy(dst, ndst, netmask);
483 		} else
484 			Bcopy(dst, ndst, dst->sa_len);
485 		rn = rnh->rnh_addaddr((caddr_t)ndst, (caddr_t)netmask,
486 					rnh, rt->rt_nodes);
487 		if (rn == 0) {
488 			if (rt->rt_gwroute)
489 				rtfree(rt->rt_gwroute);
490 			Free(rt_key(rt));
491 			pool_put(&rtentry_pool, rt);
492 			senderr(EEXIST);
493 		}
494 		IFAREF(ifa);
495 		rt->rt_ifa = ifa;
496 		rt->rt_ifp = ifa->ifa_ifp;
497 		if (req == RTM_RESOLVE) {
498 			rt->rt_rmx = (*ret_nrt)->rt_rmx; /* copy metrics */
499 		} else if (rt->rt_rmx.rmx_mtu == 0
500 			    && !(rt->rt_rmx.rmx_locks & RTV_MTU)) { /* XXX */
501 			if (rt->rt_gwroute != NULL) {
502 				rt->rt_rmx.rmx_mtu = rt->rt_gwroute->rt_rmx.rmx_mtu;
503 			} else {
504 				rt->rt_rmx.rmx_mtu = ifa->ifa_ifp->if_mtu;
505 			}
506 		}
507 		if (ifa->ifa_rtrequest)
508 			ifa->ifa_rtrequest(req, rt, SA(ret_nrt ? *ret_nrt : 0));
509 		if (ret_nrt) {
510 			*ret_nrt = rt;
511 			rt->rt_refcnt++;
512 		}
513 		break;
514 	}
515 bad:
516 	splx(s);
517 	return (error);
518 }
519 
520 int
521 rt_setgate(rt0, dst, gate)
522 	struct rtentry *rt0;
523 	struct sockaddr *dst, *gate;
524 {
525 	caddr_t new, old;
526 	int dlen = ROUNDUP(dst->sa_len), glen = ROUNDUP(gate->sa_len);
527 	register struct rtentry *rt = rt0;
528 
529 	if (rt->rt_gateway == 0 || glen > ROUNDUP(rt->rt_gateway->sa_len)) {
530 		old = (caddr_t)rt_key(rt);
531 		R_Malloc(new, caddr_t, dlen + glen);
532 		if (new == 0)
533 			return 1;
534 		Bzero(new, dlen + glen);
535 		rt->rt_nodes->rn_key = new;
536 	} else {
537 		new = rt->rt_nodes->rn_key;
538 		old = 0;
539 	}
540 	Bcopy(gate, (rt->rt_gateway = (struct sockaddr *)(new + dlen)), glen);
541 	if (old) {
542 		Bcopy(dst, new, dlen);
543 		Free(old);
544 	}
545 	if (rt->rt_gwroute) {
546 		rt = rt->rt_gwroute; RTFREE(rt);
547 		rt = rt0; rt->rt_gwroute = 0;
548 	}
549 	if (rt->rt_flags & RTF_GATEWAY) {
550 		rt->rt_gwroute = rtalloc1(gate, 1);
551 		/*
552 		 * If we switched gateways, grab the MTU from the new
553 		 * gateway route if the current MTU is 0 or greater
554 		 * than the MTU of gateway.
555 		 */
556 		if (rt->rt_gwroute
557 		    && !(rt->rt_rmx.rmx_locks & RTV_MTU)
558 		    && (rt->rt_rmx.rmx_mtu == 0
559 		    || rt->rt_rmx.rmx_mtu > rt->rt_gwroute->rt_rmx.rmx_mtu)) { /* XXX */
560 			rt->rt_rmx.rmx_mtu = rt->rt_gwroute->rt_rmx.rmx_mtu;
561 		}
562 	}
563 	return 0;
564 }
565 
566 void
567 rt_maskedcopy(src, dst, netmask)
568 	struct sockaddr *src, *dst, *netmask;
569 {
570 	register u_char *cp1 = (u_char *)src;
571 	register u_char *cp2 = (u_char *)dst;
572 	register u_char *cp3 = (u_char *)netmask;
573 	u_char *cplim = cp2 + *cp3;
574 	u_char *cplim2 = cp2 + *cp1;
575 
576 	*cp2++ = *cp1++; *cp2++ = *cp1++; /* copies sa_len & sa_family */
577 	cp3 += 2;
578 	if (cplim > cplim2)
579 		cplim = cplim2;
580 	while (cp2 < cplim)
581 		*cp2++ = *cp1++ & *cp3++;
582 	if (cp2 < cplim2)
583 		bzero((caddr_t)cp2, (unsigned)(cplim2 - cp2));
584 }
585 
586 /*
587  * Set up or tear down a routing table entry, normally
588  * for an interface.
589  */
590 int
591 rtinit(ifa, cmd, flags)
592 	register struct ifaddr *ifa;
593 	int cmd, flags;
594 {
595 	register struct rtentry *rt;
596 	register struct sockaddr *dst, *odst;
597 	struct sockaddr_storage deldst;
598 	struct rtentry *nrt = 0;
599 	int error;
600 
601 	dst = flags & RTF_HOST ? ifa->ifa_dstaddr : ifa->ifa_addr;
602 	if (cmd == RTM_DELETE) {
603 		if ((flags & RTF_HOST) == 0 && ifa->ifa_netmask) {
604 			/* Delete subnet route for this interface */
605 			odst = dst;
606 			dst = (struct sockaddr *)&deldst;
607 			rt_maskedcopy(odst, dst, ifa->ifa_netmask);
608 		}
609 		if ((rt = rtalloc1(dst, 0)) != NULL) {
610 			rt->rt_refcnt--;
611 			if (rt->rt_ifa != ifa)
612 				return (flags & RTF_HOST ? EHOSTUNREACH
613 							: ENETUNREACH);
614 		}
615 	}
616 	error = rtrequest(cmd, dst, ifa->ifa_addr, ifa->ifa_netmask,
617 			flags | ifa->ifa_flags, &nrt);
618 	if (cmd == RTM_DELETE && error == 0 && (rt = nrt)) {
619 		rt_newaddrmsg(cmd, ifa, error, nrt);
620 		if (rt->rt_refcnt <= 0) {
621 			rt->rt_refcnt++;
622 			rtfree(rt);
623 		}
624 	}
625 	if (cmd == RTM_ADD && error == 0 && (rt = nrt)) {
626 		rt->rt_refcnt--;
627 		if (rt->rt_ifa != ifa) {
628 			printf("rtinit: wrong ifa (%p) was (%p)\n", ifa,
629 				rt->rt_ifa);
630 			if (rt->rt_ifa->ifa_rtrequest)
631 			    rt->rt_ifa->ifa_rtrequest(RTM_DELETE, rt, SA(0));
632 			IFAFREE(rt->rt_ifa);
633 			rt->rt_ifa = ifa;
634 			rt->rt_ifp = ifa->ifa_ifp;
635 			rt->rt_rmx.rmx_mtu = ifa->ifa_ifp->if_mtu;	/*XXX*/
636 			IFAREF(ifa);
637 			if (ifa->ifa_rtrequest)
638 			    ifa->ifa_rtrequest(RTM_ADD, rt, SA(0));
639 		}
640 		rt_newaddrmsg(cmd, ifa, error, nrt);
641 	}
642 	return (error);
643 }
644 
645 /*
646  * Route timer routines.  These routes allow functions to be called
647  * for various routes at any time.  This is useful in supporting
648  * path MTU discovery and redirect route deletion.
649  *
650  * This is similar to some BSDI internal functions, but it provides
651  * for multiple queues for efficiency's sake...
652  */
653 
654 LIST_HEAD(, rttimer_queue) rttimer_queue_head;
655 static int rt_init_done = 0;
656 
657 #define RTTIMER_CALLOUT(r)	{				\
658 	if (r->rtt_func != NULL) {				\
659 		(*r->rtt_func)(r->rtt_rt, r);			\
660 	} else {						\
661 		rtrequest((int) RTM_DELETE,			\
662 			  (struct sockaddr *)rt_key(r->rtt_rt),	\
663 			  0, 0, 0, 0);				\
664 	}							\
665 }
666 
667 /*
668  * Some subtle order problems with domain initialization mean that
669  * we cannot count on this being run from rt_init before various
670  * protocol initializations are done.  Therefore, we make sure
671  * that this is run when the first queue is added...
672  */
673 
674 void
675 rt_timer_init()
676 {
677 	assert(rt_init_done == 0);
678 
679 	pool_init(&rttimer_pool, sizeof(struct rttimer), 0, 0, 0, "rttmrpl",
680 	    0, NULL, NULL, M_RTABLE);
681 
682 	LIST_INIT(&rttimer_queue_head);
683 	timeout(rt_timer_timer, NULL, hz);  /* every second */
684 	rt_init_done = 1;
685 }
686 
687 struct rttimer_queue *
688 rt_timer_queue_create(timeout)
689 	u_int	timeout;
690 {
691 	struct rttimer_queue *rtq;
692 
693 	if (rt_init_done == 0)
694 		rt_timer_init();
695 
696 	R_Malloc(rtq, struct rttimer_queue *, sizeof *rtq);
697 	if (rtq == NULL)
698 		return (NULL);
699 	Bzero(rtq, sizeof *rtq);
700 
701 	rtq->rtq_timeout = timeout;
702 	TAILQ_INIT(&rtq->rtq_head);
703 	LIST_INSERT_HEAD(&rttimer_queue_head, rtq, rtq_link);
704 
705 	return (rtq);
706 }
707 
708 void
709 rt_timer_queue_change(rtq, timeout)
710 	struct rttimer_queue *rtq;
711 	long timeout;
712 {
713 
714 	rtq->rtq_timeout = timeout;
715 }
716 
717 
718 void
719 rt_timer_queue_destroy(rtq, destroy)
720 	struct rttimer_queue *rtq;
721 	int destroy;
722 {
723 	struct rttimer *r;
724 
725 	while ((r = TAILQ_FIRST(&rtq->rtq_head)) != NULL) {
726 		LIST_REMOVE(r, rtt_link);
727 		TAILQ_REMOVE(&rtq->rtq_head, r, rtt_next);
728 		if (destroy)
729 			RTTIMER_CALLOUT(r);
730 		pool_put(&rttimer_pool, r);
731 	}
732 
733 	LIST_REMOVE(rtq, rtq_link);
734 
735 	/*
736 	 * Caller is responsible for freeing the rttimer_queue structure.
737 	 */
738 }
739 
740 void
741 rt_timer_remove_all(rt)
742 	struct rtentry *rt;
743 {
744 	struct rttimer *r;
745 
746 	while ((r = LIST_FIRST(&rt->rt_timer)) != NULL) {
747 		LIST_REMOVE(r, rtt_link);
748 		TAILQ_REMOVE(&r->rtt_queue->rtq_head, r, rtt_next);
749 		pool_put(&rttimer_pool, r);
750 	}
751 }
752 
753 int
754 rt_timer_add(rt, func, queue)
755 	struct rtentry *rt;
756 	void(*func) __P((struct rtentry *, struct rttimer *));
757 	struct rttimer_queue *queue;
758 {
759 	struct rttimer *r;
760 	long current_time;
761 	int s;
762 
763 	s = splclock();
764 	current_time = mono_time.tv_sec;
765 	splx(s);
766 
767 	/*
768 	 * If there's already a timer with this action, destroy it before
769 	 * we add a new one.
770 	 */
771 	for (r = LIST_FIRST(&rt->rt_timer); r != NULL;
772 	     r = LIST_NEXT(r, rtt_link)) {
773 		if (r->rtt_func == func) {
774 			LIST_REMOVE(r, rtt_link);
775 			TAILQ_REMOVE(&r->rtt_queue->rtq_head, r, rtt_next);
776 			pool_put(&rttimer_pool, r);
777 			break;  /* only one per list, so we can quit... */
778 		}
779 	}
780 
781 	r = pool_get(&rttimer_pool, PR_NOWAIT);
782 	if (r == NULL)
783 		return (ENOBUFS);
784 	Bzero(r, sizeof(*r));
785 
786 	r->rtt_rt = rt;
787 	r->rtt_time = current_time;
788 	r->rtt_func = func;
789 	r->rtt_queue = queue;
790 	LIST_INSERT_HEAD(&rt->rt_timer, r, rtt_link);
791 	TAILQ_INSERT_TAIL(&queue->rtq_head, r, rtt_next);
792 
793 	return (0);
794 }
795 
796 /* ARGSUSED */
797 void
798 rt_timer_timer(arg)
799 	void *arg;
800 {
801 	struct rttimer_queue *rtq;
802 	struct rttimer *r;
803 	long current_time;
804 	int s;
805 
806 	s = splclock();
807 	current_time = mono_time.tv_sec;
808 	splx(s);
809 
810 	s = splsoftnet();
811 	for (rtq = LIST_FIRST(&rttimer_queue_head); rtq != NULL;
812 	     rtq = LIST_NEXT(rtq, rtq_link)) {
813 		while ((r = TAILQ_FIRST(&rtq->rtq_head)) != NULL &&
814 		    (r->rtt_time + rtq->rtq_timeout) < current_time) {
815 			LIST_REMOVE(r, rtt_link);
816 			TAILQ_REMOVE(&rtq->rtq_head, r, rtt_next);
817 			RTTIMER_CALLOUT(r);
818 			pool_put(&rttimer_pool, r);
819 		}
820 	}
821 	splx(s);
822 
823 	timeout(rt_timer_timer, NULL, hz);  /* every second */
824 }
825