FreeBSD/Linux Kernel Cross Reference
sys/net/route.c
1 /* $NetBSD: route.c,v 1.58 2003/08/07 16:32:57 agc 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. Neither the name of the University nor the names of its contributors
82 * may be used to endorse or promote products derived from this software
83 * without specific prior written permission.
84 *
85 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
86 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
87 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
88 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
89 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
90 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
91 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
92 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
93 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
94 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
95 * SUCH DAMAGE.
96 *
97 * @(#)route.c 8.3 (Berkeley) 1/9/95
98 */
99
100 #include <sys/cdefs.h>
101 __KERNEL_RCSID(0, "$NetBSD: route.c,v 1.58 2003/08/07 16:32:57 agc Exp $");
102
103 #include "opt_ns.h"
104
105 #include <sys/param.h>
106 #include <sys/systm.h>
107 #include <sys/callout.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 struct route_cb route_cb;
132 struct rtstat rtstat;
133 struct radix_node_head *rt_tables[AF_MAX+1];
134
135 int rttrash; /* routes not in table but not freed */
136 struct sockaddr wildcard; /* zero valued cookie for wildcard searches */
137
138 struct pool rtentry_pool; /* pool for rtentry structures */
139 struct pool rttimer_pool; /* pool for rttimer structures */
140
141 struct callout rt_timer_ch; /* callout for rt_timer_timer() */
142
143 static int rtdeletemsg __P((struct rtentry *));
144 static int rtflushclone1 __P((struct radix_node *, void *));
145 static void rtflushclone __P((struct radix_node_head *, struct rtentry *));
146
147 void
148 rtable_init(table)
149 void **table;
150 {
151 struct domain *dom;
152 for (dom = domains; dom; dom = dom->dom_next)
153 if (dom->dom_rtattach)
154 dom->dom_rtattach(&table[dom->dom_family],
155 dom->dom_rtoffset);
156 }
157
158 void
159 route_init()
160 {
161
162 pool_init(&rtentry_pool, sizeof(struct rtentry), 0, 0, 0, "rtentpl",
163 NULL);
164
165 rn_init(); /* initialize all zeroes, all ones, mask table */
166 rtable_init((void **)rt_tables);
167 }
168
169 /*
170 * Packet routing routines.
171 */
172 void
173 rtalloc(ro)
174 struct route *ro;
175 {
176 if (ro->ro_rt && ro->ro_rt->rt_ifp && (ro->ro_rt->rt_flags & RTF_UP))
177 return; /* XXX */
178 ro->ro_rt = rtalloc1(&ro->ro_dst, 1);
179 }
180
181 struct rtentry *
182 rtalloc1(dst, report)
183 struct sockaddr *dst;
184 int report;
185 {
186 struct radix_node_head *rnh = rt_tables[dst->sa_family];
187 struct rtentry *rt;
188 struct radix_node *rn;
189 struct rtentry *newrt = 0;
190 struct rt_addrinfo info;
191 int s = splsoftnet(), err = 0, msgtype = RTM_MISS;
192
193 if (rnh && (rn = rnh->rnh_matchaddr((caddr_t)dst, rnh)) &&
194 ((rn->rn_flags & RNF_ROOT) == 0)) {
195 newrt = rt = (struct rtentry *)rn;
196 if (report && (rt->rt_flags & RTF_CLONING)) {
197 err = rtrequest(RTM_RESOLVE, dst, SA(0),
198 SA(0), 0, &newrt);
199 if (err) {
200 newrt = rt;
201 rt->rt_refcnt++;
202 goto miss;
203 }
204 if ((rt = newrt) && (rt->rt_flags & RTF_XRESOLVE)) {
205 msgtype = RTM_RESOLVE;
206 goto miss;
207 }
208 /* Inform listeners of the new route */
209 memset(&info, 0, sizeof(info));
210 info.rti_info[RTAX_DST] = rt_key(rt);
211 info.rti_info[RTAX_NETMASK] = rt_mask(rt);
212 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
213 if (rt->rt_ifp != NULL) {
214 info.rti_info[RTAX_IFP] =
215 TAILQ_FIRST(&rt->rt_ifp->if_addrlist)->ifa_addr;
216 info.rti_info[RTAX_IFA] = rt->rt_ifa->ifa_addr;
217 }
218 rt_missmsg(RTM_ADD, &info, rt->rt_flags, 0);
219 } else
220 rt->rt_refcnt++;
221 } else {
222 rtstat.rts_unreach++;
223 miss: if (report) {
224 memset((caddr_t)&info, 0, sizeof(info));
225 info.rti_info[RTAX_DST] = dst;
226 rt_missmsg(msgtype, &info, 0, err);
227 }
228 }
229 splx(s);
230 return (newrt);
231 }
232
233 void
234 rtfree(rt)
235 struct rtentry *rt;
236 {
237 struct ifaddr *ifa;
238
239 if (rt == 0)
240 panic("rtfree");
241 rt->rt_refcnt--;
242 if (rt->rt_refcnt <= 0 && (rt->rt_flags & RTF_UP) == 0) {
243 if (rt->rt_nodes->rn_flags & (RNF_ACTIVE | RNF_ROOT))
244 panic ("rtfree 2");
245 rttrash--;
246 if (rt->rt_refcnt < 0) {
247 printf("rtfree: %p not freed (neg refs)\n", rt);
248 return;
249 }
250 rt_timer_remove_all(rt, 0);
251 ifa = rt->rt_ifa;
252 IFAFREE(ifa);
253 Free(rt_key(rt));
254 pool_put(&rtentry_pool, rt);
255 }
256 }
257
258 void
259 ifafree(ifa)
260 struct ifaddr *ifa;
261 {
262
263 #ifdef DIAGNOSTIC
264 if (ifa == NULL)
265 panic("ifafree: null ifa");
266 if (ifa->ifa_refcnt != 0)
267 panic("ifafree: ifa_refcnt != 0 (%d)", ifa->ifa_refcnt);
268 #endif
269 #ifdef IFAREF_DEBUG
270 printf("ifafree: freeing ifaddr %p\n", ifa);
271 #endif
272 free(ifa, M_IFADDR);
273 }
274
275 /*
276 * Force a routing table entry to the specified
277 * destination to go through the given gateway.
278 * Normally called as a result of a routing redirect
279 * message from the network layer.
280 *
281 * N.B.: must be called at splsoftnet
282 */
283 void
284 rtredirect(dst, gateway, netmask, flags, src, rtp)
285 struct sockaddr *dst, *gateway, *netmask, *src;
286 int flags;
287 struct rtentry **rtp;
288 {
289 struct rtentry *rt;
290 int error = 0;
291 u_quad_t *stat = 0;
292 struct rt_addrinfo info;
293 struct ifaddr *ifa;
294
295 /* verify the gateway is directly reachable */
296 if ((ifa = ifa_ifwithnet(gateway)) == 0) {
297 error = ENETUNREACH;
298 goto out;
299 }
300 rt = rtalloc1(dst, 0);
301 /*
302 * If the redirect isn't from our current router for this dst,
303 * it's either old or wrong. If it redirects us to ourselves,
304 * we have a routing loop, perhaps as a result of an interface
305 * going down recently.
306 */
307 #define equal(a1, a2) \
308 ((a1)->sa_len == (a2)->sa_len && \
309 bcmp((caddr_t)(a1), (caddr_t)(a2), (a1)->sa_len) == 0)
310 if (!(flags & RTF_DONE) && rt &&
311 (!equal(src, rt->rt_gateway) || rt->rt_ifa != ifa))
312 error = EINVAL;
313 else if (ifa_ifwithaddr(gateway))
314 error = EHOSTUNREACH;
315 if (error)
316 goto done;
317 /*
318 * Create a new entry if we just got back a wildcard entry
319 * or the lookup failed. This is necessary for hosts
320 * which use routing redirects generated by smart gateways
321 * to dynamically build the routing tables.
322 */
323 if ((rt == 0) || (rt_mask(rt) && rt_mask(rt)->sa_len < 2))
324 goto create;
325 /*
326 * Don't listen to the redirect if it's
327 * for a route to an interface.
328 */
329 if (rt->rt_flags & RTF_GATEWAY) {
330 if (((rt->rt_flags & RTF_HOST) == 0) && (flags & RTF_HOST)) {
331 /*
332 * Changing from route to net => route to host.
333 * Create new route, rather than smashing route to net.
334 */
335 create:
336 if (rt)
337 rtfree(rt);
338 flags |= RTF_GATEWAY | RTF_DYNAMIC;
339 info.rti_info[RTAX_DST] = dst;
340 info.rti_info[RTAX_GATEWAY] = gateway;
341 info.rti_info[RTAX_NETMASK] = netmask;
342 info.rti_ifa = ifa;
343 info.rti_flags = flags;
344 rt = NULL;
345 error = rtrequest1(RTM_ADD, &info, &rt);
346 if (rt != NULL)
347 flags = rt->rt_flags;
348 stat = &rtstat.rts_dynamic;
349 } else {
350 /*
351 * Smash the current notion of the gateway to
352 * this destination. Should check about netmask!!!
353 */
354 rt->rt_flags |= RTF_MODIFIED;
355 flags |= RTF_MODIFIED;
356 stat = &rtstat.rts_newgateway;
357 rt_setgate(rt, rt_key(rt), gateway);
358 }
359 } else
360 error = EHOSTUNREACH;
361 done:
362 if (rt) {
363 if (rtp && !error)
364 *rtp = rt;
365 else
366 rtfree(rt);
367 }
368 out:
369 if (error)
370 rtstat.rts_badredirect++;
371 else if (stat != NULL)
372 (*stat)++;
373 memset((caddr_t)&info, 0, sizeof(info));
374 info.rti_info[RTAX_DST] = dst;
375 info.rti_info[RTAX_GATEWAY] = gateway;
376 info.rti_info[RTAX_NETMASK] = netmask;
377 info.rti_info[RTAX_AUTHOR] = src;
378 rt_missmsg(RTM_REDIRECT, &info, flags, error);
379 }
380
381 /*
382 * Delete a route and generate a message
383 */
384 static int
385 rtdeletemsg(rt)
386 struct rtentry *rt;
387 {
388 int error;
389 struct rt_addrinfo info;
390
391 /*
392 * Request the new route so that the entry is not actually
393 * deleted. That will allow the information being reported to
394 * be accurate (and consistent with route_output()).
395 */
396 memset((caddr_t)&info, 0, sizeof(info));
397 info.rti_info[RTAX_DST] = rt_key(rt);
398 info.rti_info[RTAX_NETMASK] = rt_mask(rt);
399 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
400 info.rti_flags = rt->rt_flags;
401 error = rtrequest1(RTM_DELETE, &info, &rt);
402
403 rt_missmsg(RTM_DELETE, &info, info.rti_flags, error);
404
405 /* Adjust the refcount */
406 if (error == 0 && rt->rt_refcnt <= 0) {
407 rt->rt_refcnt++;
408 rtfree(rt);
409 }
410 return (error);
411 }
412
413 static int
414 rtflushclone1(rn, arg)
415 struct radix_node *rn;
416 void *arg;
417 {
418 struct rtentry *rt, *parent;
419
420 rt = (struct rtentry *)rn;
421 parent = (struct rtentry *)arg;
422 if ((rt->rt_flags & RTF_CLONED) != 0 && rt->rt_parent == parent)
423 rtdeletemsg(rt);
424 return 0;
425 }
426
427 static void
428 rtflushclone(rnh, parent)
429 struct radix_node_head *rnh;
430 struct rtentry *parent;
431 {
432
433 #ifdef DIAGNOSTIC
434 if (!parent || (parent->rt_flags & RTF_CLONING) == 0)
435 panic("rtflushclone: called with a non-cloning route");
436 if (!rnh->rnh_walktree)
437 panic("rtflushclone: no rnh_walktree");
438 #endif
439 rnh->rnh_walktree(rnh, rtflushclone1, (void *)parent);
440 }
441
442 /*
443 * Routing table ioctl interface.
444 */
445 int
446 rtioctl(req, data, p)
447 u_long req;
448 caddr_t data;
449 struct proc *p;
450 {
451 return (EOPNOTSUPP);
452 }
453
454 struct ifaddr *
455 ifa_ifwithroute(flags, dst, gateway)
456 int flags;
457 struct sockaddr *dst, *gateway;
458 {
459 struct ifaddr *ifa;
460 if ((flags & RTF_GATEWAY) == 0) {
461 /*
462 * If we are adding a route to an interface,
463 * and the interface is a pt to pt link
464 * we should search for the destination
465 * as our clue to the interface. Otherwise
466 * we can use the local address.
467 */
468 ifa = 0;
469 if (flags & RTF_HOST)
470 ifa = ifa_ifwithdstaddr(dst);
471 if (ifa == 0)
472 ifa = ifa_ifwithaddr(gateway);
473 } else {
474 /*
475 * If we are adding a route to a remote net
476 * or host, the gateway may still be on the
477 * other end of a pt to pt link.
478 */
479 ifa = ifa_ifwithdstaddr(gateway);
480 }
481 if (ifa == 0)
482 ifa = ifa_ifwithnet(gateway);
483 if (ifa == 0) {
484 struct rtentry *rt = rtalloc1(dst, 0);
485 if (rt == 0)
486 return (0);
487 rt->rt_refcnt--;
488 if ((ifa = rt->rt_ifa) == 0)
489 return (0);
490 }
491 if (ifa->ifa_addr->sa_family != dst->sa_family) {
492 struct ifaddr *oifa = ifa;
493 ifa = ifaof_ifpforaddr(dst, ifa->ifa_ifp);
494 if (ifa == 0)
495 ifa = oifa;
496 }
497 return (ifa);
498 }
499
500 #define ROUNDUP(a) (a>0 ? (1 + (((a) - 1) | (sizeof(long) - 1))) : sizeof(long))
501
502 int
503 rtrequest(req, dst, gateway, netmask, flags, ret_nrt)
504 int req, flags;
505 struct sockaddr *dst, *gateway, *netmask;
506 struct rtentry **ret_nrt;
507 {
508 struct rt_addrinfo info;
509
510 memset(&info, 0, sizeof(info));
511 info.rti_flags = flags;
512 info.rti_info[RTAX_DST] = dst;
513 info.rti_info[RTAX_GATEWAY] = gateway;
514 info.rti_info[RTAX_NETMASK] = netmask;
515 return rtrequest1(req, &info, ret_nrt);
516 }
517
518 /*
519 * These (questionable) definitions of apparent local variables apply
520 * to the next function. XXXXXX!!!
521 */
522 #define dst info->rti_info[RTAX_DST]
523 #define gateway info->rti_info[RTAX_GATEWAY]
524 #define netmask info->rti_info[RTAX_NETMASK]
525 #define ifaaddr info->rti_info[RTAX_IFA]
526 #define ifpaddr info->rti_info[RTAX_IFP]
527 #define flags info->rti_flags
528
529 int
530 rt_getifa(info)
531 struct rt_addrinfo *info;
532 {
533 struct ifaddr *ifa;
534 int error = 0;
535
536 /*
537 * ifp may be specified by sockaddr_dl when protocol address
538 * is ambiguous
539 */
540 if (info->rti_ifp == NULL && ifpaddr != NULL
541 && ifpaddr->sa_family == AF_LINK &&
542 (ifa = ifa_ifwithnet((struct sockaddr *)ifpaddr)) != NULL)
543 info->rti_ifp = ifa->ifa_ifp;
544 if (info->rti_ifa == NULL && ifaaddr != NULL)
545 info->rti_ifa = ifa_ifwithaddr(ifaaddr);
546 if (info->rti_ifa == NULL) {
547 struct sockaddr *sa;
548
549 sa = ifaaddr != NULL ? ifaaddr :
550 (gateway != NULL ? gateway : dst);
551 if (sa != NULL && info->rti_ifp != NULL)
552 info->rti_ifa = ifaof_ifpforaddr(sa, info->rti_ifp);
553 else if (dst != NULL && gateway != NULL)
554 info->rti_ifa = ifa_ifwithroute(flags, dst, gateway);
555 else if (sa != NULL)
556 info->rti_ifa = ifa_ifwithroute(flags, sa, sa);
557 }
558 if ((ifa = info->rti_ifa) != NULL) {
559 if (info->rti_ifp == NULL)
560 info->rti_ifp = ifa->ifa_ifp;
561 } else
562 error = ENETUNREACH;
563 return (error);
564 }
565
566 int
567 rtrequest1(req, info, ret_nrt)
568 int req;
569 struct rt_addrinfo *info;
570 struct rtentry **ret_nrt;
571 {
572 int s = splsoftnet(); int error = 0;
573 struct rtentry *rt, *crt;
574 struct radix_node *rn;
575 struct radix_node_head *rnh;
576 struct ifaddr *ifa;
577 struct sockaddr *ndst;
578 #define senderr(x) { error = x ; goto bad; }
579
580 if ((rnh = rt_tables[dst->sa_family]) == 0)
581 senderr(ESRCH);
582 if (flags & RTF_HOST)
583 netmask = 0;
584 switch (req) {
585 case RTM_DELETE:
586 if ((rn = rnh->rnh_lookup(dst, netmask, rnh)) == 0)
587 senderr(ESRCH);
588 rt = (struct rtentry *)rn;
589 if ((rt->rt_flags & RTF_CLONING) != 0) {
590 /* clean up any cloned children */
591 rtflushclone(rnh, rt);
592 }
593 if ((rn = rnh->rnh_deladdr(dst, netmask, rnh)) == 0)
594 senderr(ESRCH);
595 if (rn->rn_flags & (RNF_ACTIVE | RNF_ROOT))
596 panic ("rtrequest delete");
597 rt = (struct rtentry *)rn;
598 if (rt->rt_gwroute) {
599 rt = rt->rt_gwroute; RTFREE(rt);
600 (rt = (struct rtentry *)rn)->rt_gwroute = 0;
601 }
602 if (rt->rt_parent) {
603 rt->rt_parent->rt_refcnt--;
604 rt->rt_parent = NULL;
605 }
606 rt->rt_flags &= ~RTF_UP;
607 if ((ifa = rt->rt_ifa) && ifa->ifa_rtrequest)
608 ifa->ifa_rtrequest(RTM_DELETE, rt, info);
609 rttrash++;
610 if (ret_nrt)
611 *ret_nrt = rt;
612 else if (rt->rt_refcnt <= 0) {
613 rt->rt_refcnt++;
614 rtfree(rt);
615 }
616 break;
617
618 case RTM_RESOLVE:
619 if (ret_nrt == 0 || (rt = *ret_nrt) == 0)
620 senderr(EINVAL);
621 if ((rt->rt_flags & RTF_CLONING) == 0)
622 senderr(EINVAL);
623 ifa = rt->rt_ifa;
624 flags = rt->rt_flags & ~(RTF_CLONING | RTF_STATIC);
625 flags |= RTF_CLONED;
626 gateway = rt->rt_gateway;
627 if ((netmask = rt->rt_genmask) == 0)
628 flags |= RTF_HOST;
629 goto makeroute;
630
631 case RTM_ADD:
632 if (info->rti_ifa == 0 && (error = rt_getifa(info)))
633 senderr(error);
634 ifa = info->rti_ifa;
635 makeroute:
636 rt = pool_get(&rtentry_pool, PR_NOWAIT);
637 if (rt == 0)
638 senderr(ENOBUFS);
639 Bzero(rt, sizeof(*rt));
640 rt->rt_flags = RTF_UP | flags;
641 LIST_INIT(&rt->rt_timer);
642 if (rt_setgate(rt, dst, gateway)) {
643 pool_put(&rtentry_pool, rt);
644 senderr(ENOBUFS);
645 }
646 ndst = rt_key(rt);
647 if (netmask) {
648 rt_maskedcopy(dst, ndst, netmask);
649 } else
650 Bcopy(dst, ndst, dst->sa_len);
651 IFAREF(ifa);
652 rt->rt_ifa = ifa;
653 rt->rt_ifp = ifa->ifa_ifp;
654 if (req == RTM_RESOLVE) {
655 rt->rt_rmx = (*ret_nrt)->rt_rmx; /* copy metrics */
656 rt->rt_parent = *ret_nrt;
657 rt->rt_parent->rt_refcnt++;
658 }
659 rn = rnh->rnh_addaddr((caddr_t)ndst, (caddr_t)netmask,
660 rnh, rt->rt_nodes);
661 if (rn == NULL && (crt = rtalloc1(ndst, 0)) != NULL) {
662 /* overwrite cloned route */
663 if ((crt->rt_flags & RTF_CLONED) != 0) {
664 rtdeletemsg(crt);
665 rn = rnh->rnh_addaddr((caddr_t)ndst,
666 (caddr_t)netmask, rnh, rt->rt_nodes);
667 }
668 RTFREE(crt);
669 }
670 if (rn == 0) {
671 IFAFREE(ifa);
672 if ((rt->rt_flags & RTF_CLONED) != 0 && rt->rt_parent)
673 rtfree(rt->rt_parent);
674 if (rt->rt_gwroute)
675 rtfree(rt->rt_gwroute);
676 Free(rt_key(rt));
677 pool_put(&rtentry_pool, rt);
678 senderr(EEXIST);
679 }
680 if (ifa->ifa_rtrequest)
681 ifa->ifa_rtrequest(req, rt, info);
682 if (ret_nrt) {
683 *ret_nrt = rt;
684 rt->rt_refcnt++;
685 }
686 if ((rt->rt_flags & RTF_CLONING) != 0) {
687 /* clean up any cloned children */
688 rtflushclone(rnh, rt);
689 }
690 break;
691 }
692 bad:
693 splx(s);
694 return (error);
695 }
696
697 #undef dst
698 #undef gateway
699 #undef netmask
700 #undef ifaaddr
701 #undef ifpaddr
702 #undef flags
703
704 int
705 rt_setgate(rt0, dst, gate)
706 struct rtentry *rt0;
707 struct sockaddr *dst, *gate;
708 {
709 caddr_t new, old;
710 u_int dlen = ROUNDUP(dst->sa_len), glen = ROUNDUP(gate->sa_len);
711 struct rtentry *rt = rt0;
712
713 if (rt->rt_gateway == 0 || glen > ROUNDUP(rt->rt_gateway->sa_len)) {
714 old = (caddr_t)rt_key(rt);
715 R_Malloc(new, caddr_t, dlen + glen);
716 if (new == 0)
717 return 1;
718 Bzero(new, dlen + glen);
719 rt->rt_nodes->rn_key = new;
720 } else {
721 new = rt->rt_nodes->rn_key;
722 old = 0;
723 }
724 Bcopy(gate, (rt->rt_gateway = (struct sockaddr *)(new + dlen)), glen);
725 if (old) {
726 Bcopy(dst, new, dlen);
727 Free(old);
728 }
729 if (rt->rt_gwroute) {
730 rt = rt->rt_gwroute; RTFREE(rt);
731 rt = rt0; rt->rt_gwroute = 0;
732 }
733 if (rt->rt_flags & RTF_GATEWAY) {
734 rt->rt_gwroute = rtalloc1(gate, 1);
735 /*
736 * If we switched gateways, grab the MTU from the new
737 * gateway route if the current MTU, if the current MTU is
738 * greater than the MTU of gateway.
739 * Note that, if the MTU of gateway is 0, we will reset the
740 * MTU of the route to run PMTUD again from scratch. XXX
741 */
742 if (rt->rt_gwroute
743 && !(rt->rt_rmx.rmx_locks & RTV_MTU)
744 && rt->rt_rmx.rmx_mtu
745 && rt->rt_rmx.rmx_mtu > rt->rt_gwroute->rt_rmx.rmx_mtu) {
746 rt->rt_rmx.rmx_mtu = rt->rt_gwroute->rt_rmx.rmx_mtu;
747 }
748 }
749 return 0;
750 }
751
752 void
753 rt_maskedcopy(src, dst, netmask)
754 struct sockaddr *src, *dst, *netmask;
755 {
756 u_char *cp1 = (u_char *)src;
757 u_char *cp2 = (u_char *)dst;
758 u_char *cp3 = (u_char *)netmask;
759 u_char *cplim = cp2 + *cp3;
760 u_char *cplim2 = cp2 + *cp1;
761
762 *cp2++ = *cp1++; *cp2++ = *cp1++; /* copies sa_len & sa_family */
763 cp3 += 2;
764 if (cplim > cplim2)
765 cplim = cplim2;
766 while (cp2 < cplim)
767 *cp2++ = *cp1++ & *cp3++;
768 if (cp2 < cplim2)
769 memset((caddr_t)cp2, 0, (unsigned)(cplim2 - cp2));
770 }
771
772 /*
773 * Set up or tear down a routing table entry, normally
774 * for an interface.
775 */
776 int
777 rtinit(ifa, cmd, flags)
778 struct ifaddr *ifa;
779 int cmd, flags;
780 {
781 struct rtentry *rt;
782 struct sockaddr *dst, *odst;
783 struct sockaddr_storage deldst;
784 struct rtentry *nrt = 0;
785 int error;
786 struct rt_addrinfo info;
787
788 dst = flags & RTF_HOST ? ifa->ifa_dstaddr : ifa->ifa_addr;
789 if (cmd == RTM_DELETE) {
790 if ((flags & RTF_HOST) == 0 && ifa->ifa_netmask) {
791 /* Delete subnet route for this interface */
792 odst = dst;
793 dst = (struct sockaddr *)&deldst;
794 rt_maskedcopy(odst, dst, ifa->ifa_netmask);
795 }
796 if ((rt = rtalloc1(dst, 0)) != NULL) {
797 rt->rt_refcnt--;
798 if (rt->rt_ifa != ifa)
799 return (flags & RTF_HOST ? EHOSTUNREACH
800 : ENETUNREACH);
801 }
802 }
803 memset(&info, 0, sizeof(info));
804 info.rti_ifa = ifa;
805 info.rti_flags = flags | ifa->ifa_flags;
806 info.rti_info[RTAX_DST] = dst;
807 info.rti_info[RTAX_GATEWAY] = ifa->ifa_addr;
808 /*
809 * XXX here, it seems that we are assuming that ifa_netmask is NULL
810 * for RTF_HOST. bsdi4 passes NULL explicitly (via intermediate
811 * variable) when RTF_HOST is 1. still not sure if i can safely
812 * change it to meet bsdi4 behavior.
813 */
814 info.rti_info[RTAX_NETMASK] = ifa->ifa_netmask;
815 error = rtrequest1(cmd, &info, &nrt);
816 if (cmd == RTM_DELETE && error == 0 && (rt = nrt)) {
817 rt_newaddrmsg(cmd, ifa, error, nrt);
818 if (rt->rt_refcnt <= 0) {
819 rt->rt_refcnt++;
820 rtfree(rt);
821 }
822 }
823 if (cmd == RTM_ADD && error == 0 && (rt = nrt)) {
824 rt->rt_refcnt--;
825 if (rt->rt_ifa != ifa) {
826 printf("rtinit: wrong ifa (%p) was (%p)\n", ifa,
827 rt->rt_ifa);
828 if (rt->rt_ifa->ifa_rtrequest)
829 rt->rt_ifa->ifa_rtrequest(RTM_DELETE, rt, NULL);
830 IFAFREE(rt->rt_ifa);
831 rt->rt_ifa = ifa;
832 rt->rt_ifp = ifa->ifa_ifp;
833 IFAREF(ifa);
834 if (ifa->ifa_rtrequest)
835 ifa->ifa_rtrequest(RTM_ADD, rt, NULL);
836 }
837 rt_newaddrmsg(cmd, ifa, error, nrt);
838 }
839 return (error);
840 }
841
842 /*
843 * Route timer routines. These routes allow functions to be called
844 * for various routes at any time. This is useful in supporting
845 * path MTU discovery and redirect route deletion.
846 *
847 * This is similar to some BSDI internal functions, but it provides
848 * for multiple queues for efficiency's sake...
849 */
850
851 LIST_HEAD(, rttimer_queue) rttimer_queue_head;
852 static int rt_init_done = 0;
853
854 #define RTTIMER_CALLOUT(r) { \
855 if (r->rtt_func != NULL) { \
856 (*r->rtt_func)(r->rtt_rt, r); \
857 } else { \
858 rtrequest((int) RTM_DELETE, \
859 (struct sockaddr *)rt_key(r->rtt_rt), \
860 0, 0, 0, 0); \
861 } \
862 }
863
864 /*
865 * Some subtle order problems with domain initialization mean that
866 * we cannot count on this being run from rt_init before various
867 * protocol initializations are done. Therefore, we make sure
868 * that this is run when the first queue is added...
869 */
870
871 void
872 rt_timer_init()
873 {
874 assert(rt_init_done == 0);
875
876 pool_init(&rttimer_pool, sizeof(struct rttimer), 0, 0, 0, "rttmrpl",
877 NULL);
878
879 LIST_INIT(&rttimer_queue_head);
880 callout_init(&rt_timer_ch);
881 callout_reset(&rt_timer_ch, hz, rt_timer_timer, NULL);
882 rt_init_done = 1;
883 }
884
885 struct rttimer_queue *
886 rt_timer_queue_create(timeout)
887 u_int timeout;
888 {
889 struct rttimer_queue *rtq;
890
891 if (rt_init_done == 0)
892 rt_timer_init();
893
894 R_Malloc(rtq, struct rttimer_queue *, sizeof *rtq);
895 if (rtq == NULL)
896 return (NULL);
897 Bzero(rtq, sizeof *rtq);
898
899 rtq->rtq_timeout = timeout;
900 rtq->rtq_count = 0;
901 TAILQ_INIT(&rtq->rtq_head);
902 LIST_INSERT_HEAD(&rttimer_queue_head, rtq, rtq_link);
903
904 return (rtq);
905 }
906
907 void
908 rt_timer_queue_change(rtq, timeout)
909 struct rttimer_queue *rtq;
910 long timeout;
911 {
912
913 rtq->rtq_timeout = timeout;
914 }
915
916 void
917 rt_timer_queue_remove_all(rtq, destroy)
918 struct rttimer_queue *rtq;
919 int destroy;
920 {
921 struct rttimer *r;
922
923 while ((r = TAILQ_FIRST(&rtq->rtq_head)) != NULL) {
924 LIST_REMOVE(r, rtt_link);
925 TAILQ_REMOVE(&rtq->rtq_head, r, rtt_next);
926 if (destroy)
927 RTTIMER_CALLOUT(r);
928 pool_put(&rttimer_pool, r);
929 if (rtq->rtq_count > 0)
930 rtq->rtq_count--;
931 else
932 printf("rt_timer_queue_remove_all: "
933 "rtq_count reached 0\n");
934 }
935 }
936
937 void
938 rt_timer_queue_destroy(rtq, destroy)
939 struct rttimer_queue *rtq;
940 int destroy;
941 {
942
943 rt_timer_queue_remove_all(rtq, destroy);
944
945 LIST_REMOVE(rtq, rtq_link);
946
947 /*
948 * Caller is responsible for freeing the rttimer_queue structure.
949 */
950 }
951
952 unsigned long
953 rt_timer_count(rtq)
954 struct rttimer_queue *rtq;
955 {
956
957 return rtq->rtq_count;
958 }
959
960 void
961 rt_timer_remove_all(rt, destroy)
962 struct rtentry *rt;
963 int destroy;
964 {
965 struct rttimer *r;
966
967 while ((r = LIST_FIRST(&rt->rt_timer)) != NULL) {
968 LIST_REMOVE(r, rtt_link);
969 TAILQ_REMOVE(&r->rtt_queue->rtq_head, r, rtt_next);
970 if (destroy)
971 RTTIMER_CALLOUT(r);
972 if (r->rtt_queue->rtq_count > 0)
973 r->rtt_queue->rtq_count--;
974 else
975 printf("rt_timer_remove_all: rtq_count reached 0\n");
976 pool_put(&rttimer_pool, r);
977 }
978 }
979
980 int
981 rt_timer_add(rt, func, queue)
982 struct rtentry *rt;
983 void(*func) __P((struct rtentry *, struct rttimer *));
984 struct rttimer_queue *queue;
985 {
986 struct rttimer *r;
987 long current_time;
988 int s;
989
990 s = splclock();
991 current_time = mono_time.tv_sec;
992 splx(s);
993
994 /*
995 * If there's already a timer with this action, destroy it before
996 * we add a new one.
997 */
998 for (r = LIST_FIRST(&rt->rt_timer); r != NULL;
999 r = LIST_NEXT(r, rtt_link)) {
1000 if (r->rtt_func == func) {
1001 LIST_REMOVE(r, rtt_link);
1002 TAILQ_REMOVE(&r->rtt_queue->rtq_head, r, rtt_next);
1003 if (r->rtt_queue->rtq_count > 0)
1004 r->rtt_queue->rtq_count--;
1005 else
1006 printf("rt_timer_add: rtq_count reached 0\n");
1007 pool_put(&rttimer_pool, r);
1008 break; /* only one per list, so we can quit... */
1009 }
1010 }
1011
1012 r = pool_get(&rttimer_pool, PR_NOWAIT);
1013 if (r == NULL)
1014 return (ENOBUFS);
1015 Bzero(r, sizeof(*r));
1016
1017 r->rtt_rt = rt;
1018 r->rtt_time = current_time;
1019 r->rtt_func = func;
1020 r->rtt_queue = queue;
1021 LIST_INSERT_HEAD(&rt->rt_timer, r, rtt_link);
1022 TAILQ_INSERT_TAIL(&queue->rtq_head, r, rtt_next);
1023 r->rtt_queue->rtq_count++;
1024
1025 return (0);
1026 }
1027
1028 /* ARGSUSED */
1029 void
1030 rt_timer_timer(arg)
1031 void *arg;
1032 {
1033 struct rttimer_queue *rtq;
1034 struct rttimer *r;
1035 long current_time;
1036 int s;
1037
1038 s = splclock();
1039 current_time = mono_time.tv_sec;
1040 splx(s);
1041
1042 s = splsoftnet();
1043 for (rtq = LIST_FIRST(&rttimer_queue_head); rtq != NULL;
1044 rtq = LIST_NEXT(rtq, rtq_link)) {
1045 while ((r = TAILQ_FIRST(&rtq->rtq_head)) != NULL &&
1046 (r->rtt_time + rtq->rtq_timeout) < current_time) {
1047 LIST_REMOVE(r, rtt_link);
1048 TAILQ_REMOVE(&rtq->rtq_head, r, rtt_next);
1049 RTTIMER_CALLOUT(r);
1050 pool_put(&rttimer_pool, r);
1051 if (rtq->rtq_count > 0)
1052 rtq->rtq_count--;
1053 else
1054 printf("rt_timer_timer: rtq_count reached 0\n");
1055 }
1056 }
1057 splx(s);
1058
1059 callout_reset(&rt_timer_ch, hz, rt_timer_timer, NULL);
1060 }
Cache object: c29405faca8a3d9cabc09747cbfcaa86
|