2 * SPDX-License-Identifier: BSD-3-Clause
4 * Copyright (c) 1980, 1986, 1991, 1993
5 * The Regents of the University of California. All rights reserved.
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
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 University 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.
19 * THIS SOFTWARE IS PROVIDED BY THE REGENTS 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 REGENTS 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
31 * @(#)route.c 8.3.1.1 (Berkeley) 2/23/95
34 /************************************************************************
35 * Note: In this file a 'fib' is a "forwarding information base" *
36 * Which is the new name for an in kernel routing (next hop) table. *
37 ***********************************************************************/
40 #include "opt_inet6.h"
41 #include "opt_mrouting.h"
42 #include "opt_route.h"
44 #include <sys/param.h>
45 #include <sys/systm.h>
46 #include <sys/malloc.h>
48 #include <sys/socket.h>
49 #include <sys/sysctl.h>
50 #include <sys/syslog.h>
51 #include <sys/sysproto.h>
53 #include <sys/domain.h>
54 #include <sys/eventhandler.h>
55 #include <sys/kernel.h>
57 #include <sys/rmlock.h>
60 #include <net/if_var.h>
61 #include <net/if_dl.h>
62 #include <net/route.h>
63 #include <net/route/route_ctl.h>
64 #include <net/route/route_var.h>
65 #include <net/route/nhop.h>
68 #include <netinet/in.h>
69 #include <netinet/ip_mroute.h>
71 VNET_PCPUSTAT_DEFINE(struct rtstat, rtstat);
73 VNET_PCPUSTAT_SYSINIT(rtstat);
75 VNET_PCPUSTAT_SYSUNINIT(rtstat);
78 EVENTHANDLER_LIST_DEFINE(rt_addrmsg);
80 static int rt_ifdelroute(const struct rtentry *rt, const struct nhop_object *,
82 static int rt_exportinfo(struct rtentry *rt, struct nhop_object *nh,
83 struct rt_addrinfo *info, int flags);
86 * route initialization must occur before ip6_init2(), which happenas at
95 SYSINIT(route_init, SI_SUB_PROTO_DOMAIN, SI_ORDER_THIRD, route_init, NULL);
98 rt_table_init(int offset, int family, u_int fibnum)
102 rh = malloc(sizeof(struct rib_head), M_RTABLE, M_WAITOK | M_ZERO);
104 /* TODO: These details should be hidded inside radix.c */
105 /* Init masks tree */
106 rn_inithead_internal(&rh->head, rh->rnh_nodes, offset);
107 rn_inithead_internal(&rh->rmhead.head, rh->rmhead.mask_nodes, 0);
108 rh->head.rnh_masks = &rh->rmhead;
110 /* Save metadata associated with this routing table. */
111 rh->rib_family = family;
112 rh->rib_fibnum = fibnum;
114 rh->rib_vnet = curvnet;
124 /* Init subscription system */
125 rib_init_subscriptions(rh);
127 /* Finally, set base callbacks */
128 rh->rnh_addaddr = rn_addroute;
129 rh->rnh_deladdr = rn_delete;
130 rh->rnh_matchaddr = rn_match;
131 rh->rnh_lookup = rn_lookup;
132 rh->rnh_walktree = rn_walktree;
133 rh->rnh_walktree_from = rn_walktree_from;
139 rt_freeentry(struct radix_node *rn, void *arg)
141 struct radix_head * const rnh = arg;
142 struct radix_node *x;
144 x = (struct radix_node *)rn_delete(rn + 2, NULL, rnh);
151 rt_table_destroy(struct rib_head *rh)
155 rh->rib_dying = true;
162 tmproutes_destroy(rh);
164 rn_walktree(&rh->rmhead.head, rt_freeentry, &rh->rmhead.head);
166 nhops_destroy_rib(rh);
168 rib_destroy_subscriptions(rh);
170 /* Assume table is already empty */
171 RIB_LOCK_DESTROY(rh);
176 * Adds a temporal redirect entry to the routing table.
177 * @fibnum: fib number
178 * @dst: destination to install redirect to
179 * @gateway: gateway to go via
180 * @author: sockaddr of originating router, can be NULL
181 * @ifp: interface to use for the redirected route
182 * @flags: set of flags to add. Allowed: RTF_GATEWAY
183 * @lifetime_sec: time in seconds to expire this redirect.
185 * Retuns 0 on success, errno otherwise.
188 rib_add_redirect(u_int fibnum, struct sockaddr *dst, struct sockaddr *gateway,
189 struct sockaddr *author, struct ifnet *ifp, int flags, int lifetime_sec)
191 struct rib_cmd_info rc;
193 struct rt_addrinfo info;
194 struct rt_metrics rti_rmx;
199 if (rt_tables_get_rnh(fibnum, dst->sa_family) == NULL)
200 return (EAFNOSUPPORT);
202 /* Verify the allowed flag mask. */
203 KASSERT(((flags & ~(RTF_GATEWAY)) == 0),
204 ("invalid redirect flags: %x", flags));
205 flags |= RTF_HOST | RTF_DYNAMIC;
207 /* Get the best ifa for the given interface and gateway. */
208 if ((ifa = ifaof_ifpforaddr(gateway, ifp)) == NULL)
209 return (ENETUNREACH);
211 bzero(&info, sizeof(info));
212 info.rti_info[RTAX_DST] = dst;
213 info.rti_info[RTAX_GATEWAY] = gateway;
216 info.rti_flags = flags;
218 /* Setup route metrics to define expire time. */
219 bzero(&rti_rmx, sizeof(rti_rmx));
220 /* Set expire time as absolute. */
221 rti_rmx.rmx_expire = lifetime_sec + time_second;
222 info.rti_mflags |= RTV_EXPIRE;
223 info.rti_rmx = &rti_rmx;
225 error = rib_action(fibnum, RTM_ADD, &info, &rc);
228 /* TODO: add per-fib redirect stats. */
232 RTSTAT_INC(rts_dynamic);
234 /* Send notification of a route addition to userland. */
235 bzero(&info, sizeof(info));
236 info.rti_info[RTAX_DST] = dst;
237 info.rti_info[RTAX_GATEWAY] = gateway;
238 info.rti_info[RTAX_AUTHOR] = author;
239 rt_missmsg_fib(RTM_REDIRECT, &info, flags | RTF_UP, error, fibnum);
245 * Routing table ioctl interface.
248 rtioctl_fib(u_long req, caddr_t data, u_int fibnum)
252 * If more ioctl commands are added here, make sure the proper
253 * super-user checks are being performed because it is possible for
254 * prison-root to make it this far if raw sockets have been enabled
258 /* Multicast goop, grrr... */
259 return mrt_ioctl ? mrt_ioctl(req, data, fibnum) : EOPNOTSUPP;
266 ifa_ifwithroute(int flags, const struct sockaddr *dst,
267 const struct sockaddr *gateway, u_int fibnum)
272 if ((flags & RTF_GATEWAY) == 0) {
274 * If we are adding a route to an interface,
275 * and the interface is a pt to pt link
276 * we should search for the destination
277 * as our clue to the interface. Otherwise
278 * we can use the local address.
281 if (flags & RTF_HOST)
282 ifa = ifa_ifwithdstaddr(dst, fibnum);
284 ifa = ifa_ifwithaddr(gateway);
287 * If we are adding a route to a remote net
288 * or host, the gateway may still be on the
289 * other end of a pt to pt link.
291 ifa = ifa_ifwithdstaddr(gateway, fibnum);
294 ifa = ifa_ifwithnet(gateway, 0, fibnum);
296 struct nhop_object *nh;
298 nh = rib_lookup(fibnum, gateway, NHR_NONE, 0);
301 * dismiss a gateway that is reachable only
302 * through the default router
304 if ((nh == NULL) || (nh->nh_flags & NHF_DEFAULT))
308 if (ifa->ifa_addr->sa_family != dst->sa_family) {
309 struct ifaddr *oifa = ifa;
310 ifa = ifaof_ifpforaddr(dst, ifa->ifa_ifp);
319 * Copy most of @rt data into @info.
321 * If @flags contains NHR_COPY, copies dst,netmask and gw to the
322 * pointers specified by @info structure. Assume such pointers
323 * are zeroed sockaddr-like structures with sa_len field initialized
324 * to reflect size of the provided buffer. if no NHR_COPY is specified,
325 * point dst,netmask and gw @info fields to appropriate @rt values.
327 * if @flags contains NHR_REF, do refcouting on rt_ifp and rt_ifa.
329 * Returns 0 on success.
332 rt_exportinfo(struct rtentry *rt, struct nhop_object *nh,
333 struct rt_addrinfo *info, int flags)
335 struct rt_metrics *rmx;
336 struct sockaddr *src, *dst;
339 if (flags & NHR_COPY) {
340 /* Copy destination if dst is non-zero */
342 dst = info->rti_info[RTAX_DST];
343 sa_len = src->sa_len;
345 if (src->sa_len > dst->sa_len)
347 memcpy(dst, src, src->sa_len);
348 info->rti_addrs |= RTA_DST;
351 /* Copy mask if set && dst is non-zero */
353 dst = info->rti_info[RTAX_NETMASK];
354 if (src != NULL && dst != NULL) {
356 * Radix stores different value in sa_len,
357 * assume rt_mask() to have the same length
360 if (sa_len > dst->sa_len)
362 memcpy(dst, src, src->sa_len);
363 info->rti_addrs |= RTA_NETMASK;
366 /* Copy gateway is set && dst is non-zero */
368 dst = info->rti_info[RTAX_GATEWAY];
369 if ((nhop_get_rtflags(nh) & RTF_GATEWAY) &&
370 src != NULL && dst != NULL) {
371 if (src->sa_len > dst->sa_len)
373 memcpy(dst, src, src->sa_len);
374 info->rti_addrs |= RTA_GATEWAY;
377 info->rti_info[RTAX_DST] = rt_key(rt);
378 info->rti_addrs |= RTA_DST;
379 if (rt_mask(rt) != NULL) {
380 info->rti_info[RTAX_NETMASK] = rt_mask(rt);
381 info->rti_addrs |= RTA_NETMASK;
383 if (nhop_get_rtflags(nh) & RTF_GATEWAY) {
384 info->rti_info[RTAX_GATEWAY] = &nh->gw_sa;
385 info->rti_addrs |= RTA_GATEWAY;
391 info->rti_mflags |= RTV_MTU;
392 rmx->rmx_mtu = nh->nh_mtu;
395 info->rti_flags = rt->rte_flags | nhop_get_rtflags(nh);
396 info->rti_ifp = nh->nh_ifp;
397 info->rti_ifa = nh->nh_ifa;
398 if (flags & NHR_REF) {
399 if_ref(info->rti_ifp);
400 ifa_ref(info->rti_ifa);
407 * Lookups up route entry for @dst in RIB database for fib @fibnum.
408 * Exports entry data to @info using rt_exportinfo().
410 * If @flags contains NHR_REF, refcouting is performed on rt_ifp and rt_ifa.
411 * All references can be released later by calling rib_free_info().
413 * Returns 0 on success.
414 * Returns ENOENT for lookup failure, ENOMEM for export failure.
417 rib_lookup_info(uint32_t fibnum, const struct sockaddr *dst, uint32_t flags,
418 uint32_t flowid, struct rt_addrinfo *info)
422 struct radix_node *rn;
424 struct nhop_object *nh;
427 KASSERT((fibnum < rt_numfibs), ("rib_lookup_rte: bad fibnum"));
428 rh = rt_tables_get_rnh(fibnum, dst->sa_family);
433 rn = rh->rnh_matchaddr(__DECONST(void *, dst), &rh->head);
434 if (rn != NULL && ((rn->rn_flags & RNF_ROOT) == 0)) {
436 nh = nhop_select(rt->rt_nhop, flowid);
437 /* Ensure route & ifp is UP */
438 if (RT_LINK_IS_UP(nh->nh_ifp)) {
439 flags = (flags & NHR_REF) | NHR_COPY;
440 error = rt_exportinfo(rt, nh, info, flags);
452 * Releases all references acquired by rib_lookup_info() when
453 * called with NHR_REF flags.
456 rib_free_info(struct rt_addrinfo *info)
459 ifa_free(info->rti_ifa);
460 if_rele(info->rti_ifp);
464 * Delete Routes for a Network Interface
466 * Called for each routing entry via the rnh->rnh_walktree() call above
467 * to delete all route entries referencing a detaching network interface.
470 * rt pointer to rtentry
472 * arg argument passed to rnh->rnh_walktree() - detaching interface
476 * errno failed - reason indicated
479 rt_ifdelroute(const struct rtentry *rt, const struct nhop_object *nh, void *arg)
481 struct ifnet *ifp = arg;
483 if (nh->nh_ifp != ifp)
487 * Protect (sorta) against walktree recursion problems
490 if ((rt->rte_flags & RTF_UP) == 0)
497 rt_flushifroutes(struct ifnet *ifp)
500 rib_foreach_table_walk_del(AF_UNSPEC, rt_ifdelroute, ifp);
504 * Look up rt_addrinfo for a specific fib.
506 * Assume basic consistency checks are executed by callers:
507 * RTAX_DST exists, if RTF_GATEWAY is set, RTAX_GATEWAY exists as well.
510 rt_getifa_fib(struct rt_addrinfo *info, u_int fibnum)
512 const struct sockaddr *dst, *gateway, *ifpaddr, *ifaaddr;
515 dst = info->rti_info[RTAX_DST];
516 gateway = info->rti_info[RTAX_GATEWAY];
517 ifpaddr = info->rti_info[RTAX_IFP];
518 ifaaddr = info->rti_info[RTAX_IFA];
519 flags = info->rti_flags;
522 * ifp may be specified by sockaddr_dl
523 * when protocol address is ambiguous.
527 /* If we have interface specified by the ifindex in the address, use it */
528 if (info->rti_ifp == NULL && ifpaddr != NULL &&
529 ifpaddr->sa_family == AF_LINK) {
530 const struct sockaddr_dl *sdl = (const struct sockaddr_dl *)ifpaddr;
531 if (sdl->sdl_index != 0)
532 info->rti_ifp = ifnet_byindex(sdl->sdl_index);
535 * If we have source address specified, try to find it
536 * TODO: avoid enumerating all ifas on all interfaces.
538 if (info->rti_ifa == NULL && ifaaddr != NULL)
539 info->rti_ifa = ifa_ifwithaddr(ifaaddr);
540 if (info->rti_ifa == NULL) {
541 const struct sockaddr *sa;
544 * Most common use case for the userland-supplied routes.
546 * Choose sockaddr to select ifa.
547 * -- if ifp is set --
548 * Order of preference:
551 * Note: for interface routes link-level gateway address
552 * is specified to indicate the interface index without
553 * specifying RTF_GATEWAY. In this case, ignore gateway
554 * Note: gateway AF may be different from dst AF. In this case,
556 * 3) final destination.
557 * 4) if all of these fails, try to get at least link-level ifa.
559 * try to lookup gateway or dst in the routing table to get ifa
561 if (info->rti_info[RTAX_IFA] != NULL)
562 sa = info->rti_info[RTAX_IFA];
563 else if ((info->rti_flags & RTF_GATEWAY) != 0 &&
564 gateway->sa_family == dst->sa_family)
568 if (info->rti_ifp != NULL) {
569 info->rti_ifa = ifaof_ifpforaddr(sa, info->rti_ifp);
571 if (info->rti_ifa == NULL && gateway != NULL)
572 info->rti_ifa = ifaof_ifpforaddr(gateway, info->rti_ifp);
573 } else if (dst != NULL && gateway != NULL)
574 info->rti_ifa = ifa_ifwithroute(flags, dst, gateway,
577 info->rti_ifa = ifa_ifwithroute(flags, sa, sa,
580 if (info->rti_ifa != NULL) {
581 if (info->rti_ifp == NULL)
582 info->rti_ifp = info->rti_ifa->ifa_ifp;
589 rt_updatemtu(struct ifnet *ifp)
591 struct rib_head *rnh;
596 * Try to update rt_mtu for all routes using this interface
597 * Unfortunately the only way to do this is to traverse all
598 * routing tables in all fibs/domains.
600 for (i = 1; i <= AF_MAX; i++) {
601 mtu = if_getmtu_family(ifp, i);
602 for (j = 0; j < rt_numfibs; j++) {
603 rnh = rt_tables_get_rnh(j, i);
606 nhops_update_ifmtu(rnh, ifp, mtu);
612 int p_sockaddr(char *buf, int buflen, struct sockaddr *s);
613 int rt_print(char *buf, int buflen, struct rtentry *rt);
616 p_sockaddr(char *buf, int buflen, struct sockaddr *s)
620 switch (s->sa_family) {
622 paddr = &((struct sockaddr_in *)s)->sin_addr;
625 paddr = &((struct sockaddr_in6 *)s)->sin6_addr;
632 if (inet_ntop(s->sa_family, paddr, buf, buflen) == NULL)
635 return (strlen(buf));
639 rt_print(char *buf, int buflen, struct rtentry *rt)
641 struct sockaddr *addr, *mask;
647 i = p_sockaddr(buf, buflen, addr);
648 if (!(rt->rt_flags & RTF_HOST)) {
650 i += p_sockaddr(buf + i, buflen - i, mask);
653 if (rt->rt_flags & RTF_GATEWAY) {
655 i += p_sockaddr(buf + i, buflen - i, &rt->rt_nhop->gw_sa);
663 rt_maskedcopy(struct sockaddr *src, struct sockaddr *dst, struct sockaddr *netmask)
665 u_char *cp1 = (u_char *)src;
666 u_char *cp2 = (u_char *)dst;
667 u_char *cp3 = (u_char *)netmask;
668 u_char *cplim = cp2 + *cp3;
669 u_char *cplim2 = cp2 + *cp1;
671 *cp2++ = *cp1++; *cp2++ = *cp1++; /* copies sa_len & sa_family */
676 *cp2++ = *cp1++ & *cp3++;
678 bzero((caddr_t)cp2, (unsigned)(cplim2 - cp2));
682 * Announce interface address arrival/withdraw
683 * Returns 0 on success.
686 rt_addrmsg(int cmd, struct ifaddr *ifa, int fibnum)
689 KASSERT(cmd == RTM_ADD || cmd == RTM_DELETE,
690 ("unexpected cmd %d", cmd));
691 KASSERT((fibnum >= 0 && fibnum < rt_numfibs),
692 ("%s: fib out of range 0 <=%d<%d", __func__, fibnum, rt_numfibs));
694 EVENTHANDLER_DIRECT_INVOKE(rt_addrmsg, ifa, cmd);
696 if (V_rt_add_addr_allfibs)
697 fibnum = RT_ALL_FIBS;
698 return (rtsock_addrmsg(cmd, ifa, fibnum));
702 * Announce kernel-originated route addition/removal to rtsock based on @rt data.
705 * @nh: nhop object to announce
706 * @fibnum: fib id or RT_ALL_FIBS
708 * Returns 0 on success.
711 rt_routemsg(int cmd, struct rtentry *rt, struct nhop_object *nh,
715 KASSERT(cmd == RTM_ADD || cmd == RTM_DELETE,
716 ("unexpected cmd %d", cmd));
718 KASSERT(fibnum == RT_ALL_FIBS || (fibnum >= 0 && fibnum < rt_numfibs),
719 ("%s: fib out of range 0 <=%d<%d", __func__, fibnum, rt_numfibs));
721 KASSERT(rt_key(rt) != NULL, (":%s: rt_key must be supplied", __func__));
723 return (rtsock_routemsg(cmd, rt, nh, fibnum));
727 * Announce kernel-originated route addition/removal to rtsock based on @rt data.
729 * @info: addrinfo structure with valid data.
730 * @fibnum: fib id or RT_ALL_FIBS
732 * Returns 0 on success.
735 rt_routemsg_info(int cmd, struct rt_addrinfo *info, int fibnum)
738 KASSERT(cmd == RTM_ADD || cmd == RTM_DELETE || cmd == RTM_CHANGE,
739 ("unexpected cmd %d", cmd));
741 KASSERT(fibnum == RT_ALL_FIBS || (fibnum >= 0 && fibnum < rt_numfibs),
742 ("%s: fib out of range 0 <=%d<%d", __func__, fibnum, rt_numfibs));
744 KASSERT(info->rti_info[RTAX_DST] != NULL, (":%s: RTAX_DST must be supplied", __func__));
746 return (rtsock_routemsg_info(cmd, info, fibnum));