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 rt_addrinfo *info,
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)
154 tmproutes_destroy(rh);
156 rn_walktree(&rh->rmhead.head, rt_freeentry, &rh->rmhead.head);
158 nhops_destroy_rib(rh);
160 rib_destroy_subscriptions(rh);
162 /* Assume table is already empty */
163 RIB_LOCK_DESTROY(rh);
168 * Adds a temporal redirect entry to the routing table.
169 * @fibnum: fib number
170 * @dst: destination to install redirect to
171 * @gateway: gateway to go via
172 * @author: sockaddr of originating router, can be NULL
173 * @ifp: interface to use for the redirected route
174 * @flags: set of flags to add. Allowed: RTF_GATEWAY
175 * @lifetime_sec: time in seconds to expire this redirect.
177 * Retuns 0 on success, errno otherwise.
180 rib_add_redirect(u_int fibnum, struct sockaddr *dst, struct sockaddr *gateway,
181 struct sockaddr *author, struct ifnet *ifp, int flags, int lifetime_sec)
183 struct rib_cmd_info rc;
185 struct rt_addrinfo info;
186 struct rt_metrics rti_rmx;
191 if (rt_tables_get_rnh(fibnum, dst->sa_family) == NULL)
192 return (EAFNOSUPPORT);
194 /* Verify the allowed flag mask. */
195 KASSERT(((flags & ~(RTF_GATEWAY)) == 0),
196 ("invalid redirect flags: %x", flags));
197 flags |= RTF_HOST | RTF_DYNAMIC;
199 /* Get the best ifa for the given interface and gateway. */
200 if ((ifa = ifaof_ifpforaddr(gateway, ifp)) == NULL)
201 return (ENETUNREACH);
204 bzero(&info, sizeof(info));
205 info.rti_info[RTAX_DST] = dst;
206 info.rti_info[RTAX_GATEWAY] = gateway;
209 info.rti_flags = flags;
211 /* Setup route metrics to define expire time. */
212 bzero(&rti_rmx, sizeof(rti_rmx));
213 /* Set expire time as absolute. */
214 rti_rmx.rmx_expire = lifetime_sec + time_second;
215 info.rti_mflags |= RTV_EXPIRE;
216 info.rti_rmx = &rti_rmx;
218 error = rib_action(fibnum, RTM_ADD, &info, &rc);
222 /* TODO: add per-fib redirect stats. */
226 RTSTAT_INC(rts_dynamic);
228 /* Send notification of a route addition to userland. */
229 bzero(&info, sizeof(info));
230 info.rti_info[RTAX_DST] = dst;
231 info.rti_info[RTAX_GATEWAY] = gateway;
232 info.rti_info[RTAX_AUTHOR] = author;
233 rt_missmsg_fib(RTM_REDIRECT, &info, flags | RTF_UP, error, fibnum);
239 * Routing table ioctl interface.
242 rtioctl_fib(u_long req, caddr_t data, u_int fibnum)
246 * If more ioctl commands are added here, make sure the proper
247 * super-user checks are being performed because it is possible for
248 * prison-root to make it this far if raw sockets have been enabled
252 /* Multicast goop, grrr... */
253 return mrt_ioctl ? mrt_ioctl(req, data, fibnum) : EOPNOTSUPP;
260 ifa_ifwithroute(int flags, const struct sockaddr *dst,
261 const struct sockaddr *gateway, u_int fibnum)
266 if ((flags & RTF_GATEWAY) == 0) {
268 * If we are adding a route to an interface,
269 * and the interface is a pt to pt link
270 * we should search for the destination
271 * as our clue to the interface. Otherwise
272 * we can use the local address.
275 if (flags & RTF_HOST)
276 ifa = ifa_ifwithdstaddr(dst, fibnum);
278 ifa = ifa_ifwithaddr(gateway);
281 * If we are adding a route to a remote net
282 * or host, the gateway may still be on the
283 * other end of a pt to pt link.
285 ifa = ifa_ifwithdstaddr(gateway, fibnum);
288 ifa = ifa_ifwithnet(gateway, 0, fibnum);
290 struct nhop_object *nh;
292 nh = rib_lookup(fibnum, gateway, NHR_NONE, 0);
295 * dismiss a gateway that is reachable only
296 * through the default router
298 if ((nh == NULL) || (nh->nh_flags & NHF_DEFAULT))
302 if (ifa->ifa_addr->sa_family != dst->sa_family) {
303 struct ifaddr *oifa = ifa;
304 ifa = ifaof_ifpforaddr(dst, ifa->ifa_ifp);
313 * Copy most of @rt data into @info.
315 * If @flags contains NHR_COPY, copies dst,netmask and gw to the
316 * pointers specified by @info structure. Assume such pointers
317 * are zeroed sockaddr-like structures with sa_len field initialized
318 * to reflect size of the provided buffer. if no NHR_COPY is specified,
319 * point dst,netmask and gw @info fields to appropriate @rt values.
321 * if @flags contains NHR_REF, do refcouting on rt_ifp and rt_ifa.
323 * Returns 0 on success.
326 rt_exportinfo(struct rtentry *rt, struct rt_addrinfo *info, int flags)
328 struct rt_metrics *rmx;
329 struct sockaddr *src, *dst;
330 struct nhop_object *nh;
334 if (flags & NHR_COPY) {
335 /* Copy destination if dst is non-zero */
337 dst = info->rti_info[RTAX_DST];
338 sa_len = src->sa_len;
340 if (src->sa_len > dst->sa_len)
342 memcpy(dst, src, src->sa_len);
343 info->rti_addrs |= RTA_DST;
346 /* Copy mask if set && dst is non-zero */
348 dst = info->rti_info[RTAX_NETMASK];
349 if (src != NULL && dst != NULL) {
351 * Radix stores different value in sa_len,
352 * assume rt_mask() to have the same length
355 if (sa_len > dst->sa_len)
357 memcpy(dst, src, src->sa_len);
358 info->rti_addrs |= RTA_NETMASK;
361 /* Copy gateway is set && dst is non-zero */
363 dst = info->rti_info[RTAX_GATEWAY];
364 if ((nhop_get_rtflags(nh) & RTF_GATEWAY) &&
365 src != NULL && dst != NULL) {
366 if (src->sa_len > dst->sa_len)
368 memcpy(dst, src, src->sa_len);
369 info->rti_addrs |= RTA_GATEWAY;
372 info->rti_info[RTAX_DST] = rt_key(rt);
373 info->rti_addrs |= RTA_DST;
374 if (rt_mask(rt) != NULL) {
375 info->rti_info[RTAX_NETMASK] = rt_mask(rt);
376 info->rti_addrs |= RTA_NETMASK;
378 if (nhop_get_rtflags(nh) & RTF_GATEWAY) {
379 info->rti_info[RTAX_GATEWAY] = &nh->gw_sa;
380 info->rti_addrs |= RTA_GATEWAY;
386 info->rti_mflags |= RTV_MTU;
387 rmx->rmx_mtu = nh->nh_mtu;
390 info->rti_flags = rt->rte_flags | nhop_get_rtflags(nh);
391 info->rti_ifp = nh->nh_ifp;
392 info->rti_ifa = nh->nh_ifa;
393 if (flags & NHR_REF) {
394 if_ref(info->rti_ifp);
395 ifa_ref(info->rti_ifa);
402 * Lookups up route entry for @dst in RIB database for fib @fibnum.
403 * Exports entry data to @info using rt_exportinfo().
405 * If @flags contains NHR_REF, refcouting is performed on rt_ifp and rt_ifa.
406 * All references can be released later by calling rib_free_info().
408 * Returns 0 on success.
409 * Returns ENOENT for lookup failure, ENOMEM for export failure.
412 rib_lookup_info(uint32_t fibnum, const struct sockaddr *dst, uint32_t flags,
413 uint32_t flowid, struct rt_addrinfo *info)
417 struct radix_node *rn;
421 KASSERT((fibnum < rt_numfibs), ("rib_lookup_rte: bad fibnum"));
422 rh = rt_tables_get_rnh(fibnum, dst->sa_family);
427 rn = rh->rnh_matchaddr(__DECONST(void *, dst), &rh->head);
428 if (rn != NULL && ((rn->rn_flags & RNF_ROOT) == 0)) {
430 /* Ensure route & ifp is UP */
431 if (RT_LINK_IS_UP(rt->rt_nhop->nh_ifp)) {
432 flags = (flags & NHR_REF) | NHR_COPY;
433 error = rt_exportinfo(rt, info, flags);
445 * Releases all references acquired by rib_lookup_info() when
446 * called with NHR_REF flags.
449 rib_free_info(struct rt_addrinfo *info)
452 ifa_free(info->rti_ifa);
453 if_rele(info->rti_ifp);
457 * Delete Routes for a Network Interface
459 * Called for each routing entry via the rnh->rnh_walktree() call above
460 * to delete all route entries referencing a detaching network interface.
463 * rt pointer to rtentry
465 * arg argument passed to rnh->rnh_walktree() - detaching interface
469 * errno failed - reason indicated
472 rt_ifdelroute(const struct rtentry *rt, const struct nhop_object *nh, void *arg)
474 struct ifnet *ifp = arg;
476 if (nh->nh_ifp != ifp)
480 * Protect (sorta) against walktree recursion problems
483 if ((rt->rte_flags & RTF_UP) == 0)
490 * Delete all remaining routes using this interface
491 * Unfortuneatly the only way to do this is to slog through
492 * the entire routing table looking for routes which point
493 * to this interface...oh well...
496 rt_flushifroutes_af(struct ifnet *ifp, int af)
498 KASSERT((af >= 1 && af <= AF_MAX), ("%s: af %d not >= 1 and <= %d",
499 __func__, af, AF_MAX));
501 rib_foreach_table_walk_del(af, rt_ifdelroute, ifp);
505 rt_flushifroutes(struct ifnet *ifp)
508 rib_foreach_table_walk_del(AF_UNSPEC, rt_ifdelroute, ifp);
512 * Look up rt_addrinfo for a specific fib. Note that if rti_ifa is defined,
513 * it will be referenced so the caller must free it.
515 * Assume basic consistency checks are executed by callers:
516 * RTAX_DST exists, if RTF_GATEWAY is set, RTAX_GATEWAY exists as well.
519 rt_getifa_fib(struct rt_addrinfo *info, u_int fibnum)
521 const struct sockaddr *dst, *gateway, *ifpaddr, *ifaaddr;
522 struct epoch_tracker et;
523 int needref, error, flags;
525 dst = info->rti_info[RTAX_DST];
526 gateway = info->rti_info[RTAX_GATEWAY];
527 ifpaddr = info->rti_info[RTAX_IFP];
528 ifaaddr = info->rti_info[RTAX_IFA];
529 flags = info->rti_flags;
532 * ifp may be specified by sockaddr_dl
533 * when protocol address is ambiguous.
536 needref = (info->rti_ifa == NULL);
539 /* If we have interface specified by the ifindex in the address, use it */
540 if (info->rti_ifp == NULL && ifpaddr != NULL &&
541 ifpaddr->sa_family == AF_LINK) {
542 const struct sockaddr_dl *sdl = (const struct sockaddr_dl *)ifpaddr;
543 if (sdl->sdl_index != 0)
544 info->rti_ifp = ifnet_byindex(sdl->sdl_index);
547 * If we have source address specified, try to find it
548 * TODO: avoid enumerating all ifas on all interfaces.
550 if (info->rti_ifa == NULL && ifaaddr != NULL)
551 info->rti_ifa = ifa_ifwithaddr(ifaaddr);
552 if (info->rti_ifa == NULL) {
553 const struct sockaddr *sa;
556 * Most common use case for the userland-supplied routes.
558 * Choose sockaddr to select ifa.
559 * -- if ifp is set --
560 * Order of preference:
563 * Note: for interface routes link-level gateway address
564 * is specified to indicate the interface index without
565 * specifying RTF_GATEWAY. In this case, ignore gateway
566 * Note: gateway AF may be different from dst AF. In this case,
568 * 3) final destination.
569 * 4) if all of these fails, try to get at least link-level ifa.
571 * try to lookup gateway or dst in the routing table to get ifa
573 if (info->rti_info[RTAX_IFA] != NULL)
574 sa = info->rti_info[RTAX_IFA];
575 else if ((info->rti_flags & RTF_GATEWAY) != 0 &&
576 gateway->sa_family == dst->sa_family)
580 if (info->rti_ifp != NULL) {
581 info->rti_ifa = ifaof_ifpforaddr(sa, info->rti_ifp);
583 if (info->rti_ifa == NULL && gateway != NULL)
584 info->rti_ifa = ifaof_ifpforaddr(gateway, info->rti_ifp);
585 } else if (dst != NULL && gateway != NULL)
586 info->rti_ifa = ifa_ifwithroute(flags, dst, gateway,
589 info->rti_ifa = ifa_ifwithroute(flags, sa, sa,
592 if (needref && info->rti_ifa != NULL) {
593 if (info->rti_ifp == NULL)
594 info->rti_ifp = info->rti_ifa->ifa_ifp;
595 ifa_ref(info->rti_ifa);
603 rt_updatemtu(struct ifnet *ifp)
605 struct rib_head *rnh;
610 * Try to update rt_mtu for all routes using this interface
611 * Unfortunately the only way to do this is to traverse all
612 * routing tables in all fibs/domains.
614 for (i = 1; i <= AF_MAX; i++) {
615 mtu = if_getmtu_family(ifp, i);
616 for (j = 0; j < rt_numfibs; j++) {
617 rnh = rt_tables_get_rnh(j, i);
620 nhops_update_ifmtu(rnh, ifp, mtu);
626 int p_sockaddr(char *buf, int buflen, struct sockaddr *s);
627 int rt_print(char *buf, int buflen, struct rtentry *rt);
630 p_sockaddr(char *buf, int buflen, struct sockaddr *s)
634 switch (s->sa_family) {
636 paddr = &((struct sockaddr_in *)s)->sin_addr;
639 paddr = &((struct sockaddr_in6 *)s)->sin6_addr;
646 if (inet_ntop(s->sa_family, paddr, buf, buflen) == NULL)
649 return (strlen(buf));
653 rt_print(char *buf, int buflen, struct rtentry *rt)
655 struct sockaddr *addr, *mask;
661 i = p_sockaddr(buf, buflen, addr);
662 if (!(rt->rt_flags & RTF_HOST)) {
664 i += p_sockaddr(buf + i, buflen - i, mask);
667 if (rt->rt_flags & RTF_GATEWAY) {
669 i += p_sockaddr(buf + i, buflen - i, &rt->rt_nhop->gw_sa);
677 rt_maskedcopy(struct sockaddr *src, struct sockaddr *dst, struct sockaddr *netmask)
679 u_char *cp1 = (u_char *)src;
680 u_char *cp2 = (u_char *)dst;
681 u_char *cp3 = (u_char *)netmask;
682 u_char *cplim = cp2 + *cp3;
683 u_char *cplim2 = cp2 + *cp1;
685 *cp2++ = *cp1++; *cp2++ = *cp1++; /* copies sa_len & sa_family */
690 *cp2++ = *cp1++ & *cp3++;
692 bzero((caddr_t)cp2, (unsigned)(cplim2 - cp2));
696 * Announce interface address arrival/withdraw
697 * Returns 0 on success.
700 rt_addrmsg(int cmd, struct ifaddr *ifa, int fibnum)
703 KASSERT(cmd == RTM_ADD || cmd == RTM_DELETE,
704 ("unexpected cmd %d", cmd));
705 KASSERT(fibnum == RT_ALL_FIBS || (fibnum >= 0 && fibnum < rt_numfibs),
706 ("%s: fib out of range 0 <=%d<%d", __func__, fibnum, rt_numfibs));
708 EVENTHANDLER_DIRECT_INVOKE(rt_addrmsg, ifa, cmd);
709 return (rtsock_addrmsg(cmd, ifa, fibnum));
713 * Announce kernel-originated route addition/removal to rtsock based on @rt data.
716 * @ifp: target route interface
717 * @fibnum: fib id or RT_ALL_FIBS
719 * Returns 0 on success.
722 rt_routemsg(int cmd, struct rtentry *rt, struct ifnet *ifp, int rti_addrs,
726 KASSERT(cmd == RTM_ADD || cmd == RTM_DELETE,
727 ("unexpected cmd %d", cmd));
729 KASSERT(fibnum == RT_ALL_FIBS || (fibnum >= 0 && fibnum < rt_numfibs),
730 ("%s: fib out of range 0 <=%d<%d", __func__, fibnum, rt_numfibs));
732 KASSERT(rt_key(rt) != NULL, (":%s: rt_key must be supplied", __func__));
734 return (rtsock_routemsg(cmd, rt, ifp, 0, fibnum));
738 * Announce kernel-originated route addition/removal to rtsock based on @rt data.
740 * @info: addrinfo structure with valid data.
741 * @fibnum: fib id or RT_ALL_FIBS
743 * Returns 0 on success.
746 rt_routemsg_info(int cmd, struct rt_addrinfo *info, int fibnum)
749 KASSERT(cmd == RTM_ADD || cmd == RTM_DELETE || cmd == RTM_CHANGE,
750 ("unexpected cmd %d", cmd));
752 KASSERT(fibnum == RT_ALL_FIBS || (fibnum >= 0 && fibnum < rt_numfibs),
753 ("%s: fib out of range 0 <=%d<%d", __func__, fibnum, rt_numfibs));
755 KASSERT(info->rti_info[RTAX_DST] != NULL, (":%s: RTAX_DST must be supplied", __func__));
757 return (rtsock_routemsg_info(cmd, info, fibnum));
761 * This is called to generate messages from the routing socket
762 * indicating a network interface has had addresses associated with it.
765 rt_newaddrmsg_fib(int cmd, struct ifaddr *ifa, struct rtentry *rt, int fibnum)
768 KASSERT(cmd == RTM_ADD || cmd == RTM_DELETE,
769 ("unexpected cmd %u", cmd));
770 KASSERT(fibnum == RT_ALL_FIBS || (fibnum >= 0 && fibnum < rt_numfibs),
771 ("%s: fib out of range 0 <=%d<%d", __func__, fibnum, rt_numfibs));
773 if (cmd == RTM_ADD) {
774 rt_addrmsg(cmd, ifa, fibnum);
776 rt_routemsg(cmd, rt, ifa->ifa_ifp, 0, fibnum);
779 rt_routemsg(cmd, rt, ifa->ifa_ifp, 0, fibnum);
780 rt_addrmsg(cmd, ifa, fibnum);