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/devctl.h>
54 #include <sys/domain.h>
55 #include <sys/eventhandler.h>
56 #include <sys/kernel.h>
58 #include <sys/rmlock.h>
61 #include <net/if_var.h>
62 #include <net/if_dl.h>
63 #include <net/route.h>
64 #include <net/route/route_ctl.h>
65 #include <net/route/route_var.h>
66 #include <net/route/nhop.h>
69 #include <netinet/in.h>
70 #include <netinet/ip_mroute.h>
71 #include <netinet6/in6_var.h>
73 VNET_PCPUSTAT_DEFINE(struct rtstat, rtstat);
75 VNET_PCPUSTAT_SYSINIT(rtstat);
77 VNET_PCPUSTAT_SYSUNINIT(rtstat);
80 EVENTHANDLER_LIST_DEFINE(rt_addrmsg);
82 static int rt_ifdelroute(const struct rtentry *rt, const struct nhop_object *,
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 route_nhop_data rnd = { .rnd_weight = RT_DEFAULT_WEIGHT };
192 struct rib_cmd_info rc;
198 if (rt_tables_get_rnh(fibnum, dst->sa_family) == NULL)
199 return (EAFNOSUPPORT);
201 /* Verify the allowed flag mask. */
202 KASSERT(((flags & ~(RTF_GATEWAY)) == 0),
203 ("invalid redirect flags: %x", flags));
204 flags |= RTF_HOST | RTF_DYNAMIC;
206 /* Get the best ifa for the given interface and gateway. */
207 if ((ifa = ifaof_ifpforaddr(gateway, ifp)) == NULL)
208 return (ENETUNREACH);
210 struct nhop_object *nh = nhop_alloc(fibnum, dst->sa_family);
214 nhop_set_gw(nh, gateway, flags & RTF_GATEWAY);
215 nhop_set_transmit_ifp(nh, ifp);
216 nhop_set_src(nh, ifa);
217 nhop_set_pxtype_flag(nh, NHF_HOST);
218 nhop_set_expire(nh, lifetime_sec + time_uptime);
219 nhop_set_redirect(nh, true);
220 nhop_set_origin(nh, NH_ORIGIN_REDIRECT);
221 rnd.rnd_nhop = nhop_get_nhop(nh, &error);
223 error = rib_add_route_px(fibnum, dst, -1,
224 &rnd, RTM_F_CREATE, &rc);
228 /* TODO: add per-fib redirect stats. */
232 RTSTAT_INC(rts_dynamic);
234 /* Send notification of a route addition to userland. */
235 struct rt_addrinfo info = {
236 .rti_info[RTAX_DST] = dst,
237 .rti_info[RTAX_GATEWAY] = gateway,
238 .rti_info[RTAX_AUTHOR] = author,
240 rt_missmsg_fib(RTM_REDIRECT, &info, flags | RTF_UP, error, fibnum);
246 * Routing table ioctl interface.
249 rtioctl_fib(u_long req, caddr_t data, u_int fibnum)
253 * If more ioctl commands are added here, make sure the proper
254 * super-user checks are being performed because it is possible for
255 * prison-root to make it this far if raw sockets have been enabled
259 /* Multicast goop, grrr... */
260 return mrt_ioctl ? mrt_ioctl(req, data, fibnum) : EOPNOTSUPP;
267 ifa_ifwithroute(int flags, const struct sockaddr *dst,
268 const struct sockaddr *gateway, u_int fibnum)
273 if ((flags & RTF_GATEWAY) == 0) {
275 * If we are adding a route to an interface,
276 * and the interface is a pt to pt link
277 * we should search for the destination
278 * as our clue to the interface. Otherwise
279 * we can use the local address.
282 if (flags & RTF_HOST)
283 ifa = ifa_ifwithdstaddr(dst, fibnum);
285 ifa = ifa_ifwithaddr(gateway);
288 * If we are adding a route to a remote net
289 * or host, the gateway may still be on the
290 * other end of a pt to pt link.
292 ifa = ifa_ifwithdstaddr(gateway, fibnum);
295 ifa = ifa_ifwithnet(gateway, 0, fibnum);
297 struct nhop_object *nh;
299 nh = rib_lookup(fibnum, gateway, NHR_NONE, 0);
302 * dismiss a gateway that is reachable only
303 * through the default router
305 if ((nh == NULL) || (nh->nh_flags & NHF_DEFAULT))
309 if (ifa->ifa_addr->sa_family != dst->sa_family) {
310 struct ifaddr *oifa = ifa;
311 ifa = ifaof_ifpforaddr(dst, ifa->ifa_ifp);
320 * Delete Routes for a Network Interface
322 * Called for each routing entry via the rnh->rnh_walktree() call above
323 * to delete all route entries referencing a detaching network interface.
326 * rt pointer to rtentry
328 * arg argument passed to rnh->rnh_walktree() - detaching interface
332 * errno failed - reason indicated
335 rt_ifdelroute(const struct rtentry *rt, const struct nhop_object *nh, void *arg)
337 struct ifnet *ifp = arg;
339 if (nh->nh_ifp != ifp)
343 * Protect (sorta) against walktree recursion problems
346 if ((rt->rte_flags & RTF_UP) == 0)
353 rt_flushifroutes(struct ifnet *ifp)
356 rib_foreach_table_walk_del(AF_UNSPEC, rt_ifdelroute, ifp);
360 * Tries to extract interface from RTAX_IFP passed in rt_addrinfo.
361 * Interface can be specified ether as interface index (sdl_index) or
362 * the interface name (sdl_data).
364 * Returns found ifp or NULL
366 static struct ifnet *
367 info_get_ifp(struct rt_addrinfo *info)
369 const struct sockaddr_dl *sdl;
371 sdl = (const struct sockaddr_dl *)info->rti_info[RTAX_IFP];
372 if (sdl->sdl_family != AF_LINK)
375 if (sdl->sdl_index != 0)
376 return (ifnet_byindex(sdl->sdl_index));
377 if (sdl->sdl_nlen > 0) {
378 char if_name[IF_NAMESIZE];
379 if (sdl->sdl_nlen + offsetof(struct sockaddr_dl, sdl_data) > sdl->sdl_len)
381 if (sdl->sdl_nlen >= IF_NAMESIZE)
383 bzero(if_name, sizeof(if_name));
384 memcpy(if_name, sdl->sdl_data, sdl->sdl_nlen);
385 return (ifunit(if_name));
392 * Calculates proper ifa/ifp for the cases when gateway AF is different
395 * Returns 0 on success.
397 __noinline static int
398 rt_getifa_family(struct rt_addrinfo *info, uint32_t fibnum)
400 if (info->rti_ifp == NULL) {
401 struct ifaddr *ifa = NULL;
403 * No transmit interface specified. Guess it by checking gw sa.
405 const struct sockaddr *gw = info->rti_info[RTAX_GATEWAY];
406 ifa = ifa_ifwithroute(RTF_GATEWAY, gw, gw, fibnum);
408 return (ENETUNREACH);
409 info->rti_ifp = ifa->ifa_ifp;
412 /* Prefer address from outgoing interface */
413 info->rti_ifa = ifaof_ifpforaddr(info->rti_info[RTAX_DST], info->rti_ifp);
415 if (info->rti_ifa == NULL) {
416 /* Use first found IPv4 address */
417 bool loopback_ok = info->rti_ifp->if_flags & IFF_LOOPBACK;
418 info->rti_ifa = (struct ifaddr *)in_findlocal(fibnum, loopback_ok);
421 if (info->rti_ifa == NULL)
422 return (ENETUNREACH);
427 * Fills in rti_ifp and rti_ifa for the provided fib.
429 * Assume basic consistency checks are executed by callers:
430 * RTAX_DST exists, if RTF_GATEWAY is set, RTAX_GATEWAY exists as well.
433 rt_getifa_fib(struct rt_addrinfo *info, u_int fibnum)
435 const struct sockaddr *dst, *gateway, *ifaaddr;
438 dst = info->rti_info[RTAX_DST];
439 gateway = info->rti_info[RTAX_GATEWAY];
440 ifaaddr = info->rti_info[RTAX_IFA];
441 flags = info->rti_flags;
444 * ifp may be specified by sockaddr_dl
445 * when protocol address is ambiguous.
449 /* If we have interface specified by RTAX_IFP address, try to use it */
450 if ((info->rti_ifp == NULL) && (info->rti_info[RTAX_IFP] != NULL))
451 info->rti_ifp = info_get_ifp(info);
453 * If we have source address specified, try to find it
454 * TODO: avoid enumerating all ifas on all interfaces.
456 if (info->rti_ifa == NULL && ifaaddr != NULL)
457 info->rti_ifa = ifa_ifwithaddr(ifaaddr);
458 if ((info->rti_ifa == NULL) && ((info->rti_flags & RTF_GATEWAY) != 0) &&
459 (gateway->sa_family != dst->sa_family))
460 return (rt_getifa_family(info, fibnum));
461 if (info->rti_ifa == NULL) {
462 const struct sockaddr *sa;
465 * Most common use case for the userland-supplied routes.
467 * Choose sockaddr to select ifa.
468 * -- if ifp is set --
469 * Order of preference:
472 * Note: for interface routes link-level gateway address
473 * is specified to indicate the interface index without
474 * specifying RTF_GATEWAY. In this case, ignore gateway
475 * Note: gateway AF may be different from dst AF. In this case,
477 * 3) final destination.
478 * 4) if all of these fails, try to get at least link-level ifa.
480 * try to lookup gateway or dst in the routing table to get ifa
482 if (info->rti_info[RTAX_IFA] != NULL)
483 sa = info->rti_info[RTAX_IFA];
484 else if ((info->rti_flags & RTF_GATEWAY) != 0 &&
485 gateway->sa_family == dst->sa_family)
489 if (info->rti_ifp != NULL) {
490 info->rti_ifa = ifaof_ifpforaddr(sa, info->rti_ifp);
492 if (info->rti_ifa == NULL && gateway != NULL)
493 info->rti_ifa = ifaof_ifpforaddr(gateway, info->rti_ifp);
494 } else if (dst != NULL && gateway != NULL)
495 info->rti_ifa = ifa_ifwithroute(flags, dst, gateway,
498 info->rti_ifa = ifa_ifwithroute(flags, sa, sa,
501 if (info->rti_ifa != NULL) {
502 if (info->rti_ifp == NULL)
503 info->rti_ifp = info->rti_ifa->ifa_ifp;
510 rt_updatemtu(struct ifnet *ifp)
512 struct rib_head *rnh;
517 * Try to update rt_mtu for all routes using this interface
518 * Unfortunately the only way to do this is to traverse all
519 * routing tables in all fibs/domains.
521 for (i = 1; i <= AF_MAX; i++) {
522 mtu = if_getmtu_family(ifp, i);
523 for (j = 0; j < rt_numfibs; j++) {
524 rnh = rt_tables_get_rnh(j, i);
527 nhops_update_ifmtu(rnh, ifp, mtu);
533 int p_sockaddr(char *buf, int buflen, struct sockaddr *s);
534 int rt_print(char *buf, int buflen, struct rtentry *rt);
537 p_sockaddr(char *buf, int buflen, struct sockaddr *s)
541 switch (s->sa_family) {
543 paddr = &((struct sockaddr_in *)s)->sin_addr;
546 paddr = &((struct sockaddr_in6 *)s)->sin6_addr;
553 if (inet_ntop(s->sa_family, paddr, buf, buflen) == NULL)
556 return (strlen(buf));
560 rt_print(char *buf, int buflen, struct rtentry *rt)
562 struct sockaddr *addr, *mask;
568 i = p_sockaddr(buf, buflen, addr);
569 if (!(rt->rt_flags & RTF_HOST)) {
571 i += p_sockaddr(buf + i, buflen - i, mask);
574 if (rt->rt_flags & RTF_GATEWAY) {
576 i += p_sockaddr(buf + i, buflen - i, &rt->rt_nhop->gw_sa);
584 rt_maskedcopy(const struct sockaddr *src, struct sockaddr *dst,
585 const struct sockaddr *netmask)
587 const u_char *cp1 = (const u_char *)src;
588 u_char *cp2 = (u_char *)dst;
589 const u_char *cp3 = (const u_char *)netmask;
590 u_char *cplim = cp2 + *cp3;
591 u_char *cplim2 = cp2 + *cp1;
593 *cp2++ = *cp1++; *cp2++ = *cp1++; /* copies sa_len & sa_family */
598 *cp2++ = *cp1++ & *cp3++;
600 bzero((caddr_t)cp2, (unsigned)(cplim2 - cp2));
604 * Announce interface address arrival/withdraw
605 * Returns 0 on success.
608 rt_addrmsg(int cmd, struct ifaddr *ifa, int fibnum)
610 #if defined(INET) || defined(INET6)
611 struct sockaddr *sa = ifa->ifa_addr;
612 struct ifnet *ifp = ifa->ifa_ifp;
615 KASSERT(cmd == RTM_ADD || cmd == RTM_DELETE,
616 ("unexpected cmd %d", cmd));
617 KASSERT((fibnum >= 0 && fibnum < rt_numfibs),
618 ("%s: fib out of range 0 <=%d<%d", __func__, fibnum, rt_numfibs));
620 EVENTHANDLER_DIRECT_INVOKE(rt_addrmsg, ifa, cmd);
623 if (sa->sa_family == AF_INET) {
624 char addrstr[INET_ADDRSTRLEN];
625 char strbuf[INET_ADDRSTRLEN + 12];
627 inet_ntoa_r(((struct sockaddr_in *)sa)->sin_addr, addrstr);
628 snprintf(strbuf, sizeof(strbuf), "address=%s", addrstr);
629 devctl_notify("IFNET", ifp->if_xname,
630 (cmd == RTM_ADD) ? "ADDR_ADD" : "ADDR_DEL", strbuf);
634 if (sa->sa_family == AF_INET6) {
635 char addrstr[INET6_ADDRSTRLEN];
636 char strbuf[INET6_ADDRSTRLEN + 12];
638 ip6_sprintf(addrstr, IFA_IN6(ifa));
639 snprintf(strbuf, sizeof(strbuf), "address=%s", addrstr);
640 devctl_notify("IFNET", ifp->if_xname,
641 (cmd == RTM_ADD) ? "ADDR_ADD" : "ADDR_DEL", strbuf);
645 if (V_rt_add_addr_allfibs)
646 fibnum = RT_ALL_FIBS;
647 return (rtsock_addrmsg(cmd, ifa, fibnum));
651 * Announce kernel-originated route addition/removal to rtsock based on @rt data.
654 * @nh: nhop object to announce
655 * @fibnum: fib id or RT_ALL_FIBS
657 * Returns 0 on success.
660 rt_routemsg(int cmd, struct rtentry *rt, struct nhop_object *nh,
664 KASSERT(cmd == RTM_ADD || cmd == RTM_DELETE,
665 ("unexpected cmd %d", cmd));
667 KASSERT(fibnum == RT_ALL_FIBS || (fibnum >= 0 && fibnum < rt_numfibs),
668 ("%s: fib out of range 0 <=%d<%d", __func__, fibnum, rt_numfibs));
670 KASSERT(rt_key(rt) != NULL, (":%s: rt_key must be supplied", __func__));
672 return (rtsock_routemsg(cmd, rt, nh, fibnum));
676 * Announce kernel-originated route addition/removal to rtsock based on @rt data.
678 * @info: addrinfo structure with valid data.
679 * @fibnum: fib id or RT_ALL_FIBS
681 * Returns 0 on success.
684 rt_routemsg_info(int cmd, struct rt_addrinfo *info, int fibnum)
687 KASSERT(cmd == RTM_ADD || cmd == RTM_DELETE || cmd == RTM_CHANGE,
688 ("unexpected cmd %d", cmd));
690 KASSERT(fibnum == RT_ALL_FIBS || (fibnum >= 0 && fibnum < rt_numfibs),
691 ("%s: fib out of range 0 <=%d<%d", __func__, fibnum, rt_numfibs));
693 KASSERT(info->rti_info[RTAX_DST] != NULL, (":%s: RTAX_DST must be supplied", __func__));
695 return (rtsock_routemsg_info(cmd, info, fibnum));
698 /* Netlink-related callbacks needed to glue rtsock, netlink and linuxolator */
700 ignore_route_event(uint32_t fibnum, const struct rib_cmd_info *rc)
703 static struct rtbridge ignore_cb = { .route_f = ignore_route_event };
705 void *linux_netlink_p = NULL; /* Callback pointer for Linux translator functions */
706 struct rtbridge *rtsock_callback_p = &ignore_cb;
707 struct rtbridge *netlink_callback_p = &ignore_cb;