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_private.h>
63 #include <net/if_dl.h>
64 #include <net/route.h>
65 #include <net/route/route_ctl.h>
66 #include <net/route/route_var.h>
67 #include <net/route/nhop.h>
70 #include <netinet/in.h>
71 #include <netinet/ip_mroute.h>
72 #include <netinet6/in6_var.h>
74 VNET_PCPUSTAT_DEFINE(struct rtstat, rtstat);
76 VNET_PCPUSTAT_SYSINIT(rtstat);
78 VNET_PCPUSTAT_SYSUNINIT(rtstat);
81 EVENTHANDLER_LIST_DEFINE(rt_addrmsg);
83 static int rt_ifdelroute(const struct rtentry *rt, const struct nhop_object *,
87 * route initialization must occur before ip6_init2(), which happenas at
96 SYSINIT(route_init, SI_SUB_PROTO_DOMAIN, SI_ORDER_THIRD, route_init, NULL);
99 rt_table_init(int offset, int family, u_int fibnum)
103 rh = malloc(sizeof(struct rib_head), M_RTABLE, M_WAITOK | M_ZERO);
105 /* TODO: These details should be hidded inside radix.c */
106 /* Init masks tree */
107 rn_inithead_internal(&rh->head, rh->rnh_nodes, offset);
108 rn_inithead_internal(&rh->rmhead.head, rh->rmhead.mask_nodes, 0);
109 rh->head.rnh_masks = &rh->rmhead;
111 /* Save metadata associated with this routing table. */
112 rh->rib_family = family;
113 rh->rib_fibnum = fibnum;
115 rh->rib_vnet = curvnet;
125 /* Init subscription system */
126 rib_init_subscriptions(rh);
128 /* Finally, set base callbacks */
129 rh->rnh_addaddr = rn_addroute;
130 rh->rnh_deladdr = rn_delete;
131 rh->rnh_matchaddr = rn_match;
132 rh->rnh_lookup = rn_lookup;
133 rh->rnh_walktree = rn_walktree;
134 rh->rnh_walktree_from = rn_walktree_from;
140 rt_freeentry(struct radix_node *rn, void *arg)
142 struct radix_head * const rnh = arg;
143 struct radix_node *x;
145 x = (struct radix_node *)rn_delete(rn + 2, NULL, rnh);
152 rt_table_destroy(struct rib_head *rh)
156 rh->rib_dying = true;
163 tmproutes_destroy(rh);
165 rn_walktree(&rh->rmhead.head, rt_freeentry, &rh->rmhead.head);
167 nhops_destroy_rib(rh);
169 rib_destroy_subscriptions(rh);
171 /* Assume table is already empty */
172 RIB_LOCK_DESTROY(rh);
177 * Adds a temporal redirect entry to the routing table.
178 * @fibnum: fib number
179 * @dst: destination to install redirect to
180 * @gateway: gateway to go via
181 * @author: sockaddr of originating router, can be NULL
182 * @ifp: interface to use for the redirected route
183 * @flags: set of flags to add. Allowed: RTF_GATEWAY
184 * @lifetime_sec: time in seconds to expire this redirect.
186 * Retuns 0 on success, errno otherwise.
189 rib_add_redirect(u_int fibnum, struct sockaddr *dst, struct sockaddr *gateway,
190 struct sockaddr *author, struct ifnet *ifp, int flags, int lifetime_sec)
192 struct route_nhop_data rnd = { .rnd_weight = RT_DEFAULT_WEIGHT };
193 struct rib_cmd_info rc;
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 struct nhop_object *nh = nhop_alloc(fibnum, dst->sa_family);
215 nhop_set_gw(nh, gateway, flags & RTF_GATEWAY);
216 nhop_set_transmit_ifp(nh, ifp);
217 nhop_set_src(nh, ifa);
218 nhop_set_pxtype_flag(nh, NHF_HOST);
219 nhop_set_expire(nh, lifetime_sec + time_uptime);
220 nhop_set_redirect(nh, true);
221 nhop_set_origin(nh, NH_ORIGIN_REDIRECT);
222 rnd.rnd_nhop = nhop_get_nhop(nh, &error);
224 error = rib_add_route_px(fibnum, dst, -1,
225 &rnd, RTM_F_CREATE, &rc);
229 /* TODO: add per-fib redirect stats. */
233 RTSTAT_INC(rts_dynamic);
235 /* Send notification of a route addition to userland. */
236 struct rt_addrinfo info = {
237 .rti_info[RTAX_DST] = dst,
238 .rti_info[RTAX_GATEWAY] = gateway,
239 .rti_info[RTAX_AUTHOR] = author,
241 rt_missmsg_fib(RTM_REDIRECT, &info, flags | RTF_UP, error, fibnum);
247 * Routing table ioctl interface.
250 rtioctl_fib(u_long req, caddr_t data, u_int fibnum)
254 * If more ioctl commands are added here, make sure the proper
255 * super-user checks are being performed because it is possible for
256 * prison-root to make it this far if raw sockets have been enabled
260 /* Multicast goop, grrr... */
261 return mrt_ioctl ? mrt_ioctl(req, data, fibnum) : EOPNOTSUPP;
268 ifa_ifwithroute(int flags, const struct sockaddr *dst,
269 const struct sockaddr *gateway, u_int fibnum)
274 if ((flags & RTF_GATEWAY) == 0) {
276 * If we are adding a route to an interface,
277 * and the interface is a pt to pt link
278 * we should search for the destination
279 * as our clue to the interface. Otherwise
280 * we can use the local address.
283 if (flags & RTF_HOST)
284 ifa = ifa_ifwithdstaddr(dst, fibnum);
286 ifa = ifa_ifwithaddr(gateway);
289 * If we are adding a route to a remote net
290 * or host, the gateway may still be on the
291 * other end of a pt to pt link.
293 ifa = ifa_ifwithdstaddr(gateway, fibnum);
296 ifa = ifa_ifwithnet(gateway, 0, fibnum);
298 struct nhop_object *nh;
300 nh = rib_lookup(fibnum, gateway, NHR_NONE, 0);
303 * dismiss a gateway that is reachable only
304 * through the default router
306 if ((nh == NULL) || (nh->nh_flags & NHF_DEFAULT))
310 if (ifa->ifa_addr->sa_family != dst->sa_family) {
311 struct ifaddr *oifa = ifa;
312 ifa = ifaof_ifpforaddr(dst, ifa->ifa_ifp);
321 * Delete Routes for a Network Interface
323 * Called for each routing entry via the rnh->rnh_walktree() call above
324 * to delete all route entries referencing a detaching network interface.
327 * rt pointer to rtentry
329 * arg argument passed to rnh->rnh_walktree() - detaching interface
333 * errno failed - reason indicated
336 rt_ifdelroute(const struct rtentry *rt, const struct nhop_object *nh, void *arg)
338 struct ifnet *ifp = arg;
340 if (nh->nh_ifp != ifp)
344 * Protect (sorta) against walktree recursion problems
347 if ((rt->rte_flags & RTF_UP) == 0)
354 rt_flushifroutes(struct ifnet *ifp)
357 rib_foreach_table_walk_del(AF_UNSPEC, rt_ifdelroute, ifp);
361 * Tries to extract interface from RTAX_IFP passed in rt_addrinfo.
362 * Interface can be specified ether as interface index (sdl_index) or
363 * the interface name (sdl_data).
365 * Returns found ifp or NULL
367 static struct ifnet *
368 info_get_ifp(struct rt_addrinfo *info)
370 const struct sockaddr_dl *sdl;
372 sdl = (const struct sockaddr_dl *)info->rti_info[RTAX_IFP];
373 if (sdl->sdl_family != AF_LINK)
376 if (sdl->sdl_index != 0)
377 return (ifnet_byindex(sdl->sdl_index));
378 if (sdl->sdl_nlen > 0) {
379 char if_name[IF_NAMESIZE];
380 if (sdl->sdl_nlen + offsetof(struct sockaddr_dl, sdl_data) > sdl->sdl_len)
382 if (sdl->sdl_nlen >= IF_NAMESIZE)
384 bzero(if_name, sizeof(if_name));
385 memcpy(if_name, sdl->sdl_data, sdl->sdl_nlen);
386 return (ifunit(if_name));
393 * Calculates proper ifa/ifp for the cases when gateway AF is different
396 * Returns 0 on success.
398 __noinline static int
399 rt_getifa_family(struct rt_addrinfo *info, uint32_t fibnum)
401 if (info->rti_ifp == NULL) {
402 struct ifaddr *ifa = NULL;
404 * No transmit interface specified. Guess it by checking gw sa.
406 const struct sockaddr *gw = info->rti_info[RTAX_GATEWAY];
407 ifa = ifa_ifwithroute(RTF_GATEWAY, gw, gw, fibnum);
409 return (ENETUNREACH);
410 info->rti_ifp = ifa->ifa_ifp;
413 /* Prefer address from outgoing interface */
414 info->rti_ifa = ifaof_ifpforaddr(info->rti_info[RTAX_DST], info->rti_ifp);
416 if (info->rti_ifa == NULL) {
417 /* Use first found IPv4 address */
418 bool loopback_ok = info->rti_ifp->if_flags & IFF_LOOPBACK;
419 info->rti_ifa = (struct ifaddr *)in_findlocal(fibnum, loopback_ok);
422 if (info->rti_ifa == NULL)
423 return (ENETUNREACH);
428 * Fills in rti_ifp and rti_ifa for the provided fib.
430 * Assume basic consistency checks are executed by callers:
431 * RTAX_DST exists, if RTF_GATEWAY is set, RTAX_GATEWAY exists as well.
434 rt_getifa_fib(struct rt_addrinfo *info, u_int fibnum)
436 const struct sockaddr *dst, *gateway, *ifaaddr;
439 dst = info->rti_info[RTAX_DST];
440 gateway = info->rti_info[RTAX_GATEWAY];
441 ifaaddr = info->rti_info[RTAX_IFA];
442 flags = info->rti_flags;
445 * ifp may be specified by sockaddr_dl
446 * when protocol address is ambiguous.
450 /* If we have interface specified by RTAX_IFP address, try to use it */
451 if ((info->rti_ifp == NULL) && (info->rti_info[RTAX_IFP] != NULL))
452 info->rti_ifp = info_get_ifp(info);
454 * If we have source address specified, try to find it
455 * TODO: avoid enumerating all ifas on all interfaces.
457 if (info->rti_ifa == NULL && ifaaddr != NULL)
458 info->rti_ifa = ifa_ifwithaddr(ifaaddr);
459 if ((info->rti_ifa == NULL) && ((info->rti_flags & RTF_GATEWAY) != 0) &&
460 (gateway->sa_family != dst->sa_family))
461 return (rt_getifa_family(info, fibnum));
462 if (info->rti_ifa == NULL) {
463 const struct sockaddr *sa;
466 * Most common use case for the userland-supplied routes.
468 * Choose sockaddr to select ifa.
469 * -- if ifp is set --
470 * Order of preference:
473 * Note: for interface routes link-level gateway address
474 * is specified to indicate the interface index without
475 * specifying RTF_GATEWAY. In this case, ignore gateway
476 * Note: gateway AF may be different from dst AF. In this case,
478 * 3) final destination.
479 * 4) if all of these fails, try to get at least link-level ifa.
481 * try to lookup gateway or dst in the routing table to get ifa
483 if (info->rti_info[RTAX_IFA] != NULL)
484 sa = info->rti_info[RTAX_IFA];
485 else if ((info->rti_flags & RTF_GATEWAY) != 0 &&
486 gateway->sa_family == dst->sa_family)
490 if (info->rti_ifp != NULL) {
491 info->rti_ifa = ifaof_ifpforaddr(sa, info->rti_ifp);
493 if (info->rti_ifa == NULL && gateway != NULL)
494 info->rti_ifa = ifaof_ifpforaddr(gateway, info->rti_ifp);
495 } else if (dst != NULL && gateway != NULL)
496 info->rti_ifa = ifa_ifwithroute(flags, dst, gateway,
499 info->rti_ifa = ifa_ifwithroute(flags, sa, sa,
502 if (info->rti_ifa != NULL) {
503 if (info->rti_ifp == NULL)
504 info->rti_ifp = info->rti_ifa->ifa_ifp;
511 rt_updatemtu(struct ifnet *ifp)
513 struct rib_head *rnh;
518 * Try to update rt_mtu for all routes using this interface
519 * Unfortunately the only way to do this is to traverse all
520 * routing tables in all fibs/domains.
522 for (i = 1; i <= AF_MAX; i++) {
523 mtu = if_getmtu_family(ifp, i);
524 for (j = 0; j < rt_numfibs; j++) {
525 rnh = rt_tables_get_rnh(j, i);
528 nhops_update_ifmtu(rnh, ifp, mtu);
534 int p_sockaddr(char *buf, int buflen, struct sockaddr *s);
535 int rt_print(char *buf, int buflen, struct rtentry *rt);
538 p_sockaddr(char *buf, int buflen, struct sockaddr *s)
542 switch (s->sa_family) {
544 paddr = &((struct sockaddr_in *)s)->sin_addr;
547 paddr = &((struct sockaddr_in6 *)s)->sin6_addr;
554 if (inet_ntop(s->sa_family, paddr, buf, buflen) == NULL)
557 return (strlen(buf));
561 rt_print(char *buf, int buflen, struct rtentry *rt)
563 struct sockaddr *addr, *mask;
569 i = p_sockaddr(buf, buflen, addr);
570 if (!(rt->rt_flags & RTF_HOST)) {
572 i += p_sockaddr(buf + i, buflen - i, mask);
575 if (rt->rt_flags & RTF_GATEWAY) {
577 i += p_sockaddr(buf + i, buflen - i, &rt->rt_nhop->gw_sa);
585 rt_maskedcopy(const struct sockaddr *src, struct sockaddr *dst,
586 const struct sockaddr *netmask)
588 const u_char *cp1 = (const u_char *)src;
589 u_char *cp2 = (u_char *)dst;
590 const u_char *cp3 = (const u_char *)netmask;
591 u_char *cplim = cp2 + *cp3;
592 u_char *cplim2 = cp2 + *cp1;
594 *cp2++ = *cp1++; *cp2++ = *cp1++; /* copies sa_len & sa_family */
599 *cp2++ = *cp1++ & *cp3++;
601 bzero((caddr_t)cp2, (unsigned)(cplim2 - cp2));
605 * Announce interface address arrival/withdraw
606 * Returns 0 on success.
609 rt_addrmsg(int cmd, struct ifaddr *ifa, int fibnum)
611 #if defined(INET) || defined(INET6)
612 struct sockaddr *sa = ifa->ifa_addr;
613 struct ifnet *ifp = ifa->ifa_ifp;
616 KASSERT(cmd == RTM_ADD || cmd == RTM_DELETE,
617 ("unexpected cmd %d", cmd));
618 KASSERT((fibnum >= 0 && fibnum < rt_numfibs),
619 ("%s: fib out of range 0 <=%d<%d", __func__, fibnum, rt_numfibs));
621 EVENTHANDLER_DIRECT_INVOKE(rt_addrmsg, ifa, cmd);
624 if (sa->sa_family == AF_INET) {
625 char addrstr[INET_ADDRSTRLEN];
626 char strbuf[INET_ADDRSTRLEN + 12];
628 inet_ntoa_r(((struct sockaddr_in *)sa)->sin_addr, addrstr);
629 snprintf(strbuf, sizeof(strbuf), "address=%s", addrstr);
630 devctl_notify("IFNET", ifp->if_xname,
631 (cmd == RTM_ADD) ? "ADDR_ADD" : "ADDR_DEL", strbuf);
635 if (sa->sa_family == AF_INET6) {
636 char addrstr[INET6_ADDRSTRLEN];
637 char strbuf[INET6_ADDRSTRLEN + 12];
639 ip6_sprintf(addrstr, IFA_IN6(ifa));
640 snprintf(strbuf, sizeof(strbuf), "address=%s", addrstr);
641 devctl_notify("IFNET", ifp->if_xname,
642 (cmd == RTM_ADD) ? "ADDR_ADD" : "ADDR_DEL", strbuf);
646 if (V_rt_add_addr_allfibs)
647 fibnum = RT_ALL_FIBS;
648 return (rtsock_addrmsg(cmd, ifa, fibnum));
652 * Announce kernel-originated route addition/removal to rtsock based on @rt data.
655 * @nh: nhop object to announce
656 * @fibnum: fib id or RT_ALL_FIBS
658 * Returns 0 on success.
661 rt_routemsg(int cmd, struct rtentry *rt, struct nhop_object *nh,
665 KASSERT(cmd == RTM_ADD || cmd == RTM_DELETE || cmd == RTM_CHANGE,
666 ("unexpected cmd %d", cmd));
668 KASSERT(fibnum == RT_ALL_FIBS || (fibnum >= 0 && fibnum < rt_numfibs),
669 ("%s: fib out of range 0 <=%d<%d", __func__, fibnum, rt_numfibs));
671 KASSERT(rt_key(rt) != NULL, (":%s: rt_key must be supplied", __func__));
673 return (rtsock_routemsg(cmd, rt, nh, fibnum));
677 * Announce kernel-originated route addition/removal to rtsock based on @rt data.
679 * @info: addrinfo structure with valid data.
680 * @fibnum: fib id or RT_ALL_FIBS
682 * Returns 0 on success.
685 rt_routemsg_info(int cmd, struct rt_addrinfo *info, int fibnum)
688 KASSERT(cmd == RTM_ADD || cmd == RTM_DELETE || cmd == RTM_CHANGE,
689 ("unexpected cmd %d", cmd));
691 KASSERT(fibnum == RT_ALL_FIBS || (fibnum >= 0 && fibnum < rt_numfibs),
692 ("%s: fib out of range 0 <=%d<%d", __func__, fibnum, rt_numfibs));
694 KASSERT(info->rti_info[RTAX_DST] != NULL, (":%s: RTAX_DST must be supplied", __func__));
696 return (rtsock_routemsg_info(cmd, info, fibnum));
700 rt_ifmsg(struct ifnet *ifp, int if_flags_mask)
702 rtsock_callback_p->ifmsg_f(ifp, if_flags_mask);
703 netlink_callback_p->ifmsg_f(ifp, if_flags_mask);