2 * Copyright (c) 1988, 1991, 1993
3 * The Regents of the University of California. All rights reserved.
5 * Redistribution and use in source and binary forms, with or without
6 * modification, are permitted provided that the following conditions
8 * 1. Redistributions of source code must retain the above copyright
9 * notice, this list of conditions and the following disclaimer.
10 * 2. Redistributions in binary form must reproduce the above copyright
11 * notice, this list of conditions and the following disclaimer in the
12 * documentation and/or other materials provided with the distribution.
13 * 4. Neither the name of the University nor the names of its contributors
14 * may be used to endorse or promote products derived from this software
15 * without specific prior written permission.
17 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29 * @(#)rtsock.c 8.7 (Berkeley) 10/12/95
32 #include "opt_compat.h"
34 #include "opt_mpath.h"
36 #include "opt_inet6.h"
38 #include <sys/param.h>
40 #include <sys/kernel.h>
41 #include <sys/domain.h>
43 #include <sys/malloc.h>
47 #include <sys/protosw.h>
48 #include <sys/rwlock.h>
49 #include <sys/signalvar.h>
50 #include <sys/socket.h>
51 #include <sys/socketvar.h>
52 #include <sys/sysctl.h>
53 #include <sys/systm.h>
55 #define _IN_NET_RTSOCK_C
57 #include <net/if_dl.h>
58 #include <net/if_llatbl.h>
59 #include <net/if_types.h>
60 #include <net/netisr.h>
61 #include <net/raw_cb.h>
62 #include <net/route.h>
65 #include <netinet/in.h>
66 #include <netinet/if_ether.h>
67 #include <netinet/ip_carp.h>
69 #include <netinet6/ip6_var.h>
70 #include <netinet6/scope6_var.h>
73 #if defined(INET) || defined(INET6)
75 extern void sctp_addr_change(struct ifaddr *ifa, int cmd);
79 #ifdef COMPAT_FREEBSD32
80 #include <sys/mount.h>
81 #include <compat/freebsd32/freebsd32.h>
88 uint8_t ifi_link_state;
90 uint8_t ifi_baudrate_pf;
94 uint32_t ifi_baudrate;
95 uint32_t ifi_ipackets;
97 uint32_t ifi_opackets;
99 uint32_t ifi_collisions;
102 uint32_t ifi_imcasts;
103 uint32_t ifi_omcasts;
104 uint32_t ifi_iqdrops;
105 uint32_t ifi_noproto;
106 uint32_t ifi_hwassist;
108 struct timeval32 ifi_lastchange;
109 uint32_t ifi_oqdrops;
119 struct if_data32 ifm_data;
122 struct if_msghdrl32 {
129 uint16_t _ifm_spare1;
131 uint16_t ifm_data_off;
132 struct if_data32 ifm_data;
135 struct ifa_msghdrl32 {
136 uint16_t ifam_msglen;
137 uint8_t ifam_version;
142 uint16_t _ifam_spare1;
144 uint16_t ifam_data_off;
146 struct if_data32 ifam_data;
148 #endif /* COMPAT_FREEBSD32 */
150 MALLOC_DEFINE(M_RTABLE, "routetbl", "routing tables");
152 /* NB: these are not modified */
153 static struct sockaddr route_src = { 2, PF_ROUTE, };
154 static struct sockaddr sa_zero = { sizeof(sa_zero), AF_INET, };
156 /* These are external hooks for CARP. */
157 int (*carp_get_vhid_p)(struct ifaddr *);
160 * Used by rtsock/raw_input callback code to decide whether to filter the update
161 * notification to a socket bound to a particular FIB.
163 #define RTS_FILTER_FIB M_PROTO8
164 #define RTS_ALLFIBS -1
167 int ip_count; /* attached w/ AF_INET */
168 int ip6_count; /* attached w/ AF_INET6 */
169 int ipx_count; /* attached w/ AF_IPX */
170 int any_count; /* total attached */
173 struct mtx rtsock_mtx;
174 MTX_SYSINIT(rtsock, &rtsock_mtx, "rtsock route_cb lock", MTX_DEF);
176 #define RTSOCK_LOCK() mtx_lock(&rtsock_mtx)
177 #define RTSOCK_UNLOCK() mtx_unlock(&rtsock_mtx)
178 #define RTSOCK_LOCK_ASSERT() mtx_assert(&rtsock_mtx, MA_OWNED)
180 static SYSCTL_NODE(_net, OID_AUTO, route, CTLFLAG_RD, 0, "");
186 struct sysctl_req *w_req;
189 static void rts_input(struct mbuf *m);
190 static struct mbuf *rt_msg1(int type, struct rt_addrinfo *rtinfo);
191 static int rt_msg2(int type, struct rt_addrinfo *rtinfo,
192 caddr_t cp, struct walkarg *w);
193 static int rt_xaddrs(caddr_t cp, caddr_t cplim,
194 struct rt_addrinfo *rtinfo);
195 static int sysctl_dumpentry(struct radix_node *rn, void *vw);
196 static int sysctl_iflist(int af, struct walkarg *w);
197 static int sysctl_ifmalist(int af, struct walkarg *w);
198 static int route_output(struct mbuf *m, struct socket *so);
199 static void rt_setmetrics(const struct rt_msghdr *rtm, struct rtentry *rt);
200 static void rt_getmetrics(const struct rtentry *rt, struct rt_metrics *out);
201 static void rt_dispatch(struct mbuf *, sa_family_t);
203 static struct netisr_handler rtsock_nh = {
205 .nh_handler = rts_input,
206 .nh_proto = NETISR_ROUTE,
207 .nh_policy = NETISR_POLICY_SOURCE,
211 sysctl_route_netisr_maxqlen(SYSCTL_HANDLER_ARGS)
215 netisr_getqlimit(&rtsock_nh, &qlimit);
216 error = sysctl_handle_int(oidp, &qlimit, 0, req);
217 if (error || !req->newptr)
221 return (netisr_setqlimit(&rtsock_nh, qlimit));
223 SYSCTL_PROC(_net_route, OID_AUTO, netisr_maxqlen, CTLTYPE_INT|CTLFLAG_RW,
224 0, 0, sysctl_route_netisr_maxqlen, "I",
225 "maximum routing socket dispatch queue length");
232 if (TUNABLE_INT_FETCH("net.route.netisr_maxqlen", &tmp))
233 rtsock_nh.nh_qlimit = tmp;
234 netisr_register(&rtsock_nh);
236 SYSINIT(rtsock, SI_SUB_PROTO_DOMAIN, SI_ORDER_THIRD, rts_init, 0);
239 raw_input_rts_cb(struct mbuf *m, struct sockproto *proto, struct sockaddr *src,
244 KASSERT(m != NULL, ("%s: m is NULL", __func__));
245 KASSERT(proto != NULL, ("%s: proto is NULL", __func__));
246 KASSERT(rp != NULL, ("%s: rp is NULL", __func__));
248 /* No filtering requested. */
249 if ((m->m_flags & RTS_FILTER_FIB) == 0)
252 /* Check if it is a rts and the fib matches the one of the socket. */
253 fibnum = M_GETFIB(m);
254 if (proto->sp_family != PF_ROUTE ||
255 rp->rcb_socket == NULL ||
256 rp->rcb_socket->so_fibnum == fibnum)
259 /* Filtering requested and no match, the socket shall be skipped. */
264 rts_input(struct mbuf *m)
266 struct sockproto route_proto;
267 unsigned short *family;
270 route_proto.sp_family = PF_ROUTE;
271 tag = m_tag_find(m, PACKET_TAG_RTSOCKFAM, NULL);
273 family = (unsigned short *)(tag + 1);
274 route_proto.sp_protocol = *family;
275 m_tag_delete(m, tag);
277 route_proto.sp_protocol = 0;
279 raw_input_ext(m, &route_proto, &route_src, raw_input_rts_cb);
283 * It really doesn't make any sense at all for this code to share much
284 * with raw_usrreq.c, since its functionality is so restricted. XXX
287 rts_abort(struct socket *so)
290 raw_usrreqs.pru_abort(so);
294 rts_close(struct socket *so)
297 raw_usrreqs.pru_close(so);
300 /* pru_accept is EOPNOTSUPP */
303 rts_attach(struct socket *so, int proto, struct thread *td)
308 KASSERT(so->so_pcb == NULL, ("rts_attach: so_pcb != NULL"));
311 rp = malloc(sizeof *rp, M_PCB, M_WAITOK | M_ZERO);
315 so->so_pcb = (caddr_t)rp;
316 so->so_fibnum = td->td_proc->p_fibnum;
317 error = raw_attach(so, proto);
325 switch(rp->rcb_proto.sp_protocol) {
330 route_cb.ip6_count++;
333 route_cb.ipx_count++;
336 route_cb.any_count++;
339 so->so_options |= SO_USELOOPBACK;
344 rts_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
347 return (raw_usrreqs.pru_bind(so, nam, td)); /* xxx just EINVAL */
351 rts_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
354 return (raw_usrreqs.pru_connect(so, nam, td)); /* XXX just EINVAL */
357 /* pru_connect2 is EOPNOTSUPP */
358 /* pru_control is EOPNOTSUPP */
361 rts_detach(struct socket *so)
363 struct rawcb *rp = sotorawcb(so);
365 KASSERT(rp != NULL, ("rts_detach: rp == NULL"));
368 switch(rp->rcb_proto.sp_protocol) {
373 route_cb.ip6_count--;
376 route_cb.ipx_count--;
379 route_cb.any_count--;
381 raw_usrreqs.pru_detach(so);
385 rts_disconnect(struct socket *so)
388 return (raw_usrreqs.pru_disconnect(so));
391 /* pru_listen is EOPNOTSUPP */
394 rts_peeraddr(struct socket *so, struct sockaddr **nam)
397 return (raw_usrreqs.pru_peeraddr(so, nam));
400 /* pru_rcvd is EOPNOTSUPP */
401 /* pru_rcvoob is EOPNOTSUPP */
404 rts_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
405 struct mbuf *control, struct thread *td)
408 return (raw_usrreqs.pru_send(so, flags, m, nam, control, td));
411 /* pru_sense is null */
414 rts_shutdown(struct socket *so)
417 return (raw_usrreqs.pru_shutdown(so));
421 rts_sockaddr(struct socket *so, struct sockaddr **nam)
424 return (raw_usrreqs.pru_sockaddr(so, nam));
427 static struct pr_usrreqs route_usrreqs = {
428 .pru_abort = rts_abort,
429 .pru_attach = rts_attach,
430 .pru_bind = rts_bind,
431 .pru_connect = rts_connect,
432 .pru_detach = rts_detach,
433 .pru_disconnect = rts_disconnect,
434 .pru_peeraddr = rts_peeraddr,
435 .pru_send = rts_send,
436 .pru_shutdown = rts_shutdown,
437 .pru_sockaddr = rts_sockaddr,
438 .pru_close = rts_close,
441 #ifndef _SOCKADDR_UNION_DEFINED
442 #define _SOCKADDR_UNION_DEFINED
444 * The union of all possible address formats we handle.
446 union sockaddr_union {
448 struct sockaddr_in sin;
449 struct sockaddr_in6 sin6;
451 #endif /* _SOCKADDR_UNION_DEFINED */
454 rtm_get_jailed(struct rt_addrinfo *info, struct ifnet *ifp,
455 struct rtentry *rt, union sockaddr_union *saun, struct ucred *cred)
458 /* First, see if the returned address is part of the jail. */
459 if (prison_if(cred, rt->rt_ifa->ifa_addr) == 0) {
460 info->rti_info[RTAX_IFA] = rt->rt_ifa->ifa_addr;
464 switch (info->rti_info[RTAX_DST]->sa_family) {
474 * Try to find an address on the given outgoing interface
475 * that belongs to the jail.
478 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
481 if (sa->sa_family != AF_INET)
483 ia = ((struct sockaddr_in *)sa)->sin_addr;
484 if (prison_check_ip4(cred, &ia) == 0) {
489 IF_ADDR_RUNLOCK(ifp);
492 * As a last resort return the 'default' jail address.
494 ia = ((struct sockaddr_in *)rt->rt_ifa->ifa_addr)->
496 if (prison_get_ip4(cred, &ia) != 0)
499 bzero(&saun->sin, sizeof(struct sockaddr_in));
500 saun->sin.sin_len = sizeof(struct sockaddr_in);
501 saun->sin.sin_family = AF_INET;
502 saun->sin.sin_addr.s_addr = ia.s_addr;
503 info->rti_info[RTAX_IFA] = (struct sockaddr *)&saun->sin;
516 * Try to find an address on the given outgoing interface
517 * that belongs to the jail.
520 TAILQ_FOREACH(ifa, &ifp->if_addrhead, ifa_link) {
523 if (sa->sa_family != AF_INET6)
525 bcopy(&((struct sockaddr_in6 *)sa)->sin6_addr,
526 &ia6, sizeof(struct in6_addr));
527 if (prison_check_ip6(cred, &ia6) == 0) {
532 IF_ADDR_RUNLOCK(ifp);
535 * As a last resort return the 'default' jail address.
537 ia6 = ((struct sockaddr_in6 *)rt->rt_ifa->ifa_addr)->
539 if (prison_get_ip6(cred, &ia6) != 0)
542 bzero(&saun->sin6, sizeof(struct sockaddr_in6));
543 saun->sin6.sin6_len = sizeof(struct sockaddr_in6);
544 saun->sin6.sin6_family = AF_INET6;
545 bcopy(&ia6, &saun->sin6.sin6_addr, sizeof(struct in6_addr));
546 if (sa6_recoverscope(&saun->sin6) != 0)
548 info->rti_info[RTAX_IFA] = (struct sockaddr *)&saun->sin6;
560 route_output(struct mbuf *m, struct socket *so)
562 #define sa_equal(a1, a2) (bcmp((a1), (a2), (a1)->sa_len) == 0)
563 struct rt_msghdr *rtm = NULL;
564 struct rtentry *rt = NULL;
565 struct radix_node_head *rnh;
566 struct rt_addrinfo info;
568 struct sockaddr_storage ss;
569 struct sockaddr_in6 *sin6;
570 int i, rti_need_deembed = 0;
573 struct ifnet *ifp = NULL;
574 union sockaddr_union saun;
575 sa_family_t saf = AF_UNSPEC;
577 #define senderr(e) { error = e; goto flush;}
578 if (m == NULL || ((m->m_len < sizeof(long)) &&
579 (m = m_pullup(m, sizeof(long))) == NULL))
581 if ((m->m_flags & M_PKTHDR) == 0)
582 panic("route_output");
583 len = m->m_pkthdr.len;
584 if (len < sizeof(*rtm) ||
585 len != mtod(m, struct rt_msghdr *)->rtm_msglen) {
586 info.rti_info[RTAX_DST] = NULL;
589 R_Malloc(rtm, struct rt_msghdr *, len);
591 info.rti_info[RTAX_DST] = NULL;
594 m_copydata(m, 0, len, (caddr_t)rtm);
595 if (rtm->rtm_version != RTM_VERSION) {
596 info.rti_info[RTAX_DST] = NULL;
597 senderr(EPROTONOSUPPORT);
599 rtm->rtm_pid = curproc->p_pid;
600 bzero(&info, sizeof(info));
601 info.rti_addrs = rtm->rtm_addrs;
603 * rt_xaddrs() performs s6_addr[2] := sin6_scope_id for AF_INET6
604 * link-local address because rtrequest requires addresses with
607 if (rt_xaddrs((caddr_t)(rtm + 1), len + (caddr_t)rtm, &info)) {
608 info.rti_info[RTAX_DST] = NULL;
611 info.rti_flags = rtm->rtm_flags;
612 if (info.rti_info[RTAX_DST] == NULL ||
613 info.rti_info[RTAX_DST]->sa_family >= AF_MAX ||
614 (info.rti_info[RTAX_GATEWAY] != NULL &&
615 info.rti_info[RTAX_GATEWAY]->sa_family >= AF_MAX))
617 saf = info.rti_info[RTAX_DST]->sa_family;
619 * Verify that the caller has the appropriate privilege; RTM_GET
620 * is the only operation the non-superuser is allowed.
622 if (rtm->rtm_type != RTM_GET) {
623 error = priv_check(curthread, PRIV_NET_ROUTE);
629 * The given gateway address may be an interface address.
630 * For example, issuing a "route change" command on a route
631 * entry that was created from a tunnel, and the gateway
632 * address given is the local end point. In this case the
633 * RTF_GATEWAY flag must be cleared or the destination will
634 * not be reachable even though there is no error message.
636 if (info.rti_info[RTAX_GATEWAY] != NULL &&
637 info.rti_info[RTAX_GATEWAY]->sa_family != AF_LINK) {
640 bzero(&gw_ro, sizeof(gw_ro));
641 gw_ro.ro_dst = *info.rti_info[RTAX_GATEWAY];
642 rtalloc_ign_fib(&gw_ro, 0, so->so_fibnum);
644 * A host route through the loopback interface is
645 * installed for each interface adddress. In pre 8.0
646 * releases the interface address of a PPP link type
647 * is not reachable locally. This behavior is fixed as
648 * part of the new L2/L3 redesign and rewrite work. The
649 * signature of this interface address route is the
650 * AF_LINK sa_family type of the rt_gateway, and the
651 * rt_ifp has the IFF_LOOPBACK flag set.
653 if (gw_ro.ro_rt != NULL &&
654 gw_ro.ro_rt->rt_gateway->sa_family == AF_LINK &&
655 gw_ro.ro_rt->rt_ifp->if_flags & IFF_LOOPBACK) {
656 info.rti_flags &= ~RTF_GATEWAY;
657 info.rti_flags |= RTF_GWFLAG_COMPAT;
659 if (gw_ro.ro_rt != NULL)
663 switch (rtm->rtm_type) {
664 struct rtentry *saved_nrt;
667 if (info.rti_info[RTAX_GATEWAY] == NULL)
671 /* support for new ARP code */
672 if (info.rti_info[RTAX_GATEWAY]->sa_family == AF_LINK &&
673 (rtm->rtm_flags & RTF_LLDATA) != 0) {
674 error = lla_rt_output(rtm, &info);
677 rti_need_deembed = (V_deembed_scopeid) ? 1 : 0;
681 error = rtrequest1_fib(RTM_ADD, &info, &saved_nrt,
683 if (error == 0 && saved_nrt) {
685 rti_need_deembed = (V_deembed_scopeid) ? 1 : 0;
688 rt_setmetrics(rtm, saved_nrt);
689 rtm->rtm_index = saved_nrt->rt_ifp->if_index;
690 RT_REMREF(saved_nrt);
691 RT_UNLOCK(saved_nrt);
697 /* support for new ARP code */
698 if (info.rti_info[RTAX_GATEWAY] &&
699 (info.rti_info[RTAX_GATEWAY]->sa_family == AF_LINK) &&
700 (rtm->rtm_flags & RTF_LLDATA) != 0) {
701 error = lla_rt_output(rtm, &info);
704 rti_need_deembed = (V_deembed_scopeid) ? 1 : 0;
708 error = rtrequest1_fib(RTM_DELETE, &info, &saved_nrt,
716 /* rt_msg2() will not be used when RTM_DELETE fails. */
717 rti_need_deembed = (V_deembed_scopeid) ? 1 : 0;
724 rnh = rt_tables_get_rnh(so->so_fibnum,
725 info.rti_info[RTAX_DST]->sa_family);
727 senderr(EAFNOSUPPORT);
728 RADIX_NODE_HEAD_RLOCK(rnh);
729 rt = (struct rtentry *) rnh->rnh_lookup(info.rti_info[RTAX_DST],
730 info.rti_info[RTAX_NETMASK], rnh);
731 if (rt == NULL) { /* XXX looks bogus */
732 RADIX_NODE_HEAD_RUNLOCK(rnh);
737 * for RTM_CHANGE/LOCK, if we got multipath routes,
738 * we require users to specify a matching RTAX_GATEWAY.
740 * for RTM_GET, gate is optional even with multipath.
741 * if gate == NULL the first match is returned.
742 * (no need to call rt_mpath_matchgate if gate == NULL)
744 if (rn_mpath_capable(rnh) &&
745 (rtm->rtm_type != RTM_GET || info.rti_info[RTAX_GATEWAY])) {
746 rt = rt_mpath_matchgate(rt, info.rti_info[RTAX_GATEWAY]);
748 RADIX_NODE_HEAD_RUNLOCK(rnh);
754 * If performing proxied L2 entry insertion, and
755 * the actual PPP host entry is found, perform
756 * another search to retrieve the prefix route of
757 * the local end point of the PPP link.
759 if (rtm->rtm_flags & RTF_ANNOUNCE) {
760 struct sockaddr laddr;
762 if (rt->rt_ifp != NULL &&
763 rt->rt_ifp->if_type == IFT_PROPVIRTUAL) {
766 ifa = ifa_ifwithnet(info.rti_info[RTAX_DST], 1);
768 rt_maskedcopy(ifa->ifa_addr,
772 rt_maskedcopy(rt->rt_ifa->ifa_addr,
774 rt->rt_ifa->ifa_netmask);
776 * refactor rt and no lock operation necessary
778 rt = (struct rtentry *)rnh->rnh_matchaddr(&laddr, rnh);
780 RADIX_NODE_HEAD_RUNLOCK(rnh);
786 RADIX_NODE_HEAD_RUNLOCK(rnh);
791 * RTM_CHANGE/LOCK need a perfect match, rn_lookup()
792 * returns a perfect match in case a netmask is
793 * specified. For host routes only a longest prefix
794 * match is returned so it is necessary to compare the
795 * existence of the netmask. If both have a netmask
796 * rnh_lookup() did a perfect match and if none of them
797 * have a netmask both are host routes which is also a
801 if (rtm->rtm_type != RTM_GET &&
802 (!rt_mask(rt) != !info.rti_info[RTAX_NETMASK])) {
807 switch(rtm->rtm_type) {
812 if ((rt->rt_flags & RTF_HOST) == 0
813 ? jailed_without_vnet(curthread->td_ucred)
814 : prison_if(curthread->td_ucred,
819 info.rti_info[RTAX_DST] = rt_key(rt);
820 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
821 info.rti_info[RTAX_NETMASK] = rt_mask(rt);
822 info.rti_info[RTAX_GENMASK] = 0;
823 if (rtm->rtm_addrs & (RTA_IFP | RTA_IFA)) {
826 info.rti_info[RTAX_IFP] =
827 ifp->if_addr->ifa_addr;
828 error = rtm_get_jailed(&info, ifp, rt,
829 &saun, curthread->td_ucred);
834 if (ifp->if_flags & IFF_POINTOPOINT)
835 info.rti_info[RTAX_BRD] =
836 rt->rt_ifa->ifa_dstaddr;
837 rtm->rtm_index = ifp->if_index;
839 info.rti_info[RTAX_IFP] = NULL;
840 info.rti_info[RTAX_IFA] = NULL;
842 } else if ((ifp = rt->rt_ifp) != NULL) {
843 rtm->rtm_index = ifp->if_index;
845 len = rt_msg2(rtm->rtm_type, &info, NULL, NULL);
846 if (len > rtm->rtm_msglen) {
847 struct rt_msghdr *new_rtm;
848 R_Malloc(new_rtm, struct rt_msghdr *, len);
849 if (new_rtm == NULL) {
853 bcopy(rtm, new_rtm, rtm->rtm_msglen);
854 Free(rtm); rtm = new_rtm;
856 (void)rt_msg2(rtm->rtm_type, &info, (caddr_t)rtm, NULL);
857 if (rt->rt_flags & RTF_GWFLAG_COMPAT)
858 rtm->rtm_flags = RTF_GATEWAY |
859 (rt->rt_flags & ~RTF_GWFLAG_COMPAT);
861 rtm->rtm_flags = rt->rt_flags;
862 rt_getmetrics(rt, &rtm->rtm_rmx);
863 rtm->rtm_addrs = info.rti_addrs;
868 * New gateway could require new ifaddr, ifp;
869 * flags may also be different; ifp may be specified
870 * by ll sockaddr when protocol address is ambiguous
872 if (((rt->rt_flags & RTF_GATEWAY) &&
873 info.rti_info[RTAX_GATEWAY] != NULL) ||
874 info.rti_info[RTAX_IFP] != NULL ||
875 (info.rti_info[RTAX_IFA] != NULL &&
876 !sa_equal(info.rti_info[RTAX_IFA],
877 rt->rt_ifa->ifa_addr))) {
879 RADIX_NODE_HEAD_LOCK(rnh);
880 error = rt_getifa_fib(&info, rt->rt_fibnum);
882 * XXXRW: Really we should release this
883 * reference later, but this maintains
884 * historical behavior.
886 if (info.rti_ifa != NULL)
887 ifa_free(info.rti_ifa);
888 RADIX_NODE_HEAD_UNLOCK(rnh);
893 if (info.rti_ifa != NULL &&
894 info.rti_ifa != rt->rt_ifa &&
895 rt->rt_ifa != NULL &&
896 rt->rt_ifa->ifa_rtrequest != NULL) {
897 rt->rt_ifa->ifa_rtrequest(RTM_DELETE, rt,
899 ifa_free(rt->rt_ifa);
901 if (info.rti_info[RTAX_GATEWAY] != NULL) {
903 RADIX_NODE_HEAD_LOCK(rnh);
906 error = rt_setgate(rt, rt_key(rt),
907 info.rti_info[RTAX_GATEWAY]);
908 RADIX_NODE_HEAD_UNLOCK(rnh);
913 rt->rt_flags &= ~RTF_GATEWAY;
914 rt->rt_flags |= (RTF_GATEWAY & info.rti_flags);
916 if (info.rti_ifa != NULL &&
917 info.rti_ifa != rt->rt_ifa) {
918 ifa_ref(info.rti_ifa);
919 rt->rt_ifa = info.rti_ifa;
920 rt->rt_ifp = info.rti_ifp;
922 /* Allow some flags to be toggled on change. */
923 rt->rt_flags = (rt->rt_flags & ~RTF_FMASK) |
924 (rtm->rtm_flags & RTF_FMASK);
925 rt_setmetrics(rtm, rt);
926 rtm->rtm_index = rt->rt_ifp->if_index;
927 if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest)
928 rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, &info);
931 /* We don't support locks anymore */
944 rtm->rtm_errno = error;
946 rtm->rtm_flags |= RTF_DONE;
948 if (rt) /* XXX can this be true? */
951 struct rawcb *rp = NULL;
953 * Check to see if we don't want our own messages.
955 if ((so->so_options & SO_USELOOPBACK) == 0) {
956 if (route_cb.any_count <= 1) {
962 /* There is another listener, so construct message */
967 if (rti_need_deembed) {
968 /* sin6_scope_id is recovered before sending rtm. */
969 sin6 = (struct sockaddr_in6 *)&ss;
970 for (i = 0; i < RTAX_MAX; i++) {
971 if (info.rti_info[i] == NULL)
973 if (info.rti_info[i]->sa_family != AF_INET6)
975 bcopy(info.rti_info[i], sin6, sizeof(*sin6));
976 if (sa6_recoverscope(sin6) == 0)
977 bcopy(sin6, info.rti_info[i],
982 m_copyback(m, 0, rtm->rtm_msglen, (caddr_t)rtm);
983 if (m->m_pkthdr.len < rtm->rtm_msglen) {
986 } else if (m->m_pkthdr.len > rtm->rtm_msglen)
987 m_adj(m, rtm->rtm_msglen - m->m_pkthdr.len);
990 M_SETFIB(m, so->so_fibnum);
991 m->m_flags |= RTS_FILTER_FIB;
994 * XXX insure we don't get a copy by
995 * invalidating our protocol
997 unsigned short family = rp->rcb_proto.sp_family;
998 rp->rcb_proto.sp_family = 0;
1000 rp->rcb_proto.sp_family = family;
1002 rt_dispatch(m, saf);
1004 /* info.rti_info[RTAX_DST] (used above) can point inside of rtm */
1013 rt_setmetrics(const struct rt_msghdr *rtm, struct rtentry *rt)
1016 if (rtm->rtm_inits & RTV_MTU)
1017 rt->rt_mtu = rtm->rtm_rmx.rmx_mtu;
1018 if (rtm->rtm_inits & RTV_WEIGHT)
1019 rt->rt_weight = rtm->rtm_rmx.rmx_weight;
1020 /* Kernel -> userland timebase conversion. */
1021 if (rtm->rtm_inits & RTV_EXPIRE)
1022 rt->rt_expire = rtm->rtm_rmx.rmx_expire ?
1023 rtm->rtm_rmx.rmx_expire - time_second + time_uptime : 0;
1027 rt_getmetrics(const struct rtentry *rt, struct rt_metrics *out)
1030 bzero(out, sizeof(*out));
1031 out->rmx_mtu = rt->rt_mtu;
1032 out->rmx_weight = rt->rt_weight;
1033 out->rmx_pksent = counter_u64_fetch(rt->rt_pksent);
1034 /* Kernel -> userland timebase conversion. */
1035 out->rmx_expire = rt->rt_expire ?
1036 rt->rt_expire - time_uptime + time_second : 0;
1040 * Extract the addresses of the passed sockaddrs.
1041 * Do a little sanity checking so as to avoid bad memory references.
1042 * This data is derived straight from userland.
1045 rt_xaddrs(caddr_t cp, caddr_t cplim, struct rt_addrinfo *rtinfo)
1047 struct sockaddr *sa;
1050 for (i = 0; i < RTAX_MAX && cp < cplim; i++) {
1051 if ((rtinfo->rti_addrs & (1 << i)) == 0)
1053 sa = (struct sockaddr *)cp;
1057 if (cp + sa->sa_len > cplim)
1060 * there are no more.. quit now
1061 * If there are more bits, they are in error.
1062 * I've seen this. route(1) can evidently generate these.
1063 * This causes kernel to core dump.
1064 * for compatibility, If we see this, point to a safe address.
1066 if (sa->sa_len == 0) {
1067 rtinfo->rti_info[i] = &sa_zero;
1068 return (0); /* should be EINVAL but for compat */
1072 if (sa->sa_family == AF_INET6)
1073 sa6_embedscope((struct sockaddr_in6 *)sa,
1076 rtinfo->rti_info[i] = sa;
1083 * Used by the routing socket.
1085 static struct mbuf *
1086 rt_msg1(int type, struct rt_addrinfo *rtinfo)
1088 struct rt_msghdr *rtm;
1091 struct sockaddr *sa;
1093 struct sockaddr_storage ss;
1094 struct sockaddr_in6 *sin6;
1102 len = sizeof(struct ifa_msghdr);
1107 len = sizeof(struct ifma_msghdr);
1111 len = sizeof(struct if_msghdr);
1114 case RTM_IFANNOUNCE:
1116 len = sizeof(struct if_announcemsghdr);
1120 len = sizeof(struct rt_msghdr);
1123 /* XXXGL: can we use MJUMPAGESIZE cluster here? */
1124 KASSERT(len <= MCLBYTES, ("%s: message too big", __func__));
1126 m = m_getcl(M_NOWAIT, MT_DATA, M_PKTHDR);
1128 m = m_gethdr(M_NOWAIT, MT_DATA);
1132 m->m_pkthdr.len = m->m_len = len;
1133 rtm = mtod(m, struct rt_msghdr *);
1134 bzero((caddr_t)rtm, len);
1135 for (i = 0; i < RTAX_MAX; i++) {
1136 if ((sa = rtinfo->rti_info[i]) == NULL)
1138 rtinfo->rti_addrs |= (1 << i);
1141 if (V_deembed_scopeid && sa->sa_family == AF_INET6) {
1142 sin6 = (struct sockaddr_in6 *)&ss;
1143 bcopy(sa, sin6, sizeof(*sin6));
1144 if (sa6_recoverscope(sin6) == 0)
1145 sa = (struct sockaddr *)sin6;
1148 m_copyback(m, len, dlen, (caddr_t)sa);
1151 if (m->m_pkthdr.len != len) {
1155 rtm->rtm_msglen = len;
1156 rtm->rtm_version = RTM_VERSION;
1157 rtm->rtm_type = type;
1162 * Used by the sysctl code and routing socket.
1165 rt_msg2(int type, struct rt_addrinfo *rtinfo, caddr_t cp, struct walkarg *w)
1168 int len, dlen, second_time = 0;
1171 struct sockaddr_storage ss;
1172 struct sockaddr_in6 *sin6;
1175 rtinfo->rti_addrs = 0;
1181 if (w != NULL && w->w_op == NET_RT_IFLISTL) {
1182 #ifdef COMPAT_FREEBSD32
1183 if (w->w_req->flags & SCTL_MASK32)
1184 len = sizeof(struct ifa_msghdrl32);
1187 len = sizeof(struct ifa_msghdrl);
1189 len = sizeof(struct ifa_msghdr);
1193 #ifdef COMPAT_FREEBSD32
1194 if (w != NULL && w->w_req->flags & SCTL_MASK32) {
1195 if (w->w_op == NET_RT_IFLISTL)
1196 len = sizeof(struct if_msghdrl32);
1198 len = sizeof(struct if_msghdr32);
1202 if (w != NULL && w->w_op == NET_RT_IFLISTL)
1203 len = sizeof(struct if_msghdrl);
1205 len = sizeof(struct if_msghdr);
1209 len = sizeof(struct ifma_msghdr);
1213 len = sizeof(struct rt_msghdr);
1218 for (i = 0; i < RTAX_MAX; i++) {
1219 struct sockaddr *sa;
1221 if ((sa = rtinfo->rti_info[i]) == NULL)
1223 rtinfo->rti_addrs |= (1 << i);
1227 if (V_deembed_scopeid && sa->sa_family == AF_INET6) {
1228 sin6 = (struct sockaddr_in6 *)&ss;
1229 bcopy(sa, sin6, sizeof(*sin6));
1230 if (sa6_recoverscope(sin6) == 0)
1231 sa = (struct sockaddr *)sin6;
1234 bcopy((caddr_t)sa, cp, (unsigned)dlen);
1240 if (cp == NULL && w != NULL && !second_time) {
1241 struct walkarg *rw = w;
1244 if (rw->w_tmemsize < len) {
1246 free(rw->w_tmem, M_RTABLE);
1247 rw->w_tmem = (caddr_t)
1248 malloc(len, M_RTABLE, M_NOWAIT);
1250 rw->w_tmemsize = len;
1260 struct rt_msghdr *rtm = (struct rt_msghdr *)cp0;
1262 rtm->rtm_version = RTM_VERSION;
1263 rtm->rtm_type = type;
1264 rtm->rtm_msglen = len;
1270 * This routine is called to generate a message from the routing
1271 * socket indicating that a redirect has occured, a routing lookup
1272 * has failed, or that a protocol has detected timeouts to a particular
1276 rt_missmsg_fib(int type, struct rt_addrinfo *rtinfo, int flags, int error,
1279 struct rt_msghdr *rtm;
1281 struct sockaddr *sa = rtinfo->rti_info[RTAX_DST];
1283 if (route_cb.any_count == 0)
1285 m = rt_msg1(type, rtinfo);
1289 if (fibnum != RTS_ALLFIBS) {
1290 KASSERT(fibnum >= 0 && fibnum < rt_numfibs, ("%s: fibnum out "
1291 "of range 0 <= %d < %d", __func__, fibnum, rt_numfibs));
1292 M_SETFIB(m, fibnum);
1293 m->m_flags |= RTS_FILTER_FIB;
1296 rtm = mtod(m, struct rt_msghdr *);
1297 rtm->rtm_flags = RTF_DONE | flags;
1298 rtm->rtm_errno = error;
1299 rtm->rtm_addrs = rtinfo->rti_addrs;
1300 rt_dispatch(m, sa ? sa->sa_family : AF_UNSPEC);
1304 rt_missmsg(int type, struct rt_addrinfo *rtinfo, int flags, int error)
1307 rt_missmsg_fib(type, rtinfo, flags, error, RTS_ALLFIBS);
1311 * This routine is called to generate a message from the routing
1312 * socket indicating that the status of a network interface has changed.
1315 rt_ifmsg(struct ifnet *ifp)
1317 struct if_msghdr *ifm;
1319 struct rt_addrinfo info;
1321 if (route_cb.any_count == 0)
1323 bzero((caddr_t)&info, sizeof(info));
1324 m = rt_msg1(RTM_IFINFO, &info);
1327 ifm = mtod(m, struct if_msghdr *);
1328 ifm->ifm_index = ifp->if_index;
1329 ifm->ifm_flags = ifp->if_flags | ifp->if_drv_flags;
1330 ifm->ifm_data = ifp->if_data;
1332 rt_dispatch(m, AF_UNSPEC);
1336 * This is called to generate messages from the routing socket
1337 * indicating a network interface has had addresses associated with it.
1338 * if we ever reverse the logic and replace messages TO the routing
1339 * socket indicate a request to configure interfaces, then it will
1340 * be unnecessary as the routing socket will automatically generate
1344 rt_newaddrmsg_fib(int cmd, struct ifaddr *ifa, int error, struct rtentry *rt,
1347 struct rt_addrinfo info;
1348 struct sockaddr *sa = NULL;
1350 struct mbuf *m = NULL;
1351 struct ifnet *ifp = ifa->ifa_ifp;
1353 KASSERT(cmd == RTM_ADD || cmd == RTM_DELETE,
1354 ("unexpected cmd %u", cmd));
1355 #if defined(INET) || defined(INET6)
1358 * notify the SCTP stack
1359 * this will only get called when an address is added/deleted
1360 * XXX pass the ifaddr struct instead if ifa->ifa_addr...
1362 sctp_addr_change(ifa, cmd);
1365 if (route_cb.any_count == 0)
1367 for (pass = 1; pass < 3; pass++) {
1368 bzero((caddr_t)&info, sizeof(info));
1369 if ((cmd == RTM_ADD && pass == 1) ||
1370 (cmd == RTM_DELETE && pass == 2)) {
1371 struct ifa_msghdr *ifam;
1372 int ncmd = cmd == RTM_ADD ? RTM_NEWADDR : RTM_DELADDR;
1374 info.rti_info[RTAX_IFA] = sa = ifa->ifa_addr;
1375 info.rti_info[RTAX_IFP] = ifp->if_addr->ifa_addr;
1376 info.rti_info[RTAX_NETMASK] = ifa->ifa_netmask;
1377 info.rti_info[RTAX_BRD] = ifa->ifa_dstaddr;
1378 if ((m = rt_msg1(ncmd, &info)) == NULL)
1380 ifam = mtod(m, struct ifa_msghdr *);
1381 ifam->ifam_index = ifp->if_index;
1382 ifam->ifam_metric = ifa->ifa_metric;
1383 ifam->ifam_flags = ifa->ifa_flags;
1384 ifam->ifam_addrs = info.rti_addrs;
1386 if ((cmd == RTM_ADD && pass == 2) ||
1387 (cmd == RTM_DELETE && pass == 1)) {
1388 struct rt_msghdr *rtm;
1392 info.rti_info[RTAX_NETMASK] = rt_mask(rt);
1393 info.rti_info[RTAX_DST] = sa = rt_key(rt);
1394 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
1395 if ((m = rt_msg1(cmd, &info)) == NULL)
1397 rtm = mtod(m, struct rt_msghdr *);
1398 rtm->rtm_index = ifp->if_index;
1399 rtm->rtm_flags |= rt->rt_flags;
1400 rtm->rtm_errno = error;
1401 rtm->rtm_addrs = info.rti_addrs;
1403 if (fibnum != RTS_ALLFIBS) {
1404 KASSERT(fibnum >= 0 && fibnum < rt_numfibs, ("%s: "
1405 "fibnum out of range 0 <= %d < %d", __func__,
1406 fibnum, rt_numfibs));
1407 M_SETFIB(m, fibnum);
1408 m->m_flags |= RTS_FILTER_FIB;
1410 rt_dispatch(m, sa ? sa->sa_family : AF_UNSPEC);
1415 rt_newaddrmsg(int cmd, struct ifaddr *ifa, int error, struct rtentry *rt)
1418 rt_newaddrmsg_fib(cmd, ifa, error, rt, RTS_ALLFIBS);
1422 * This is the analogue to the rt_newaddrmsg which performs the same
1423 * function but for multicast group memberhips. This is easier since
1424 * there is no route state to worry about.
1427 rt_newmaddrmsg(int cmd, struct ifmultiaddr *ifma)
1429 struct rt_addrinfo info;
1430 struct mbuf *m = NULL;
1431 struct ifnet *ifp = ifma->ifma_ifp;
1432 struct ifma_msghdr *ifmam;
1434 if (route_cb.any_count == 0)
1437 bzero((caddr_t)&info, sizeof(info));
1438 info.rti_info[RTAX_IFA] = ifma->ifma_addr;
1439 info.rti_info[RTAX_IFP] = ifp ? ifp->if_addr->ifa_addr : NULL;
1441 * If a link-layer address is present, present it as a ``gateway''
1442 * (similarly to how ARP entries, e.g., are presented).
1444 info.rti_info[RTAX_GATEWAY] = ifma->ifma_lladdr;
1445 m = rt_msg1(cmd, &info);
1448 ifmam = mtod(m, struct ifma_msghdr *);
1449 KASSERT(ifp != NULL, ("%s: link-layer multicast address w/o ifp\n",
1451 ifmam->ifmam_index = ifp->if_index;
1452 ifmam->ifmam_addrs = info.rti_addrs;
1453 rt_dispatch(m, ifma->ifma_addr ? ifma->ifma_addr->sa_family : AF_UNSPEC);
1456 static struct mbuf *
1457 rt_makeifannouncemsg(struct ifnet *ifp, int type, int what,
1458 struct rt_addrinfo *info)
1460 struct if_announcemsghdr *ifan;
1463 if (route_cb.any_count == 0)
1465 bzero((caddr_t)info, sizeof(*info));
1466 m = rt_msg1(type, info);
1468 ifan = mtod(m, struct if_announcemsghdr *);
1469 ifan->ifan_index = ifp->if_index;
1470 strlcpy(ifan->ifan_name, ifp->if_xname,
1471 sizeof(ifan->ifan_name));
1472 ifan->ifan_what = what;
1478 * This is called to generate routing socket messages indicating
1479 * IEEE80211 wireless events.
1480 * XXX we piggyback on the RTM_IFANNOUNCE msg format in a clumsy way.
1483 rt_ieee80211msg(struct ifnet *ifp, int what, void *data, size_t data_len)
1486 struct rt_addrinfo info;
1488 m = rt_makeifannouncemsg(ifp, RTM_IEEE80211, what, &info);
1491 * Append the ieee80211 data. Try to stick it in the
1492 * mbuf containing the ifannounce msg; otherwise allocate
1493 * a new mbuf and append.
1495 * NB: we assume m is a single mbuf.
1497 if (data_len > M_TRAILINGSPACE(m)) {
1498 struct mbuf *n = m_get(M_NOWAIT, MT_DATA);
1503 bcopy(data, mtod(n, void *), data_len);
1504 n->m_len = data_len;
1506 } else if (data_len > 0) {
1507 bcopy(data, mtod(m, u_int8_t *) + m->m_len, data_len);
1508 m->m_len += data_len;
1510 if (m->m_flags & M_PKTHDR)
1511 m->m_pkthdr.len += data_len;
1512 mtod(m, struct if_announcemsghdr *)->ifan_msglen += data_len;
1513 rt_dispatch(m, AF_UNSPEC);
1518 * This is called to generate routing socket messages indicating
1519 * network interface arrival and departure.
1522 rt_ifannouncemsg(struct ifnet *ifp, int what)
1525 struct rt_addrinfo info;
1527 m = rt_makeifannouncemsg(ifp, RTM_IFANNOUNCE, what, &info);
1529 rt_dispatch(m, AF_UNSPEC);
1533 rt_dispatch(struct mbuf *m, sa_family_t saf)
1538 * Preserve the family from the sockaddr, if any, in an m_tag for
1539 * use when injecting the mbuf into the routing socket buffer from
1542 if (saf != AF_UNSPEC) {
1543 tag = m_tag_get(PACKET_TAG_RTSOCKFAM, sizeof(unsigned short),
1549 *(unsigned short *)(tag + 1) = saf;
1550 m_tag_prepend(m, tag);
1554 m->m_pkthdr.rcvif = V_loif;
1560 netisr_queue(NETISR_ROUTE, m); /* mbuf is free'd on failure. */
1564 * This is used in dumping the kernel table via sysctl().
1567 sysctl_dumpentry(struct radix_node *rn, void *vw)
1569 struct walkarg *w = vw;
1570 struct rtentry *rt = (struct rtentry *)rn;
1571 int error = 0, size;
1572 struct rt_addrinfo info;
1574 if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg))
1576 if ((rt->rt_flags & RTF_HOST) == 0
1577 ? jailed_without_vnet(w->w_req->td->td_ucred)
1578 : prison_if(w->w_req->td->td_ucred, rt_key(rt)) != 0)
1580 bzero((caddr_t)&info, sizeof(info));
1581 info.rti_info[RTAX_DST] = rt_key(rt);
1582 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
1583 info.rti_info[RTAX_NETMASK] = rt_mask(rt);
1584 info.rti_info[RTAX_GENMASK] = 0;
1586 info.rti_info[RTAX_IFP] = rt->rt_ifp->if_addr->ifa_addr;
1587 info.rti_info[RTAX_IFA] = rt->rt_ifa->ifa_addr;
1588 if (rt->rt_ifp->if_flags & IFF_POINTOPOINT)
1589 info.rti_info[RTAX_BRD] = rt->rt_ifa->ifa_dstaddr;
1591 size = rt_msg2(RTM_GET, &info, NULL, w);
1592 if (w->w_req && w->w_tmem) {
1593 struct rt_msghdr *rtm = (struct rt_msghdr *)w->w_tmem;
1595 if (rt->rt_flags & RTF_GWFLAG_COMPAT)
1596 rtm->rtm_flags = RTF_GATEWAY |
1597 (rt->rt_flags & ~RTF_GWFLAG_COMPAT);
1599 rtm->rtm_flags = rt->rt_flags;
1600 rt_getmetrics(rt, &rtm->rtm_rmx);
1601 rtm->rtm_index = rt->rt_ifp->if_index;
1602 rtm->rtm_errno = rtm->rtm_pid = rtm->rtm_seq = 0;
1603 rtm->rtm_addrs = info.rti_addrs;
1604 error = SYSCTL_OUT(w->w_req, (caddr_t)rtm, size);
1610 #ifdef COMPAT_FREEBSD32
1612 copy_ifdata32(struct if_data *src, struct if_data32 *dst)
1615 bzero(dst, sizeof(*dst));
1616 CP(*src, *dst, ifi_type);
1617 CP(*src, *dst, ifi_physical);
1618 CP(*src, *dst, ifi_addrlen);
1619 CP(*src, *dst, ifi_hdrlen);
1620 CP(*src, *dst, ifi_link_state);
1621 CP(*src, *dst, ifi_vhid);
1622 CP(*src, *dst, ifi_baudrate_pf);
1623 dst->ifi_datalen = sizeof(struct if_data32);
1624 CP(*src, *dst, ifi_mtu);
1625 CP(*src, *dst, ifi_metric);
1626 CP(*src, *dst, ifi_baudrate);
1627 CP(*src, *dst, ifi_ipackets);
1628 CP(*src, *dst, ifi_ierrors);
1629 CP(*src, *dst, ifi_opackets);
1630 CP(*src, *dst, ifi_oerrors);
1631 CP(*src, *dst, ifi_collisions);
1632 CP(*src, *dst, ifi_ibytes);
1633 CP(*src, *dst, ifi_obytes);
1634 CP(*src, *dst, ifi_imcasts);
1635 CP(*src, *dst, ifi_omcasts);
1636 CP(*src, *dst, ifi_iqdrops);
1637 CP(*src, *dst, ifi_noproto);
1638 CP(*src, *dst, ifi_hwassist);
1639 CP(*src, *dst, ifi_epoch);
1640 TV_CP(*src, *dst, ifi_lastchange);
1645 sysctl_iflist_ifml(struct ifnet *ifp, struct rt_addrinfo *info,
1646 struct walkarg *w, int len)
1648 struct if_msghdrl *ifm;
1650 #ifdef COMPAT_FREEBSD32
1651 if (w->w_req->flags & SCTL_MASK32) {
1652 struct if_msghdrl32 *ifm32;
1654 ifm32 = (struct if_msghdrl32 *)w->w_tmem;
1655 ifm32->ifm_addrs = info->rti_addrs;
1656 ifm32->ifm_flags = ifp->if_flags | ifp->if_drv_flags;
1657 ifm32->ifm_index = ifp->if_index;
1658 ifm32->_ifm_spare1 = 0;
1659 ifm32->ifm_len = sizeof(*ifm32);
1660 ifm32->ifm_data_off = offsetof(struct if_msghdrl32, ifm_data);
1662 copy_ifdata32(&ifp->if_data, &ifm32->ifm_data);
1663 /* Fixup if_data carp(4) vhid. */
1664 if (carp_get_vhid_p != NULL)
1665 ifm32->ifm_data.ifi_vhid =
1666 (*carp_get_vhid_p)(ifp->if_addr);
1667 ifm32->ifm_data.ifi_oqdrops = ifp->if_snd.ifq_drops;
1669 return (SYSCTL_OUT(w->w_req, (caddr_t)ifm32, len));
1672 ifm = (struct if_msghdrl *)w->w_tmem;
1673 ifm->ifm_addrs = info->rti_addrs;
1674 ifm->ifm_flags = ifp->if_flags | ifp->if_drv_flags;
1675 ifm->ifm_index = ifp->if_index;
1676 ifm->_ifm_spare1 = 0;
1677 ifm->ifm_len = sizeof(*ifm);
1678 ifm->ifm_data_off = offsetof(struct if_msghdrl, ifm_data);
1680 ifm->ifm_data = ifp->if_data;
1681 /* Fixup if_data carp(4) vhid. */
1682 if (carp_get_vhid_p != NULL)
1683 ifm->ifm_data.ifi_vhid = (*carp_get_vhid_p)(ifp->if_addr);
1685 ifm->ifm_data.ifi_datalen += sizeof(u_long);
1686 ifm->ifi_oqdrops = ifp->if_snd.ifq_drops;
1688 return (SYSCTL_OUT(w->w_req, (caddr_t)ifm, len));
1692 sysctl_iflist_ifm(struct ifnet *ifp, struct rt_addrinfo *info,
1693 struct walkarg *w, int len)
1695 struct if_msghdr *ifm;
1697 #ifdef COMPAT_FREEBSD32
1698 if (w->w_req->flags & SCTL_MASK32) {
1699 struct if_msghdr32 *ifm32;
1701 ifm32 = (struct if_msghdr32 *)w->w_tmem;
1702 ifm32->ifm_addrs = info->rti_addrs;
1703 ifm32->ifm_flags = ifp->if_flags | ifp->if_drv_flags;
1704 ifm32->ifm_index = ifp->if_index;
1706 copy_ifdata32(&ifp->if_data, &ifm32->ifm_data);
1707 /* Fixup if_data carp(4) vhid. */
1708 if (carp_get_vhid_p != NULL)
1709 ifm32->ifm_data.ifi_vhid =
1710 (*carp_get_vhid_p)(ifp->if_addr);
1712 return (SYSCTL_OUT(w->w_req, (caddr_t)ifm32, len));
1715 ifm = (struct if_msghdr *)w->w_tmem;
1716 ifm->ifm_addrs = info->rti_addrs;
1717 ifm->ifm_flags = ifp->if_flags | ifp->if_drv_flags;
1718 ifm->ifm_index = ifp->if_index;
1720 ifm->ifm_data = ifp->if_data;
1721 /* Fixup if_data carp(4) vhid. */
1722 if (carp_get_vhid_p != NULL)
1723 ifm->ifm_data.ifi_vhid = (*carp_get_vhid_p)(ifp->if_addr);
1725 return (SYSCTL_OUT(w->w_req, (caddr_t)ifm, len));
1729 sysctl_iflist_ifaml(struct ifaddr *ifa, struct rt_addrinfo *info,
1730 struct walkarg *w, int len)
1732 struct ifa_msghdrl *ifam;
1734 #ifdef COMPAT_FREEBSD32
1735 if (w->w_req->flags & SCTL_MASK32) {
1736 struct ifa_msghdrl32 *ifam32;
1738 ifam32 = (struct ifa_msghdrl32 *)w->w_tmem;
1739 ifam32->ifam_addrs = info->rti_addrs;
1740 ifam32->ifam_flags = ifa->ifa_flags;
1741 ifam32->ifam_index = ifa->ifa_ifp->if_index;
1742 ifam32->_ifam_spare1 = 0;
1743 ifam32->ifam_len = sizeof(*ifam32);
1744 ifam32->ifam_data_off =
1745 offsetof(struct ifa_msghdrl32, ifam_data);
1746 ifam32->ifam_metric = ifa->ifa_metric;
1748 copy_ifdata32(&ifa->ifa_ifp->if_data, &ifam32->ifam_data);
1749 /* Fixup if_data carp(4) vhid. */
1750 if (carp_get_vhid_p != NULL)
1751 ifam32->ifam_data.ifi_vhid = (*carp_get_vhid_p)(ifa);
1753 return (SYSCTL_OUT(w->w_req, (caddr_t)ifam32, len));
1757 ifam = (struct ifa_msghdrl *)w->w_tmem;
1758 ifam->ifam_addrs = info->rti_addrs;
1759 ifam->ifam_flags = ifa->ifa_flags;
1760 ifam->ifam_index = ifa->ifa_ifp->if_index;
1761 ifam->_ifam_spare1 = 0;
1762 ifam->ifam_len = sizeof(*ifam);
1763 ifam->ifam_data_off = offsetof(struct ifa_msghdrl, ifam_data);
1764 ifam->ifam_metric = ifa->ifa_metric;
1766 ifam->ifam_data = ifa->if_data;
1767 /* Fixup if_data carp(4) vhid. */
1768 if (carp_get_vhid_p != NULL)
1769 ifam->ifam_data.ifi_vhid = (*carp_get_vhid_p)(ifa);
1771 return (SYSCTL_OUT(w->w_req, w->w_tmem, len));
1775 sysctl_iflist_ifam(struct ifaddr *ifa, struct rt_addrinfo *info,
1776 struct walkarg *w, int len)
1778 struct ifa_msghdr *ifam;
1780 ifam = (struct ifa_msghdr *)w->w_tmem;
1781 ifam->ifam_addrs = info->rti_addrs;
1782 ifam->ifam_flags = ifa->ifa_flags;
1783 ifam->ifam_index = ifa->ifa_ifp->if_index;
1784 ifam->ifam_metric = ifa->ifa_metric;
1786 return (SYSCTL_OUT(w->w_req, w->w_tmem, len));
1790 sysctl_iflist(int af, struct walkarg *w)
1794 struct rt_addrinfo info;
1797 bzero((caddr_t)&info, sizeof(info));
1798 IFNET_RLOCK_NOSLEEP();
1799 TAILQ_FOREACH(ifp, &V_ifnet, if_link) {
1800 if (w->w_arg && w->w_arg != ifp->if_index)
1804 info.rti_info[RTAX_IFP] = ifa->ifa_addr;
1805 len = rt_msg2(RTM_IFINFO, &info, NULL, w);
1806 info.rti_info[RTAX_IFP] = NULL;
1807 if (w->w_req && w->w_tmem) {
1808 if (w->w_op == NET_RT_IFLISTL)
1809 error = sysctl_iflist_ifml(ifp, &info, w, len);
1811 error = sysctl_iflist_ifm(ifp, &info, w, len);
1815 while ((ifa = TAILQ_NEXT(ifa, ifa_link)) != NULL) {
1816 if (af && af != ifa->ifa_addr->sa_family)
1818 if (prison_if(w->w_req->td->td_ucred,
1819 ifa->ifa_addr) != 0)
1821 info.rti_info[RTAX_IFA] = ifa->ifa_addr;
1822 info.rti_info[RTAX_NETMASK] = ifa->ifa_netmask;
1823 info.rti_info[RTAX_BRD] = ifa->ifa_dstaddr;
1824 len = rt_msg2(RTM_NEWADDR, &info, NULL, w);
1825 if (w->w_req && w->w_tmem) {
1826 if (w->w_op == NET_RT_IFLISTL)
1827 error = sysctl_iflist_ifaml(ifa, &info,
1830 error = sysctl_iflist_ifam(ifa, &info,
1836 IF_ADDR_RUNLOCK(ifp);
1837 info.rti_info[RTAX_IFA] = info.rti_info[RTAX_NETMASK] =
1838 info.rti_info[RTAX_BRD] = NULL;
1842 IF_ADDR_RUNLOCK(ifp);
1843 IFNET_RUNLOCK_NOSLEEP();
1848 sysctl_ifmalist(int af, struct walkarg *w)
1851 struct ifmultiaddr *ifma;
1852 struct rt_addrinfo info;
1856 bzero((caddr_t)&info, sizeof(info));
1857 IFNET_RLOCK_NOSLEEP();
1858 TAILQ_FOREACH(ifp, &V_ifnet, if_link) {
1859 if (w->w_arg && w->w_arg != ifp->if_index)
1862 info.rti_info[RTAX_IFP] = ifa ? ifa->ifa_addr : NULL;
1864 TAILQ_FOREACH(ifma, &ifp->if_multiaddrs, ifma_link) {
1865 if (af && af != ifma->ifma_addr->sa_family)
1867 if (prison_if(w->w_req->td->td_ucred,
1868 ifma->ifma_addr) != 0)
1870 info.rti_info[RTAX_IFA] = ifma->ifma_addr;
1871 info.rti_info[RTAX_GATEWAY] =
1872 (ifma->ifma_addr->sa_family != AF_LINK) ?
1873 ifma->ifma_lladdr : NULL;
1874 len = rt_msg2(RTM_NEWMADDR, &info, NULL, w);
1875 if (w->w_req && w->w_tmem) {
1876 struct ifma_msghdr *ifmam;
1878 ifmam = (struct ifma_msghdr *)w->w_tmem;
1879 ifmam->ifmam_index = ifma->ifma_ifp->if_index;
1880 ifmam->ifmam_flags = 0;
1881 ifmam->ifmam_addrs = info.rti_addrs;
1882 error = SYSCTL_OUT(w->w_req, w->w_tmem, len);
1884 IF_ADDR_RUNLOCK(ifp);
1889 IF_ADDR_RUNLOCK(ifp);
1892 IFNET_RUNLOCK_NOSLEEP();
1897 sysctl_rtsock(SYSCTL_HANDLER_ARGS)
1899 int *name = (int *)arg1;
1900 u_int namelen = arg2;
1901 struct radix_node_head *rnh = NULL; /* silence compiler. */
1902 int i, lim, error = EINVAL;
1911 if (name[1] == NET_RT_DUMP) {
1913 fib = req->td->td_proc->p_fibnum;
1914 else if (namelen == 4)
1915 fib = (name[3] == -1) ?
1916 req->td->td_proc->p_fibnum : name[3];
1918 return ((namelen < 3) ? EISDIR : ENOTDIR);
1919 if (fib < 0 || fib >= rt_numfibs)
1921 } else if (namelen != 3)
1922 return ((namelen < 3) ? EISDIR : ENOTDIR);
1926 bzero(&w, sizeof(w));
1931 error = sysctl_wire_old_buffer(req, 0);
1938 if (af == 0) { /* dump all tables */
1941 } else /* dump only one table */
1945 * take care of llinfo entries, the caller must
1948 if (w.w_op == NET_RT_FLAGS &&
1949 (w.w_arg == 0 || w.w_arg & RTF_LLINFO)) {
1951 error = lltable_sysctl_dumparp(af, w.w_req);
1957 * take care of routing entries
1959 for (error = 0; error == 0 && i <= lim; i++) {
1960 rnh = rt_tables_get_rnh(fib, i);
1962 RADIX_NODE_HEAD_RLOCK(rnh);
1963 error = rnh->rnh_walktree(rnh,
1964 sysctl_dumpentry, &w);
1965 RADIX_NODE_HEAD_RUNLOCK(rnh);
1967 error = EAFNOSUPPORT;
1972 case NET_RT_IFLISTL:
1973 error = sysctl_iflist(af, &w);
1976 case NET_RT_IFMALIST:
1977 error = sysctl_ifmalist(af, &w);
1981 free(w.w_tmem, M_RTABLE);
1985 static SYSCTL_NODE(_net, PF_ROUTE, routetable, CTLFLAG_RD, sysctl_rtsock, "");
1988 * Definitions of protocols supported in the ROUTE domain.
1991 static struct domain routedomain; /* or at least forward */
1993 static struct protosw routesw[] = {
1995 .pr_type = SOCK_RAW,
1996 .pr_domain = &routedomain,
1997 .pr_flags = PR_ATOMIC|PR_ADDR,
1998 .pr_output = route_output,
1999 .pr_ctlinput = raw_ctlinput,
2000 .pr_init = raw_init,
2001 .pr_usrreqs = &route_usrreqs
2005 static struct domain routedomain = {
2006 .dom_family = PF_ROUTE,
2007 .dom_name = "route",
2008 .dom_protosw = routesw,
2009 .dom_protoswNPROTOSW = &routesw[sizeof(routesw)/sizeof(routesw[0])]
2012 VNET_DOMAIN_SET(route);