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
33 #include <sys/param.h>
34 #include <sys/domain.h>
35 #include <sys/kernel.h>
37 #include <sys/malloc.h>
40 #include <sys/protosw.h>
41 #include <sys/signalvar.h>
42 #include <sys/socket.h>
43 #include <sys/socketvar.h>
44 #include <sys/sysctl.h>
45 #include <sys/systm.h>
48 #include <net/netisr.h>
49 #include <net/raw_cb.h>
50 #include <net/route.h>
52 #include <netinet/in.h>
54 MALLOC_DEFINE(M_RTABLE, "routetbl", "routing tables");
56 /* NB: these are not modified */
57 static struct sockaddr route_dst = { 2, PF_ROUTE, };
58 static struct sockaddr route_src = { 2, PF_ROUTE, };
59 static struct sockaddr sa_zero = { sizeof(sa_zero), AF_INET, };
62 int ip_count; /* attached w/ AF_INET */
63 int ip6_count; /* attached w/ AF_INET6 */
64 int ipx_count; /* attached w/ AF_IPX */
65 int any_count; /* total attached */
68 struct mtx rtsock_mtx;
69 MTX_SYSINIT(rtsock, &rtsock_mtx, "rtsock route_cb lock", MTX_DEF);
71 #define RTSOCK_LOCK() mtx_lock(&rtsock_mtx)
72 #define RTSOCK_UNLOCK() mtx_unlock(&rtsock_mtx)
73 #define RTSOCK_LOCK_ASSERT() mtx_assert(&rtsock_mtx, MA_OWNED)
75 static struct ifqueue rtsintrq;
77 SYSCTL_NODE(_net, OID_AUTO, route, CTLFLAG_RD, 0, "");
78 SYSCTL_INT(_net_route, OID_AUTO, netisr_maxqlen, CTLFLAG_RW,
79 &rtsintrq.ifq_maxlen, 0, "maximum routing socket dispatch queue length");
85 struct sysctl_req *w_req;
88 static void rts_input(struct mbuf *m);
89 static struct mbuf *rt_msg1(int type, struct rt_addrinfo *rtinfo);
90 static int rt_msg2(int type, struct rt_addrinfo *rtinfo,
91 caddr_t cp, struct walkarg *w);
92 static int rt_xaddrs(caddr_t cp, caddr_t cplim,
93 struct rt_addrinfo *rtinfo);
94 static int sysctl_dumpentry(struct radix_node *rn, void *vw);
95 static int sysctl_iflist(int af, struct walkarg *w);
96 static int sysctl_ifmalist(int af, struct walkarg *w);
97 static int route_output(struct mbuf *m, struct socket *so);
98 static void rt_setmetrics(u_long which, const struct rt_metrics *in,
99 struct rt_metrics_lite *out);
100 static void rt_getmetrics(const struct rt_metrics_lite *in,
101 struct rt_metrics *out);
102 static void rt_dispatch(struct mbuf *, const struct sockaddr *);
109 rtsintrq.ifq_maxlen = 256;
110 if (TUNABLE_INT_FETCH("net.route.netisr_maxqlen", &tmp))
111 rtsintrq.ifq_maxlen = tmp;
112 mtx_init(&rtsintrq.ifq_mtx, "rts_inq", NULL, MTX_DEF);
113 netisr_register(NETISR_ROUTE, rts_input, &rtsintrq, NETISR_MPSAFE);
115 SYSINIT(rtsock, SI_SUB_PROTO_DOMAIN, SI_ORDER_THIRD, rts_init, 0)
118 rts_input(struct mbuf *m)
120 struct sockproto route_proto;
121 unsigned short *family;
124 route_proto.sp_family = PF_ROUTE;
125 tag = m_tag_find(m, PACKET_TAG_RTSOCKFAM, NULL);
127 family = (unsigned short *)(tag + 1);
128 route_proto.sp_protocol = *family;
129 m_tag_delete(m, tag);
131 route_proto.sp_protocol = 0;
133 raw_input(m, &route_proto, &route_src, &route_dst);
137 * It really doesn't make any sense at all for this code to share much
138 * with raw_usrreq.c, since its functionality is so restricted. XXX
141 rts_abort(struct socket *so)
144 return (raw_usrreqs.pru_abort(so));
147 /* pru_accept is EOPNOTSUPP */
150 rts_attach(struct socket *so, int proto, struct thread *td)
155 if (sotorawcb(so) != NULL)
156 return EISCONN; /* XXX panic? */
158 MALLOC(rp, struct rawcb *, sizeof *rp, M_PCB, M_WAITOK | M_ZERO);
163 * The splnet() is necessary to block protocols from sending
164 * error notifications (like RTM_REDIRECT or RTM_LOSING) while
165 * this PCB is extant but incompletely initialized.
166 * Probably we should try to do more of this work beforehand and
170 so->so_pcb = (caddr_t)rp;
171 error = raw_attach(so, proto);
180 switch(rp->rcb_proto.sp_protocol) {
185 route_cb.ip6_count++;
188 route_cb.ipx_count++;
191 rp->rcb_faddr = &route_src;
192 route_cb.any_count++;
195 so->so_options |= SO_USELOOPBACK;
201 rts_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
204 return (raw_usrreqs.pru_bind(so, nam, td)); /* xxx just EINVAL */
208 rts_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
211 return (raw_usrreqs.pru_connect(so, nam, td)); /* XXX just EINVAL */
214 /* pru_connect2 is EOPNOTSUPP */
215 /* pru_control is EOPNOTSUPP */
218 rts_detach(struct socket *so)
220 struct rawcb *rp = sotorawcb(so);
226 switch(rp->rcb_proto.sp_protocol) {
231 route_cb.ip6_count--;
234 route_cb.ipx_count--;
237 route_cb.any_count--;
240 error = raw_usrreqs.pru_detach(so);
246 rts_disconnect(struct socket *so)
249 return (raw_usrreqs.pru_disconnect(so));
252 /* pru_listen is EOPNOTSUPP */
255 rts_peeraddr(struct socket *so, struct sockaddr **nam)
258 return (raw_usrreqs.pru_peeraddr(so, nam));
261 /* pru_rcvd is EOPNOTSUPP */
262 /* pru_rcvoob is EOPNOTSUPP */
265 rts_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
266 struct mbuf *control, struct thread *td)
269 return (raw_usrreqs.pru_send(so, flags, m, nam, control, td));
272 /* pru_sense is null */
275 rts_shutdown(struct socket *so)
278 return (raw_usrreqs.pru_shutdown(so));
282 rts_sockaddr(struct socket *so, struct sockaddr **nam)
285 return (raw_usrreqs.pru_sockaddr(so, nam));
288 static struct pr_usrreqs route_usrreqs = {
289 .pru_abort = rts_abort,
290 .pru_attach = rts_attach,
291 .pru_bind = rts_bind,
292 .pru_connect = rts_connect,
293 .pru_detach = rts_detach,
294 .pru_disconnect = rts_disconnect,
295 .pru_peeraddr = rts_peeraddr,
296 .pru_send = rts_send,
297 .pru_shutdown = rts_shutdown,
298 .pru_sockaddr = rts_sockaddr,
303 route_output(struct mbuf *m, struct socket *so)
305 #define sa_equal(a1, a2) (bcmp((a1), (a2), (a1)->sa_len) == 0)
306 struct rt_msghdr *rtm = NULL;
307 struct rtentry *rt = NULL;
308 struct radix_node_head *rnh;
309 struct rt_addrinfo info;
311 struct ifnet *ifp = NULL;
312 struct ifaddr *ifa = NULL;
313 struct sockaddr_in jail;
315 #define senderr(e) { error = e; goto flush;}
316 if (m == NULL || ((m->m_len < sizeof(long)) &&
317 (m = m_pullup(m, sizeof(long))) == NULL))
319 if ((m->m_flags & M_PKTHDR) == 0)
320 panic("route_output");
321 len = m->m_pkthdr.len;
322 if (len < sizeof(*rtm) ||
323 len != mtod(m, struct rt_msghdr *)->rtm_msglen) {
324 info.rti_info[RTAX_DST] = NULL;
327 R_Malloc(rtm, struct rt_msghdr *, len);
329 info.rti_info[RTAX_DST] = NULL;
332 m_copydata(m, 0, len, (caddr_t)rtm);
333 if (rtm->rtm_version != RTM_VERSION) {
334 info.rti_info[RTAX_DST] = NULL;
335 senderr(EPROTONOSUPPORT);
337 rtm->rtm_pid = curproc->p_pid;
338 bzero(&info, sizeof(info));
339 info.rti_addrs = rtm->rtm_addrs;
340 if (rt_xaddrs((caddr_t)(rtm + 1), len + (caddr_t)rtm, &info)) {
341 info.rti_info[RTAX_DST] = NULL;
344 info.rti_flags = rtm->rtm_flags;
345 if (info.rti_info[RTAX_DST] == NULL ||
346 info.rti_info[RTAX_DST]->sa_family >= AF_MAX ||
347 (info.rti_info[RTAX_GATEWAY] != NULL &&
348 info.rti_info[RTAX_GATEWAY]->sa_family >= AF_MAX))
350 if (info.rti_info[RTAX_GENMASK]) {
351 struct radix_node *t;
352 t = rn_addmask((caddr_t) info.rti_info[RTAX_GENMASK], 0, 1);
354 bcmp((char *)(void *)info.rti_info[RTAX_GENMASK] + 1,
355 (char *)(void *)t->rn_key + 1,
356 ((struct sockaddr *)t->rn_key)->sa_len - 1) == 0)
357 info.rti_info[RTAX_GENMASK] =
358 (struct sockaddr *)t->rn_key;
364 * Verify that the caller has the appropriate privilege; RTM_GET
365 * is the only operation the non-superuser is allowed.
367 if (rtm->rtm_type != RTM_GET && (error = suser(curthread)) != 0)
370 switch (rtm->rtm_type) {
371 struct rtentry *saved_nrt;
374 if (info.rti_info[RTAX_GATEWAY] == NULL)
377 error = rtrequest1(RTM_ADD, &info, &saved_nrt);
378 if (error == 0 && saved_nrt) {
380 rt_setmetrics(rtm->rtm_inits,
381 &rtm->rtm_rmx, &saved_nrt->rt_rmx);
382 rtm->rtm_index = saved_nrt->rt_ifp->if_index;
383 RT_REMREF(saved_nrt);
384 saved_nrt->rt_genmask = info.rti_info[RTAX_GENMASK];
385 RT_UNLOCK(saved_nrt);
391 error = rtrequest1(RTM_DELETE, &info, &saved_nrt);
402 rnh = rt_tables[info.rti_info[RTAX_DST]->sa_family];
404 senderr(EAFNOSUPPORT);
405 RADIX_NODE_HEAD_LOCK(rnh);
406 rt = (struct rtentry *) rnh->rnh_lookup(info.rti_info[RTAX_DST],
407 info.rti_info[RTAX_NETMASK], rnh);
408 if (rt == NULL) { /* XXX looks bogus */
409 RADIX_NODE_HEAD_UNLOCK(rnh);
414 RADIX_NODE_HEAD_UNLOCK(rnh);
419 * RTM_CHANGE/LOCK need a perfect match, rn_lookup()
420 * returns a perfect match in case a netmask is
421 * specified. For host routes only a longest prefix
422 * match is returned so it is necessary to compare the
423 * existence of the netmask. If both have a netmask
424 * rnh_lookup() did a perfect match and if none of them
425 * have a netmask both are host routes which is also a
429 if (rtm->rtm_type != RTM_GET &&
430 (!rt_mask(rt) != !info.rti_info[RTAX_NETMASK])) {
435 switch(rtm->rtm_type) {
440 info.rti_info[RTAX_DST] = rt_key(rt);
441 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
442 info.rti_info[RTAX_NETMASK] = rt_mask(rt);
443 info.rti_info[RTAX_GENMASK] = rt->rt_genmask;
444 if (rtm->rtm_addrs & (RTA_IFP | RTA_IFA)) {
447 info.rti_info[RTAX_IFP] =
448 ifaddr_byindex(ifp->if_index)->ifa_addr;
449 if (jailed(so->so_cred)) {
450 bzero(&jail, sizeof(jail));
451 jail.sin_family = PF_INET;
452 jail.sin_len = sizeof(jail);
453 jail.sin_addr.s_addr =
454 htonl(prison_getip(so->so_cred));
455 info.rti_info[RTAX_IFA] =
456 (struct sockaddr *)&jail;
458 info.rti_info[RTAX_IFA] =
459 rt->rt_ifa->ifa_addr;
460 if (ifp->if_flags & IFF_POINTOPOINT)
461 info.rti_info[RTAX_BRD] =
462 rt->rt_ifa->ifa_dstaddr;
463 rtm->rtm_index = ifp->if_index;
465 info.rti_info[RTAX_IFP] = NULL;
466 info.rti_info[RTAX_IFA] = NULL;
468 } else if ((ifp = rt->rt_ifp) != NULL) {
469 rtm->rtm_index = ifp->if_index;
471 len = rt_msg2(rtm->rtm_type, &info, NULL, NULL);
472 if (len > rtm->rtm_msglen) {
473 struct rt_msghdr *new_rtm;
474 R_Malloc(new_rtm, struct rt_msghdr *, len);
475 if (new_rtm == NULL) {
479 bcopy(rtm, new_rtm, rtm->rtm_msglen);
480 Free(rtm); rtm = new_rtm;
482 (void)rt_msg2(rtm->rtm_type, &info, (caddr_t)rtm, NULL);
483 rtm->rtm_flags = rt->rt_flags;
485 rt_getmetrics(&rt->rt_rmx, &rtm->rtm_rmx);
486 rtm->rtm_addrs = info.rti_addrs;
491 * New gateway could require new ifaddr, ifp;
492 * flags may also be different; ifp may be specified
493 * by ll sockaddr when protocol address is ambiguous
495 if (((rt->rt_flags & RTF_GATEWAY) &&
496 info.rti_info[RTAX_GATEWAY] != NULL) ||
497 info.rti_info[RTAX_IFP] != NULL ||
498 (info.rti_info[RTAX_IFA] != NULL &&
499 !sa_equal(info.rti_info[RTAX_IFA],
500 rt->rt_ifa->ifa_addr))) {
502 if ((error = rt_getifa(&info)) != 0)
506 if (info.rti_info[RTAX_GATEWAY] != NULL &&
507 (error = rt_setgate(rt, rt_key(rt),
508 info.rti_info[RTAX_GATEWAY])) != 0) {
512 if ((ifa = info.rti_ifa) != NULL) {
513 struct ifaddr *oifa = rt->rt_ifa;
516 if (oifa->ifa_rtrequest)
524 rt->rt_ifp = info.rti_ifp;
527 /* Allow some flags to be toggled on change. */
528 if (rtm->rtm_fmask & RTF_FMASK)
529 rt->rt_flags = (rt->rt_flags &
531 (rtm->rtm_flags & rtm->rtm_fmask);
532 rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx,
534 rtm->rtm_index = rt->rt_ifp->if_index;
535 if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest)
536 rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, &info);
537 if (info.rti_info[RTAX_GENMASK])
538 rt->rt_genmask = info.rti_info[RTAX_GENMASK];
541 /* We don't support locks anymore */
554 rtm->rtm_errno = error;
556 rtm->rtm_flags |= RTF_DONE;
558 if (rt) /* XXX can this be true? */
561 struct rawcb *rp = NULL;
563 * Check to see if we don't want our own messages.
565 if ((so->so_options & SO_USELOOPBACK) == 0) {
566 if (route_cb.any_count <= 1) {
572 /* There is another listener, so construct message */
576 m_copyback(m, 0, rtm->rtm_msglen, (caddr_t)rtm);
577 if (m->m_pkthdr.len < rtm->rtm_msglen) {
580 } else if (m->m_pkthdr.len > rtm->rtm_msglen)
581 m_adj(m, rtm->rtm_msglen - m->m_pkthdr.len);
587 * XXX insure we don't get a copy by
588 * invalidating our protocol
590 unsigned short family = rp->rcb_proto.sp_family;
591 rp->rcb_proto.sp_family = 0;
592 rt_dispatch(m, info.rti_info[RTAX_DST]);
593 rp->rcb_proto.sp_family = family;
595 rt_dispatch(m, info.rti_info[RTAX_DST]);
603 rt_setmetrics(u_long which, const struct rt_metrics *in,
604 struct rt_metrics_lite *out)
606 #define metric(f, e) if (which & (f)) out->e = in->e;
608 * Only these are stored in the routing entry since introduction
609 * of tcp hostcache. The rest is ignored.
611 metric(RTV_MTU, rmx_mtu);
612 metric(RTV_EXPIRE, rmx_expire);
617 rt_getmetrics(const struct rt_metrics_lite *in, struct rt_metrics *out)
619 #define metric(e) out->e = in->e;
620 bzero(out, sizeof(*out));
627 * Extract the addresses of the passed sockaddrs.
628 * Do a little sanity checking so as to avoid bad memory references.
629 * This data is derived straight from userland.
632 rt_xaddrs(caddr_t cp, caddr_t cplim, struct rt_addrinfo *rtinfo)
637 for (i = 0; i < RTAX_MAX && cp < cplim; i++) {
638 if ((rtinfo->rti_addrs & (1 << i)) == 0)
640 sa = (struct sockaddr *)cp;
644 if (cp + sa->sa_len > cplim)
647 * there are no more.. quit now
648 * If there are more bits, they are in error.
649 * I've seen this. route(1) can evidently generate these.
650 * This causes kernel to core dump.
651 * for compatibility, If we see this, point to a safe address.
653 if (sa->sa_len == 0) {
654 rtinfo->rti_info[i] = &sa_zero;
655 return (0); /* should be EINVAL but for compat */
658 rtinfo->rti_info[i] = sa;
665 rt_msg1(int type, struct rt_addrinfo *rtinfo)
667 struct rt_msghdr *rtm;
677 len = sizeof(struct ifa_msghdr);
682 len = sizeof(struct ifma_msghdr);
686 len = sizeof(struct if_msghdr);
691 len = sizeof(struct if_announcemsghdr);
695 len = sizeof(struct rt_msghdr);
699 m = m_gethdr(M_DONTWAIT, MT_DATA);
700 if (m && len > MHLEN) {
701 MCLGET(m, M_DONTWAIT);
702 if ((m->m_flags & M_EXT) == 0) {
709 m->m_pkthdr.len = m->m_len = len;
710 m->m_pkthdr.rcvif = NULL;
711 rtm = mtod(m, struct rt_msghdr *);
712 bzero((caddr_t)rtm, len);
713 for (i = 0; i < RTAX_MAX; i++) {
714 if ((sa = rtinfo->rti_info[i]) == NULL)
716 rtinfo->rti_addrs |= (1 << i);
718 m_copyback(m, len, dlen, (caddr_t)sa);
721 if (m->m_pkthdr.len != len) {
725 rtm->rtm_msglen = len;
726 rtm->rtm_version = RTM_VERSION;
727 rtm->rtm_type = type;
732 rt_msg2(int type, struct rt_addrinfo *rtinfo, caddr_t cp, struct walkarg *w)
735 int len, dlen, second_time = 0;
738 rtinfo->rti_addrs = 0;
744 len = sizeof(struct ifa_msghdr);
748 len = sizeof(struct if_msghdr);
752 len = sizeof(struct ifma_msghdr);
756 len = sizeof(struct rt_msghdr);
761 for (i = 0; i < RTAX_MAX; i++) {
764 if ((sa = rtinfo->rti_info[i]) == NULL)
766 rtinfo->rti_addrs |= (1 << i);
769 bcopy((caddr_t)sa, cp, (unsigned)dlen);
775 if (cp == NULL && w != NULL && !second_time) {
776 struct walkarg *rw = w;
779 if (rw->w_tmemsize < len) {
781 free(rw->w_tmem, M_RTABLE);
782 rw->w_tmem = (caddr_t)
783 malloc(len, M_RTABLE, M_NOWAIT);
785 rw->w_tmemsize = len;
795 struct rt_msghdr *rtm = (struct rt_msghdr *)cp0;
797 rtm->rtm_version = RTM_VERSION;
798 rtm->rtm_type = type;
799 rtm->rtm_msglen = len;
805 * This routine is called to generate a message from the routing
806 * socket indicating that a redirect has occured, a routing lookup
807 * has failed, or that a protocol has detected timeouts to a particular
811 rt_missmsg(int type, struct rt_addrinfo *rtinfo, int flags, int error)
813 struct rt_msghdr *rtm;
815 struct sockaddr *sa = rtinfo->rti_info[RTAX_DST];
817 if (route_cb.any_count == 0)
819 m = rt_msg1(type, rtinfo);
822 rtm = mtod(m, struct rt_msghdr *);
823 rtm->rtm_flags = RTF_DONE | flags;
824 rtm->rtm_errno = error;
825 rtm->rtm_addrs = rtinfo->rti_addrs;
830 * This routine is called to generate a message from the routing
831 * socket indicating that the status of a network interface has changed.
834 rt_ifmsg(struct ifnet *ifp)
836 struct if_msghdr *ifm;
838 struct rt_addrinfo info;
840 if (route_cb.any_count == 0)
842 bzero((caddr_t)&info, sizeof(info));
843 m = rt_msg1(RTM_IFINFO, &info);
846 ifm = mtod(m, struct if_msghdr *);
847 ifm->ifm_index = ifp->if_index;
848 ifm->ifm_flags = ifp->if_flags | ifp->if_drv_flags;
849 ifm->ifm_data = ifp->if_data;
851 rt_dispatch(m, NULL);
855 * This is called to generate messages from the routing socket
856 * indicating a network interface has had addresses associated with it.
857 * if we ever reverse the logic and replace messages TO the routing
858 * socket indicate a request to configure interfaces, then it will
859 * be unnecessary as the routing socket will automatically generate
863 rt_newaddrmsg(int cmd, struct ifaddr *ifa, int error, struct rtentry *rt)
865 struct rt_addrinfo info;
866 struct sockaddr *sa = NULL;
868 struct mbuf *m = NULL;
869 struct ifnet *ifp = ifa->ifa_ifp;
871 KASSERT(cmd == RTM_ADD || cmd == RTM_DELETE,
872 ("unexpected cmd %u", cmd));
874 if (route_cb.any_count == 0)
876 for (pass = 1; pass < 3; pass++) {
877 bzero((caddr_t)&info, sizeof(info));
878 if ((cmd == RTM_ADD && pass == 1) ||
879 (cmd == RTM_DELETE && pass == 2)) {
880 struct ifa_msghdr *ifam;
881 int ncmd = cmd == RTM_ADD ? RTM_NEWADDR : RTM_DELADDR;
883 info.rti_info[RTAX_IFA] = sa = ifa->ifa_addr;
884 info.rti_info[RTAX_IFP] =
885 ifaddr_byindex(ifp->if_index)->ifa_addr;
886 info.rti_info[RTAX_NETMASK] = ifa->ifa_netmask;
887 info.rti_info[RTAX_BRD] = ifa->ifa_dstaddr;
888 if ((m = rt_msg1(ncmd, &info)) == NULL)
890 ifam = mtod(m, struct ifa_msghdr *);
891 ifam->ifam_index = ifp->if_index;
892 ifam->ifam_metric = ifa->ifa_metric;
893 ifam->ifam_flags = ifa->ifa_flags;
894 ifam->ifam_addrs = info.rti_addrs;
896 if ((cmd == RTM_ADD && pass == 2) ||
897 (cmd == RTM_DELETE && pass == 1)) {
898 struct rt_msghdr *rtm;
902 info.rti_info[RTAX_NETMASK] = rt_mask(rt);
903 info.rti_info[RTAX_DST] = sa = rt_key(rt);
904 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
905 if ((m = rt_msg1(cmd, &info)) == NULL)
907 rtm = mtod(m, struct rt_msghdr *);
908 rtm->rtm_index = ifp->if_index;
909 rtm->rtm_flags |= rt->rt_flags;
910 rtm->rtm_errno = error;
911 rtm->rtm_addrs = info.rti_addrs;
918 * This is the analogue to the rt_newaddrmsg which performs the same
919 * function but for multicast group memberhips. This is easier since
920 * there is no route state to worry about.
923 rt_newmaddrmsg(int cmd, struct ifmultiaddr *ifma)
925 struct rt_addrinfo info;
926 struct mbuf *m = NULL;
927 struct ifnet *ifp = ifma->ifma_ifp;
928 struct ifma_msghdr *ifmam;
930 if (route_cb.any_count == 0)
933 bzero((caddr_t)&info, sizeof(info));
934 info.rti_info[RTAX_IFA] = ifma->ifma_addr;
935 info.rti_info[RTAX_IFP] =
936 ifp ? ifaddr_byindex(ifp->if_index)->ifa_addr : NULL;
938 * If a link-layer address is present, present it as a ``gateway''
939 * (similarly to how ARP entries, e.g., are presented).
941 info.rti_info[RTAX_GATEWAY] = ifma->ifma_lladdr;
942 m = rt_msg1(cmd, &info);
945 ifmam = mtod(m, struct ifma_msghdr *);
946 ifmam->ifmam_index = ifp->if_index;
947 ifmam->ifmam_addrs = info.rti_addrs;
948 rt_dispatch(m, ifma->ifma_addr);
952 rt_makeifannouncemsg(struct ifnet *ifp, int type, int what,
953 struct rt_addrinfo *info)
955 struct if_announcemsghdr *ifan;
958 if (route_cb.any_count == 0)
960 bzero((caddr_t)info, sizeof(*info));
961 m = rt_msg1(type, info);
963 ifan = mtod(m, struct if_announcemsghdr *);
964 ifan->ifan_index = ifp->if_index;
965 strlcpy(ifan->ifan_name, ifp->if_xname,
966 sizeof(ifan->ifan_name));
967 ifan->ifan_what = what;
973 * This is called to generate routing socket messages indicating
974 * IEEE80211 wireless events.
975 * XXX we piggyback on the RTM_IFANNOUNCE msg format in a clumsy way.
978 rt_ieee80211msg(struct ifnet *ifp, int what, void *data, size_t data_len)
981 struct rt_addrinfo info;
983 m = rt_makeifannouncemsg(ifp, RTM_IEEE80211, what, &info);
986 * Append the ieee80211 data. Try to stick it in the
987 * mbuf containing the ifannounce msg; otherwise allocate
988 * a new mbuf and append.
990 * NB: we assume m is a single mbuf.
992 if (data_len > M_TRAILINGSPACE(m)) {
993 struct mbuf *n = m_get(M_NOWAIT, MT_DATA);
998 bcopy(data, mtod(n, void *), data_len);
1001 } else if (data_len > 0) {
1002 bcopy(data, mtod(m, u_int8_t *) + m->m_len, data_len);
1003 m->m_len += data_len;
1005 if (m->m_flags & M_PKTHDR)
1006 m->m_pkthdr.len += data_len;
1007 mtod(m, struct if_announcemsghdr *)->ifan_msglen += data_len;
1008 rt_dispatch(m, NULL);
1013 * This is called to generate routing socket messages indicating
1014 * network interface arrival and departure.
1017 rt_ifannouncemsg(struct ifnet *ifp, int what)
1020 struct rt_addrinfo info;
1022 m = rt_makeifannouncemsg(ifp, RTM_IFANNOUNCE, what, &info);
1024 rt_dispatch(m, NULL);
1028 rt_dispatch(struct mbuf *m, const struct sockaddr *sa)
1033 * Preserve the family from the sockaddr, if any, in an m_tag for
1034 * use when injecting the mbuf into the routing socket buffer from
1038 tag = m_tag_get(PACKET_TAG_RTSOCKFAM, sizeof(unsigned short),
1044 *(unsigned short *)(tag + 1) = sa->sa_family;
1045 m_tag_prepend(m, tag);
1047 netisr_queue(NETISR_ROUTE, m); /* mbuf is free'd on failure. */
1051 * This is used in dumping the kernel table via sysctl().
1054 sysctl_dumpentry(struct radix_node *rn, void *vw)
1056 struct walkarg *w = vw;
1057 struct rtentry *rt = (struct rtentry *)rn;
1058 int error = 0, size;
1059 struct rt_addrinfo info;
1061 if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg))
1063 bzero((caddr_t)&info, sizeof(info));
1064 info.rti_info[RTAX_DST] = rt_key(rt);
1065 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
1066 info.rti_info[RTAX_NETMASK] = rt_mask(rt);
1067 info.rti_info[RTAX_GENMASK] = rt->rt_genmask;
1069 info.rti_info[RTAX_IFP] =
1070 ifaddr_byindex(rt->rt_ifp->if_index)->ifa_addr;
1071 info.rti_info[RTAX_IFA] = rt->rt_ifa->ifa_addr;
1072 if (rt->rt_ifp->if_flags & IFF_POINTOPOINT)
1073 info.rti_info[RTAX_BRD] = rt->rt_ifa->ifa_dstaddr;
1075 size = rt_msg2(RTM_GET, &info, NULL, w);
1076 if (w->w_req && w->w_tmem) {
1077 struct rt_msghdr *rtm = (struct rt_msghdr *)w->w_tmem;
1079 rtm->rtm_flags = rt->rt_flags;
1080 rtm->rtm_use = rt->rt_rmx.rmx_pksent;
1081 rt_getmetrics(&rt->rt_rmx, &rtm->rtm_rmx);
1082 rtm->rtm_index = rt->rt_ifp->if_index;
1083 rtm->rtm_errno = rtm->rtm_pid = rtm->rtm_seq = 0;
1084 rtm->rtm_addrs = info.rti_addrs;
1085 error = SYSCTL_OUT(w->w_req, (caddr_t)rtm, size);
1092 sysctl_iflist(int af, struct walkarg *w)
1096 struct rt_addrinfo info;
1099 bzero((caddr_t)&info, sizeof(info));
1101 TAILQ_FOREACH(ifp, &ifnet, if_link) {
1102 if (w->w_arg && w->w_arg != ifp->if_index)
1104 ifa = ifaddr_byindex(ifp->if_index);
1105 info.rti_info[RTAX_IFP] = ifa->ifa_addr;
1106 len = rt_msg2(RTM_IFINFO, &info, NULL, w);
1107 info.rti_info[RTAX_IFP] = NULL;
1108 if (w->w_req && w->w_tmem) {
1109 struct if_msghdr *ifm;
1111 ifm = (struct if_msghdr *)w->w_tmem;
1112 ifm->ifm_index = ifp->if_index;
1113 ifm->ifm_flags = ifp->if_flags | ifp->if_drv_flags;
1114 ifm->ifm_data = ifp->if_data;
1115 ifm->ifm_addrs = info.rti_addrs;
1116 error = SYSCTL_OUT(w->w_req,(caddr_t)ifm, len);
1120 while ((ifa = TAILQ_NEXT(ifa, ifa_link)) != NULL) {
1121 if (af && af != ifa->ifa_addr->sa_family)
1123 if (jailed(curthread->td_ucred) &&
1124 prison_if(curthread->td_ucred, ifa->ifa_addr))
1126 info.rti_info[RTAX_IFA] = ifa->ifa_addr;
1127 info.rti_info[RTAX_NETMASK] = ifa->ifa_netmask;
1128 info.rti_info[RTAX_BRD] = ifa->ifa_dstaddr;
1129 len = rt_msg2(RTM_NEWADDR, &info, NULL, w);
1130 if (w->w_req && w->w_tmem) {
1131 struct ifa_msghdr *ifam;
1133 ifam = (struct ifa_msghdr *)w->w_tmem;
1134 ifam->ifam_index = ifa->ifa_ifp->if_index;
1135 ifam->ifam_flags = ifa->ifa_flags;
1136 ifam->ifam_metric = ifa->ifa_metric;
1137 ifam->ifam_addrs = info.rti_addrs;
1138 error = SYSCTL_OUT(w->w_req, w->w_tmem, len);
1143 info.rti_info[RTAX_IFA] = info.rti_info[RTAX_NETMASK] =
1144 info.rti_info[RTAX_BRD] = NULL;
1152 sysctl_ifmalist(int af, struct walkarg *w)
1155 struct ifmultiaddr *ifma;
1156 struct rt_addrinfo info;
1160 bzero((caddr_t)&info, sizeof(info));
1162 TAILQ_FOREACH(ifp, &ifnet, if_link) {
1163 if (w->w_arg && w->w_arg != ifp->if_index)
1165 ifa = ifaddr_byindex(ifp->if_index);
1166 info.rti_info[RTAX_IFP] = ifa ? ifa->ifa_addr : NULL;
1168 TAILQ_FOREACH(ifma, &ifp->if_multiaddrs, ifma_link) {
1169 if (af && af != ifma->ifma_addr->sa_family)
1171 if (jailed(curproc->p_ucred) &&
1172 prison_if(curproc->p_ucred, ifma->ifma_addr))
1174 info.rti_info[RTAX_IFA] = ifma->ifma_addr;
1175 info.rti_info[RTAX_GATEWAY] =
1176 (ifma->ifma_addr->sa_family != AF_LINK) ?
1177 ifma->ifma_lladdr : NULL;
1178 len = rt_msg2(RTM_NEWMADDR, &info, NULL, w);
1179 if (w->w_req && w->w_tmem) {
1180 struct ifma_msghdr *ifmam;
1182 ifmam = (struct ifma_msghdr *)w->w_tmem;
1183 ifmam->ifmam_index = ifma->ifma_ifp->if_index;
1184 ifmam->ifmam_flags = 0;
1185 ifmam->ifmam_addrs = info.rti_addrs;
1186 error = SYSCTL_OUT(w->w_req, w->w_tmem, len);
1188 IF_ADDR_UNLOCK(ifp);
1193 IF_ADDR_UNLOCK(ifp);
1201 sysctl_rtsock(SYSCTL_HANDLER_ARGS)
1203 int *name = (int *)arg1;
1204 u_int namelen = arg2;
1205 struct radix_node_head *rnh;
1206 int i, lim, error = EINVAL;
1215 return ((namelen < 3) ? EISDIR : ENOTDIR);
1219 bzero(&w, sizeof(w));
1224 error = sysctl_wire_old_buffer(req, 0);
1231 if (af == 0) { /* dump all tables */
1234 } else /* dump only one table */
1236 for (error = 0; error == 0 && i <= lim; i++)
1237 if ((rnh = rt_tables[i]) != NULL) {
1238 RADIX_NODE_HEAD_LOCK(rnh);
1239 error = rnh->rnh_walktree(rnh,
1240 sysctl_dumpentry, &w);
1241 RADIX_NODE_HEAD_UNLOCK(rnh);
1243 error = EAFNOSUPPORT;
1247 error = sysctl_iflist(af, &w);
1250 case NET_RT_IFMALIST:
1251 error = sysctl_ifmalist(af, &w);
1255 free(w.w_tmem, M_RTABLE);
1259 SYSCTL_NODE(_net, PF_ROUTE, routetable, CTLFLAG_RD, sysctl_rtsock, "");
1262 * Definitions of protocols supported in the ROUTE domain.
1265 extern struct domain routedomain; /* or at least forward */
1267 static struct protosw routesw[] = {
1269 .pr_type = SOCK_RAW,
1270 .pr_domain = &routedomain,
1271 .pr_flags = PR_ATOMIC|PR_ADDR,
1272 .pr_output = route_output,
1273 .pr_ctlinput = raw_ctlinput,
1274 .pr_init = raw_init,
1275 .pr_usrreqs = &route_usrreqs
1279 static struct domain routedomain = {
1280 .dom_family = PF_ROUTE,
1281 .dom_name = "route",
1282 .dom_protosw = routesw,
1283 .dom_protoswNPROTOSW = &routesw[sizeof(routesw)/sizeof(routesw[0])]