2 * Copyright (c) 2004, 2005 The DragonFly Project. All rights reserved.
4 * This code is derived from software contributed to The DragonFly Project
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 DragonFly Project nor the names of its
16 * contributors may be used to endorse or promote products derived
17 * from this software without specific, prior written permission.
19 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
20 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
21 * LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS
22 * FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE
23 * COPYRIGHT HOLDERS OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT,
24 * INCIDENTAL, SPECIAL, EXEMPLARY OR CONSEQUENTIAL DAMAGES (INCLUDING,
25 * BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES;
26 * LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED
27 * AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY,
28 * OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT
29 * OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
34 * Copyright (c) 1988, 1991, 1993
35 * The Regents of the University of California. All rights reserved.
37 * Redistribution and use in source and binary forms, with or without
38 * modification, are permitted provided that the following conditions
40 * 1. Redistributions of source code must retain the above copyright
41 * notice, this list of conditions and the following disclaimer.
42 * 2. Redistributions in binary form must reproduce the above copyright
43 * notice, this list of conditions and the following disclaimer in the
44 * documentation and/or other materials provided with the distribution.
45 * 3. All advertising materials mentioning features or use of this software
46 * must display the following acknowledgement:
47 * This product includes software developed by the University of
48 * California, Berkeley and its contributors.
49 * 4. Neither the name of the University nor the names of its contributors
50 * may be used to endorse or promote products derived from this software
51 * without specific prior written permission.
53 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
54 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
55 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
56 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
57 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
58 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
59 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
60 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
61 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
62 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
65 * @(#)rtsock.c 8.7 (Berkeley) 10/12/95
66 * $FreeBSD: src/sys/net/rtsock.c,v 1.44.2.11 2002/12/04 14:05:41 ru Exp $
67 * $DragonFly: src/sys/net/rtsock.c,v 1.45 2008/10/27 02:56:30 sephe Exp $
72 #include <sys/param.h>
73 #include <sys/systm.h>
74 #include <sys/kernel.h>
75 #include <sys/sysctl.h>
78 #include <sys/malloc.h>
80 #include <sys/protosw.h>
81 #include <sys/socket.h>
82 #include <sys/socketvar.h>
83 #include <sys/domain.h>
84 #include <sys/thread2.h>
87 #include <net/route.h>
88 #include <net/raw_cb.h>
89 #include <net/netmsg2.h>
92 extern void sctp_add_ip_address(struct ifaddr *ifa);
93 extern void sctp_delete_ip_address(struct ifaddr *ifa);
96 MALLOC_DEFINE(M_RTABLE, "routetbl", "routing tables");
98 static struct route_cb {
106 static const struct sockaddr route_src = { 2, PF_ROUTE, };
112 struct sysctl_req *w_req;
116 rt_msg_mbuf (int, struct rt_addrinfo *);
117 static void rt_msg_buffer (int, struct rt_addrinfo *, void *buf, int len);
118 static int rt_msgsize (int type, struct rt_addrinfo *rtinfo);
119 static int rt_xaddrs (char *, char *, struct rt_addrinfo *);
120 static int sysctl_dumpentry (struct radix_node *rn, void *vw);
121 static int sysctl_iflist (int af, struct walkarg *w);
122 static int route_output(struct mbuf *, struct socket *, ...);
123 static void rt_setmetrics (u_long, struct rt_metrics *,
124 struct rt_metrics *);
127 * It really doesn't make any sense at all for this code to share much
128 * with raw_usrreq.c, since its functionality is so restricted. XXX
131 rts_abort(struct socket *so)
136 error = raw_usrreqs.pru_abort(so);
141 /* pru_accept is EOPNOTSUPP */
144 rts_attach(struct socket *so, int proto, struct pru_attach_info *ai)
149 if (sotorawcb(so) != NULL)
150 return EISCONN; /* XXX panic? */
152 rp = kmalloc(sizeof *rp, M_PCB, M_WAITOK | M_ZERO);
155 * The critical section is necessary to block protocols from sending
156 * error notifications (like RTM_REDIRECT or RTM_LOSING) while
157 * this PCB is extant but incompletely initialized.
158 * Probably we should try to do more of this work beforehand and
159 * eliminate the critical section.
163 error = raw_attach(so, proto, ai->sb_rlimit);
170 switch(rp->rcb_proto.sp_protocol) {
175 route_cb.ip6_count++;
178 route_cb.ipx_count++;
184 rp->rcb_faddr = &route_src;
185 route_cb.any_count++;
187 so->so_options |= SO_USELOOPBACK;
193 rts_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
198 error = raw_usrreqs.pru_bind(so, nam, td); /* xxx just EINVAL */
204 rts_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
209 error = 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);
225 switch(rp->rcb_proto.sp_protocol) {
230 route_cb.ip6_count--;
233 route_cb.ipx_count--;
239 route_cb.any_count--;
241 error = raw_usrreqs.pru_detach(so);
247 rts_disconnect(struct socket *so)
252 error = raw_usrreqs.pru_disconnect(so);
257 /* pru_listen is EOPNOTSUPP */
260 rts_peeraddr(struct socket *so, struct sockaddr **nam)
265 error = raw_usrreqs.pru_peeraddr(so, nam);
270 /* pru_rcvd is EOPNOTSUPP */
271 /* pru_rcvoob is EOPNOTSUPP */
274 rts_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
275 struct mbuf *control, struct thread *td)
280 error = raw_usrreqs.pru_send(so, flags, m, nam, control, td);
285 /* pru_sense is null */
288 rts_shutdown(struct socket *so)
293 error = raw_usrreqs.pru_shutdown(so);
299 rts_sockaddr(struct socket *so, struct sockaddr **nam)
304 error = raw_usrreqs.pru_sockaddr(so, nam);
309 static struct pr_usrreqs route_usrreqs = {
310 .pru_abort = rts_abort,
311 .pru_accept = pru_accept_notsupp,
312 .pru_attach = rts_attach,
313 .pru_bind = rts_bind,
314 .pru_connect = rts_connect,
315 .pru_connect2 = pru_connect2_notsupp,
316 .pru_control = pru_control_notsupp,
317 .pru_detach = rts_detach,
318 .pru_disconnect = rts_disconnect,
319 .pru_listen = pru_listen_notsupp,
320 .pru_peeraddr = rts_peeraddr,
321 .pru_rcvd = pru_rcvd_notsupp,
322 .pru_rcvoob = pru_rcvoob_notsupp,
323 .pru_send = rts_send,
324 .pru_sense = pru_sense_null,
325 .pru_shutdown = rts_shutdown,
326 .pru_sockaddr = rts_sockaddr,
327 .pru_sosend = sosend,
328 .pru_soreceive = soreceive,
332 static __inline sa_family_t
333 familyof(struct sockaddr *sa)
335 return (sa != NULL ? sa->sa_family : 0);
339 * Routing socket input function. The packet must be serialized onto cpu 0.
340 * We use the cpu0_soport() netisr processing loop to handle it.
342 * This looks messy but it means that anyone, including interrupt code,
343 * can send a message to the routing socket.
346 rts_input_handler(struct netmsg *msg)
348 static const struct sockaddr route_dst = { 2, PF_ROUTE, };
349 struct sockproto route_proto;
350 struct netmsg_packet *pmsg;
356 family = pmsg->nm_netmsg.nm_lmsg.u.ms_result;
357 route_proto.sp_family = PF_ROUTE;
358 route_proto.sp_protocol = family;
363 skip = m->m_pkthdr.header;
364 m->m_pkthdr.header = NULL;
366 raw_input(m, &route_proto, &route_src, &route_dst, skip);
370 rts_input_skip(struct mbuf *m, sa_family_t family, struct rawcb *skip)
372 struct netmsg_packet *pmsg;
377 port = cpu0_soport(NULL, NULL, NULL, 0);
378 pmsg = &m->m_hdr.mh_netmsg;
379 netmsg_init(&pmsg->nm_netmsg, &netisr_apanic_rport,
380 0, rts_input_handler);
382 pmsg->nm_netmsg.nm_lmsg.u.ms_result = family;
383 m->m_pkthdr.header = skip; /* XXX steal field in pkthdr */
384 lwkt_sendmsg(port, &pmsg->nm_netmsg.nm_lmsg);
388 rts_input(struct mbuf *m, sa_family_t family)
390 rts_input_skip(m, family, NULL);
394 reallocbuf(void *ptr, size_t len, size_t olen)
398 newptr = kmalloc(len, M_RTABLE, M_INTWAIT | M_NULLOK);
401 bcopy(ptr, newptr, olen);
402 kfree(ptr, M_RTABLE);
407 * Internal helper routine for route_output().
410 fillrtmsg(struct rt_msghdr **prtm, struct rtentry *rt,
411 struct rt_addrinfo *rtinfo)
414 struct rt_msghdr *rtm = *prtm;
416 /* Fill in rt_addrinfo for call to rt_msg_buffer(). */
417 rtinfo->rti_dst = rt_key(rt);
418 rtinfo->rti_gateway = rt->rt_gateway;
419 rtinfo->rti_netmask = rt_mask(rt); /* might be NULL */
420 rtinfo->rti_genmask = rt->rt_genmask; /* might be NULL */
421 if (rtm->rtm_addrs & (RTA_IFP | RTA_IFA)) {
422 if (rt->rt_ifp != NULL) {
423 rtinfo->rti_ifpaddr =
424 TAILQ_FIRST(&rt->rt_ifp->if_addrheads[mycpuid])
426 rtinfo->rti_ifaaddr = rt->rt_ifa->ifa_addr;
427 if (rt->rt_ifp->if_flags & IFF_POINTOPOINT)
428 rtinfo->rti_bcastaddr = rt->rt_ifa->ifa_dstaddr;
429 rtm->rtm_index = rt->rt_ifp->if_index;
431 rtinfo->rti_ifpaddr = NULL;
432 rtinfo->rti_ifaaddr = NULL;
434 } else if (rt->rt_ifp != NULL) {
435 rtm->rtm_index = rt->rt_ifp->if_index;
438 msglen = rt_msgsize(rtm->rtm_type, rtinfo);
439 if (rtm->rtm_msglen < msglen) {
440 rtm = reallocbuf(rtm, msglen, rtm->rtm_msglen);
445 rt_msg_buffer(rtm->rtm_type, rtinfo, rtm, msglen);
447 rtm->rtm_flags = rt->rt_flags;
448 rtm->rtm_rmx = rt->rt_rmx;
449 rtm->rtm_addrs = rtinfo->rti_addrs;
454 static void route_output_add_callback(int, int, struct rt_addrinfo *,
455 struct rtentry *, void *);
456 static void route_output_delete_callback(int, int, struct rt_addrinfo *,
457 struct rtentry *, void *);
458 static int route_output_get_callback(int, struct rt_addrinfo *,
459 struct rtentry *, void *);
460 static int route_output_change_callback(int, struct rt_addrinfo *,
461 struct rtentry *, void *);
462 static int route_output_lock_callback(int, struct rt_addrinfo *,
463 struct rtentry *, void *);
467 route_output(struct mbuf *m, struct socket *so, ...)
469 struct rt_msghdr *rtm = NULL;
470 struct rawcb *rp = NULL;
471 struct pr_output_info *oi;
472 struct rt_addrinfo rtinfo;
477 oi = __va_arg(ap, struct pr_output_info *);
480 #define gotoerr(e) { error = e; goto flush;}
483 (m->m_len < sizeof(long) &&
484 (m = m_pullup(m, sizeof(long))) == NULL))
486 if (!(m->m_flags & M_PKTHDR))
487 panic("route_output");
488 len = m->m_pkthdr.len;
489 if (len < sizeof(struct rt_msghdr) ||
490 len != mtod(m, struct rt_msghdr *)->rtm_msglen) {
491 rtinfo.rti_dst = NULL;
494 rtm = kmalloc(len, M_RTABLE, M_INTWAIT | M_NULLOK);
496 rtinfo.rti_dst = NULL;
499 m_copydata(m, 0, len, (caddr_t)rtm);
500 if (rtm->rtm_version != RTM_VERSION) {
501 rtinfo.rti_dst = NULL;
502 gotoerr(EPROTONOSUPPORT);
504 rtm->rtm_pid = oi->p_pid;
505 bzero(&rtinfo, sizeof(struct rt_addrinfo));
506 rtinfo.rti_addrs = rtm->rtm_addrs;
507 if (rt_xaddrs((char *)(rtm + 1), (char *)rtm + len, &rtinfo) != 0) {
508 rtinfo.rti_dst = NULL;
511 rtinfo.rti_flags = rtm->rtm_flags;
512 if (rtinfo.rti_dst == NULL || rtinfo.rti_dst->sa_family >= AF_MAX ||
513 (rtinfo.rti_gateway && rtinfo.rti_gateway->sa_family >= AF_MAX))
516 if (rtinfo.rti_genmask != NULL) {
517 struct radix_node *n;
519 #define clen(s) (*(u_char *)(s))
520 n = rn_addmask((char *)rtinfo.rti_genmask, TRUE, 1);
522 rtinfo.rti_genmask->sa_len >= clen(n->rn_key) &&
523 bcmp((char *)rtinfo.rti_genmask + 1,
524 (char *)n->rn_key + 1, clen(n->rn_key) - 1) == 0)
525 rtinfo.rti_genmask = (struct sockaddr *)n->rn_key;
531 * Verify that the caller has the appropriate privilege; RTM_GET
532 * is the only operation the non-superuser is allowed.
534 if (rtm->rtm_type != RTM_GET && priv_check_cred(so->so_cred, PRIV_ROOT, 0) != 0)
537 switch (rtm->rtm_type) {
539 if (rtinfo.rti_gateway == NULL) {
542 error = rtrequest1_global(RTM_ADD, &rtinfo,
543 route_output_add_callback, rtm);
548 * note: &rtm passed as argument so 'rtm' can be replaced.
550 error = rtrequest1_global(RTM_DELETE, &rtinfo,
551 route_output_delete_callback, &rtm);
555 * note: &rtm passed as argument so 'rtm' can be replaced.
557 error = rtsearch_global(RTM_GET, &rtinfo,
558 route_output_get_callback, &rtm,
562 error = rtsearch_global(RTM_CHANGE, &rtinfo,
563 route_output_change_callback, rtm,
567 error = rtsearch_global(RTM_LOCK, &rtinfo,
568 route_output_lock_callback, rtm,
579 rtm->rtm_errno = error;
581 rtm->rtm_flags |= RTF_DONE;
585 * Check to see if we don't want our own messages.
587 if (!(so->so_options & SO_USELOOPBACK)) {
588 if (route_cb.any_count <= 1) {
590 kfree(rtm, M_RTABLE);
594 /* There is another listener, so construct message */
598 m_copyback(m, 0, rtm->rtm_msglen, (caddr_t)rtm);
599 if (m->m_pkthdr.len < rtm->rtm_msglen) {
602 } else if (m->m_pkthdr.len > rtm->rtm_msglen)
603 m_adj(m, rtm->rtm_msglen - m->m_pkthdr.len);
604 kfree(rtm, M_RTABLE);
607 rts_input_skip(m, familyof(rtinfo.rti_dst), rp);
612 route_output_add_callback(int cmd, int error, struct rt_addrinfo *rtinfo,
613 struct rtentry *rt, void *arg)
615 struct rt_msghdr *rtm = arg;
617 if (error == 0 && rt != NULL) {
618 rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx,
620 rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits);
621 rt->rt_rmx.rmx_locks |=
622 (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks);
623 rt->rt_genmask = rtinfo->rti_genmask;
628 route_output_delete_callback(int cmd, int error, struct rt_addrinfo *rtinfo,
629 struct rtentry *rt, void *arg)
631 struct rt_msghdr **rtm = arg;
633 if (error == 0 && rt) {
635 if (fillrtmsg(rtm, rt, rtinfo) != 0) {
637 /* XXX no way to return the error */
644 route_output_get_callback(int cmd, struct rt_addrinfo *rtinfo,
645 struct rtentry *rt, void *arg)
647 struct rt_msghdr **rtm = arg;
648 int error, found = 0;
650 if (((rtinfo->rti_flags ^ rt->rt_flags) & RTF_HOST) == 0)
653 error = fillrtmsg(rtm, rt, rtinfo);
654 if (!error && found) {
655 /* Got the exact match, we could return now! */
662 route_output_change_callback(int cmd, struct rt_addrinfo *rtinfo,
663 struct rtentry *rt, void *arg)
665 struct rt_msghdr *rtm = arg;
670 * new gateway could require new ifaddr, ifp;
671 * flags may also be different; ifp may be specified
672 * by ll sockaddr when protocol address is ambiguous
674 if (((rt->rt_flags & RTF_GATEWAY) && rtinfo->rti_gateway != NULL) ||
675 rtinfo->rti_ifpaddr != NULL ||
676 (rtinfo->rti_ifaaddr != NULL &&
677 !sa_equal(rtinfo->rti_ifaaddr, rt->rt_ifa->ifa_addr))) {
678 error = rt_getifa(rtinfo);
682 if (rtinfo->rti_gateway != NULL) {
683 error = rt_setgate(rt, rt_key(rt), rtinfo->rti_gateway);
687 if ((ifa = rtinfo->rti_ifa) != NULL) {
688 struct ifaddr *oifa = rt->rt_ifa;
691 if (oifa && oifa->ifa_rtrequest)
692 oifa->ifa_rtrequest(RTM_DELETE, rt, rtinfo);
696 rt->rt_ifp = rtinfo->rti_ifp;
699 rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx, &rt->rt_rmx);
700 if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest)
701 rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, rtinfo);
702 if (rtinfo->rti_genmask != NULL)
703 rt->rt_genmask = rtinfo->rti_genmask;
709 route_output_lock_callback(int cmd, struct rt_addrinfo *rtinfo,
710 struct rtentry *rt, void *arg)
712 struct rt_msghdr *rtm = arg;
714 rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits);
715 rt->rt_rmx.rmx_locks |=
716 (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks);
721 rt_setmetrics(u_long which, struct rt_metrics *in, struct rt_metrics *out)
723 #define setmetric(flag, elt) if (which & (flag)) out->elt = in->elt;
724 setmetric(RTV_RPIPE, rmx_recvpipe);
725 setmetric(RTV_SPIPE, rmx_sendpipe);
726 setmetric(RTV_SSTHRESH, rmx_ssthresh);
727 setmetric(RTV_RTT, rmx_rtt);
728 setmetric(RTV_RTTVAR, rmx_rttvar);
729 setmetric(RTV_HOPCOUNT, rmx_hopcount);
730 setmetric(RTV_MTU, rmx_mtu);
731 setmetric(RTV_EXPIRE, rmx_expire);
736 ((a) > 0 ? (1 + (((a) - 1) | (sizeof(long) - 1))) : sizeof(long))
739 * Extract the addresses of the passed sockaddrs.
740 * Do a little sanity checking so as to avoid bad memory references.
741 * This data is derived straight from userland.
744 rt_xaddrs(char *cp, char *cplim, struct rt_addrinfo *rtinfo)
749 for (i = 0; (i < RTAX_MAX) && (cp < cplim); i++) {
750 if ((rtinfo->rti_addrs & (1 << i)) == 0)
752 sa = (struct sockaddr *)cp;
756 if ((cp + sa->sa_len) > cplim) {
761 * There are no more... Quit now.
762 * If there are more bits, they are in error.
763 * I've seen this. route(1) can evidently generate these.
764 * This causes kernel to core dump.
765 * For compatibility, if we see this, point to a safe address.
767 if (sa->sa_len == 0) {
768 static struct sockaddr sa_zero = {
769 sizeof sa_zero, AF_INET,
772 rtinfo->rti_info[i] = &sa_zero;
773 kprintf("rtsock: received more addr bits than sockaddrs.\n");
774 return (0); /* should be EINVAL but for compat */
777 /* Accept the sockaddr. */
778 rtinfo->rti_info[i] = sa;
779 cp += ROUNDUP(sa->sa_len);
785 rt_msghdrsize(int type)
790 return sizeof(struct ifa_msghdr);
793 return sizeof(struct ifma_msghdr);
795 return sizeof(struct if_msghdr);
798 return sizeof(struct if_announcemsghdr);
800 return sizeof(struct rt_msghdr);
805 rt_msgsize(int type, struct rt_addrinfo *rtinfo)
809 len = rt_msghdrsize(type);
810 for (i = 0; i < RTAX_MAX; i++) {
811 if (rtinfo->rti_info[i] != NULL)
812 len += ROUNDUP(rtinfo->rti_info[i]->sa_len);
819 * Build a routing message in a buffer.
820 * Copy the addresses in the rtinfo->rti_info[] sockaddr array
821 * to the end of the buffer after the message header.
823 * Set the rtinfo->rti_addrs bitmask of addresses present in rtinfo->rti_info[].
824 * This side-effect can be avoided if we reorder the addrs bitmask field in all
825 * the route messages to line up so we can set it here instead of back in the
829 rt_msg_buffer(int type, struct rt_addrinfo *rtinfo, void *buf, int msglen)
831 struct rt_msghdr *rtm;
835 rtm = (struct rt_msghdr *) buf;
836 rtm->rtm_version = RTM_VERSION;
837 rtm->rtm_type = type;
838 rtm->rtm_msglen = msglen;
840 cp = (char *)buf + rt_msghdrsize(type);
841 rtinfo->rti_addrs = 0;
842 for (i = 0; i < RTAX_MAX; i++) {
845 if ((sa = rtinfo->rti_info[i]) == NULL)
847 rtinfo->rti_addrs |= (1 << i);
848 dlen = ROUNDUP(sa->sa_len);
855 * Build a routing message in a mbuf chain.
856 * Copy the addresses in the rtinfo->rti_info[] sockaddr array
857 * to the end of the mbuf after the message header.
859 * Set the rtinfo->rti_addrs bitmask of addresses present in rtinfo->rti_info[].
860 * This side-effect can be avoided if we reorder the addrs bitmask field in all
861 * the route messages to line up so we can set it here instead of back in the
865 rt_msg_mbuf(int type, struct rt_addrinfo *rtinfo)
868 struct rt_msghdr *rtm;
872 hlen = rt_msghdrsize(type);
873 KASSERT(hlen <= MCLBYTES, ("rt_msg_mbuf: hlen %d doesn't fit", hlen));
875 m = m_getl(hlen, MB_DONTWAIT, MT_DATA, M_PKTHDR, NULL);
879 m->m_pkthdr.len = m->m_len = hlen;
880 m->m_pkthdr.rcvif = NULL;
881 rtinfo->rti_addrs = 0;
883 for (i = 0; i < RTAX_MAX; i++) {
887 if ((sa = rtinfo->rti_info[i]) == NULL)
889 rtinfo->rti_addrs |= (1 << i);
890 dlen = ROUNDUP(sa->sa_len);
891 m_copyback(m, len, dlen, (caddr_t)sa); /* can grow mbuf chain */
894 if (m->m_pkthdr.len != len) { /* one of the m_copyback() calls failed */
898 rtm = mtod(m, struct rt_msghdr *);
900 rtm->rtm_msglen = len;
901 rtm->rtm_version = RTM_VERSION;
902 rtm->rtm_type = type;
907 * This routine is called to generate a message from the routing
908 * socket indicating that a redirect has occurred, a routing lookup
909 * has failed, or that a protocol has detected timeouts to a particular
913 rt_missmsg(int type, struct rt_addrinfo *rtinfo, int flags, int error)
915 struct sockaddr *dst = rtinfo->rti_info[RTAX_DST];
916 struct rt_msghdr *rtm;
919 if (route_cb.any_count == 0)
921 m = rt_msg_mbuf(type, rtinfo);
924 rtm = mtod(m, struct rt_msghdr *);
925 rtm->rtm_flags = RTF_DONE | flags;
926 rtm->rtm_errno = error;
927 rtm->rtm_addrs = rtinfo->rti_addrs;
928 rts_input(m, familyof(dst));
932 rt_dstmsg(int type, struct sockaddr *dst, int error)
934 struct rt_msghdr *rtm;
935 struct rt_addrinfo addrs;
938 if (route_cb.any_count == 0)
940 bzero(&addrs, sizeof(struct rt_addrinfo));
941 addrs.rti_info[RTAX_DST] = dst;
942 m = rt_msg_mbuf(type, &addrs);
945 rtm = mtod(m, struct rt_msghdr *);
946 rtm->rtm_flags = RTF_DONE;
947 rtm->rtm_errno = error;
948 rtm->rtm_addrs = addrs.rti_addrs;
949 rts_input(m, familyof(dst));
953 * This routine is called to generate a message from the routing
954 * socket indicating that the status of a network interface has changed.
957 rt_ifmsg(struct ifnet *ifp)
959 struct if_msghdr *ifm;
961 struct rt_addrinfo rtinfo;
963 if (route_cb.any_count == 0)
965 bzero(&rtinfo, sizeof(struct rt_addrinfo));
966 m = rt_msg_mbuf(RTM_IFINFO, &rtinfo);
969 ifm = mtod(m, struct if_msghdr *);
970 ifm->ifm_index = ifp->if_index;
971 ifm->ifm_flags = ifp->if_flags;
972 ifm->ifm_data = ifp->if_data;
978 rt_ifamsg(int cmd, struct ifaddr *ifa)
980 struct ifa_msghdr *ifam;
981 struct rt_addrinfo rtinfo;
983 struct ifnet *ifp = ifa->ifa_ifp;
985 bzero(&rtinfo, sizeof(struct rt_addrinfo));
986 rtinfo.rti_ifaaddr = ifa->ifa_addr;
988 TAILQ_FIRST(&ifp->if_addrheads[mycpuid])->ifa->ifa_addr;
989 rtinfo.rti_netmask = ifa->ifa_netmask;
990 rtinfo.rti_bcastaddr = ifa->ifa_dstaddr;
992 m = rt_msg_mbuf(cmd, &rtinfo);
996 ifam = mtod(m, struct ifa_msghdr *);
997 ifam->ifam_index = ifp->if_index;
998 ifam->ifam_metric = ifa->ifa_metric;
999 ifam->ifam_flags = ifa->ifa_flags;
1000 ifam->ifam_addrs = rtinfo.rti_addrs;
1002 rts_input(m, familyof(ifa->ifa_addr));
1006 rt_rtmsg(int cmd, struct rtentry *rt, struct ifnet *ifp, int error)
1008 struct rt_msghdr *rtm;
1009 struct rt_addrinfo rtinfo;
1011 struct sockaddr *dst;
1016 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1017 rtinfo.rti_dst = dst = rt_key(rt);
1018 rtinfo.rti_gateway = rt->rt_gateway;
1019 rtinfo.rti_netmask = rt_mask(rt);
1021 rtinfo.rti_ifpaddr =
1022 TAILQ_FIRST(&ifp->if_addrheads[mycpuid])->ifa->ifa_addr;
1024 rtinfo.rti_ifaaddr = rt->rt_ifa->ifa_addr;
1026 m = rt_msg_mbuf(cmd, &rtinfo);
1030 rtm = mtod(m, struct rt_msghdr *);
1032 rtm->rtm_index = ifp->if_index;
1033 rtm->rtm_flags |= rt->rt_flags;
1034 rtm->rtm_errno = error;
1035 rtm->rtm_addrs = rtinfo.rti_addrs;
1037 rts_input(m, familyof(dst));
1041 * This is called to generate messages from the routing socket
1042 * indicating a network interface has had addresses associated with it.
1043 * if we ever reverse the logic and replace messages TO the routing
1044 * socket indicate a request to configure interfaces, then it will
1045 * be unnecessary as the routing socket will automatically generate
1049 rt_newaddrmsg(int cmd, struct ifaddr *ifa, int error, struct rtentry *rt)
1053 * notify the SCTP stack
1054 * this will only get called when an address is added/deleted
1055 * XXX pass the ifaddr struct instead if ifa->ifa_addr...
1058 sctp_add_ip_address(ifa);
1059 else if (cmd == RTM_DELETE)
1060 sctp_delete_ip_address(ifa);
1063 if (route_cb.any_count == 0)
1066 if (cmd == RTM_ADD) {
1067 rt_ifamsg(RTM_NEWADDR, ifa);
1068 rt_rtmsg(RTM_ADD, rt, ifa->ifa_ifp, error);
1070 KASSERT((cmd == RTM_DELETE), ("unknown cmd %d", cmd));
1071 rt_rtmsg(RTM_DELETE, rt, ifa->ifa_ifp, error);
1072 rt_ifamsg(RTM_DELADDR, ifa);
1077 * This is the analogue to the rt_newaddrmsg which performs the same
1078 * function but for multicast group memberhips. This is easier since
1079 * there is no route state to worry about.
1082 rt_newmaddrmsg(int cmd, struct ifmultiaddr *ifma)
1084 struct rt_addrinfo rtinfo;
1085 struct mbuf *m = NULL;
1086 struct ifnet *ifp = ifma->ifma_ifp;
1087 struct ifma_msghdr *ifmam;
1089 if (route_cb.any_count == 0)
1092 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1093 rtinfo.rti_ifaaddr = ifma->ifma_addr;
1094 if (ifp != NULL && !TAILQ_EMPTY(&ifp->if_addrheads[mycpuid])) {
1095 rtinfo.rti_ifpaddr =
1096 TAILQ_FIRST(&ifp->if_addrheads[mycpuid])->ifa->ifa_addr;
1099 * If a link-layer address is present, present it as a ``gateway''
1100 * (similarly to how ARP entries, e.g., are presented).
1102 rtinfo.rti_gateway = ifma->ifma_lladdr;
1104 m = rt_msg_mbuf(cmd, &rtinfo);
1108 ifmam = mtod(m, struct ifma_msghdr *);
1109 ifmam->ifmam_index = ifp->if_index;
1110 ifmam->ifmam_addrs = rtinfo.rti_addrs;
1112 rts_input(m, familyof(ifma->ifma_addr));
1115 static struct mbuf *
1116 rt_makeifannouncemsg(struct ifnet *ifp, int type, int what,
1117 struct rt_addrinfo *info)
1119 struct if_announcemsghdr *ifan;
1122 if (route_cb.any_count == 0)
1125 bzero(info, sizeof(*info));
1126 m = rt_msg_mbuf(type, info);
1130 ifan = mtod(m, struct if_announcemsghdr *);
1131 ifan->ifan_index = ifp->if_index;
1132 strlcpy(ifan->ifan_name, ifp->if_xname, sizeof ifan->ifan_name);
1133 ifan->ifan_what = what;
1138 * This is called to generate routing socket messages indicating
1139 * IEEE80211 wireless events.
1140 * XXX we piggyback on the RTM_IFANNOUNCE msg format in a clumsy way.
1143 rt_ieee80211msg(struct ifnet *ifp, int what, void *data, size_t data_len)
1145 struct rt_addrinfo info;
1148 m = rt_makeifannouncemsg(ifp, RTM_IEEE80211, what, &info);
1153 * Append the ieee80211 data. Try to stick it in the
1154 * mbuf containing the ifannounce msg; otherwise allocate
1155 * a new mbuf and append.
1157 * NB: we assume m is a single mbuf.
1159 if (data_len > M_TRAILINGSPACE(m)) {
1160 struct mbuf *n = m_get(MB_DONTWAIT, MT_DATA);
1165 bcopy(data, mtod(n, void *), data_len);
1166 n->m_len = data_len;
1168 } else if (data_len > 0) {
1169 bcopy(data, mtod(m, u_int8_t *) + m->m_len, data_len);
1170 m->m_len += data_len;
1173 if (m->m_flags & M_PKTHDR)
1174 m->m_pkthdr.len += data_len;
1175 mtod(m, struct if_announcemsghdr *)->ifan_msglen += data_len;
1180 * This is called to generate routing socket messages indicating
1181 * network interface arrival and departure.
1184 rt_ifannouncemsg(struct ifnet *ifp, int what)
1186 struct rt_addrinfo addrinfo;
1189 m = rt_makeifannouncemsg(ifp, RTM_IFANNOUNCE, what, &addrinfo);
1195 resizewalkarg(struct walkarg *w, int len)
1199 newptr = kmalloc(len, M_RTABLE, M_INTWAIT | M_NULLOK);
1202 if (w->w_tmem != NULL)
1203 kfree(w->w_tmem, M_RTABLE);
1205 w->w_tmemsize = len;
1210 * This is used in dumping the kernel table via sysctl().
1213 sysctl_dumpentry(struct radix_node *rn, void *vw)
1215 struct walkarg *w = vw;
1216 struct rtentry *rt = (struct rtentry *)rn;
1217 struct rt_addrinfo rtinfo;
1220 if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg))
1223 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1224 rtinfo.rti_dst = rt_key(rt);
1225 rtinfo.rti_gateway = rt->rt_gateway;
1226 rtinfo.rti_netmask = rt_mask(rt);
1227 rtinfo.rti_genmask = rt->rt_genmask;
1228 if (rt->rt_ifp != NULL) {
1229 rtinfo.rti_ifpaddr =
1230 TAILQ_FIRST(&rt->rt_ifp->if_addrheads[mycpuid])->ifa->ifa_addr;
1231 rtinfo.rti_ifaaddr = rt->rt_ifa->ifa_addr;
1232 if (rt->rt_ifp->if_flags & IFF_POINTOPOINT)
1233 rtinfo.rti_bcastaddr = rt->rt_ifa->ifa_dstaddr;
1235 msglen = rt_msgsize(RTM_GET, &rtinfo);
1236 if (w->w_tmemsize < msglen && resizewalkarg(w, msglen) != 0)
1238 rt_msg_buffer(RTM_GET, &rtinfo, w->w_tmem, msglen);
1239 if (w->w_req != NULL) {
1240 struct rt_msghdr *rtm = w->w_tmem;
1242 rtm->rtm_flags = rt->rt_flags;
1243 rtm->rtm_use = rt->rt_use;
1244 rtm->rtm_rmx = rt->rt_rmx;
1245 rtm->rtm_index = rt->rt_ifp->if_index;
1246 rtm->rtm_errno = rtm->rtm_pid = rtm->rtm_seq = 0;
1247 rtm->rtm_addrs = rtinfo.rti_addrs;
1248 error = SYSCTL_OUT(w->w_req, rtm, msglen);
1255 sysctl_iflist(int af, struct walkarg *w)
1258 struct rt_addrinfo rtinfo;
1261 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1262 TAILQ_FOREACH(ifp, &ifnet, if_link) {
1263 struct ifaddr_container *ifac;
1266 if (w->w_arg && w->w_arg != ifp->if_index)
1268 ifac = TAILQ_FIRST(&ifp->if_addrheads[mycpuid]);
1270 rtinfo.rti_ifpaddr = ifa->ifa_addr;
1271 msglen = rt_msgsize(RTM_IFINFO, &rtinfo);
1272 if (w->w_tmemsize < msglen && resizewalkarg(w, msglen) != 0)
1274 rt_msg_buffer(RTM_IFINFO, &rtinfo, w->w_tmem, msglen);
1275 rtinfo.rti_ifpaddr = NULL;
1276 if (w->w_req != NULL && w->w_tmem != NULL) {
1277 struct if_msghdr *ifm = w->w_tmem;
1279 ifm->ifm_index = ifp->if_index;
1280 ifm->ifm_flags = ifp->if_flags;
1281 ifm->ifm_data = ifp->if_data;
1282 ifm->ifm_addrs = rtinfo.rti_addrs;
1283 error = SYSCTL_OUT(w->w_req, ifm, msglen);
1287 while ((ifac = TAILQ_NEXT(ifac, ifa_link)) != NULL) {
1290 if (af && af != ifa->ifa_addr->sa_family)
1292 if (curproc->p_ucred->cr_prison &&
1293 prison_if(curproc->p_ucred, ifa->ifa_addr))
1295 rtinfo.rti_ifaaddr = ifa->ifa_addr;
1296 rtinfo.rti_netmask = ifa->ifa_netmask;
1297 rtinfo.rti_bcastaddr = ifa->ifa_dstaddr;
1298 msglen = rt_msgsize(RTM_NEWADDR, &rtinfo);
1299 if (w->w_tmemsize < msglen &&
1300 resizewalkarg(w, msglen) != 0)
1302 rt_msg_buffer(RTM_NEWADDR, &rtinfo, w->w_tmem, msglen);
1303 if (w->w_req != NULL) {
1304 struct ifa_msghdr *ifam = w->w_tmem;
1306 ifam->ifam_index = ifa->ifa_ifp->if_index;
1307 ifam->ifam_flags = ifa->ifa_flags;
1308 ifam->ifam_metric = ifa->ifa_metric;
1309 ifam->ifam_addrs = rtinfo.rti_addrs;
1310 error = SYSCTL_OUT(w->w_req, w->w_tmem, msglen);
1315 rtinfo.rti_netmask = NULL;
1316 rtinfo.rti_ifaaddr = NULL;
1317 rtinfo.rti_bcastaddr = NULL;
1323 sysctl_rtsock(SYSCTL_HANDLER_ARGS)
1325 int *name = (int *)arg1;
1326 u_int namelen = arg2;
1327 struct radix_node_head *rnh;
1328 int i, error = EINVAL;
1337 if (namelen != 3 && namelen != 4)
1340 bzero(&w, sizeof w);
1346 * Optional third argument specifies cpu, used primarily for
1347 * debugging the route table.
1350 if (name[3] < 0 || name[3] >= ncpus)
1353 lwkt_migratecpu(name[3]);
1361 for (i = 1; i <= AF_MAX; i++)
1362 if ((rnh = rt_tables[mycpuid][i]) &&
1363 (af == 0 || af == i) &&
1364 (error = rnh->rnh_walktree(rnh,
1365 sysctl_dumpentry, &w)))
1370 error = sysctl_iflist(af, &w);
1373 if (w.w_tmem != NULL)
1374 kfree(w.w_tmem, M_RTABLE);
1376 lwkt_migratecpu(origcpu);
1380 SYSCTL_NODE(_net, PF_ROUTE, routetable, CTLFLAG_RD, sysctl_rtsock, "");
1383 * Definitions of protocols supported in the ROUTE domain.
1386 static struct domain routedomain; /* or at least forward */
1388 static struct protosw routesw[] = {
1389 { SOCK_RAW, &routedomain, 0, PR_ATOMIC|PR_ADDR,
1390 0, route_output, raw_ctlinput, 0,
1391 cpu0_soport, cpu0_ctlport,
1397 static struct domain routedomain = {
1398 PF_ROUTE, "route", NULL, NULL, NULL,
1399 routesw, &routesw[(sizeof routesw)/(sizeof routesw[0])],