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
331 static __inline sa_family_t
332 familyof(struct sockaddr *sa)
334 return (sa != NULL ? sa->sa_family : 0);
338 * Routing socket input function. The packet must be serialized onto cpu 0.
339 * We use the cpu0_soport() netisr processing loop to handle it.
341 * This looks messy but it means that anyone, including interrupt code,
342 * can send a message to the routing socket.
345 rts_input_handler(struct netmsg *msg)
347 static const struct sockaddr route_dst = { 2, PF_ROUTE, };
348 struct sockproto route_proto;
349 struct netmsg_packet *pmsg;
355 family = pmsg->nm_netmsg.nm_lmsg.u.ms_result;
356 route_proto.sp_family = PF_ROUTE;
357 route_proto.sp_protocol = family;
362 skip = m->m_pkthdr.header;
363 m->m_pkthdr.header = NULL;
365 raw_input(m, &route_proto, &route_src, &route_dst, skip);
369 rts_input_skip(struct mbuf *m, sa_family_t family, struct rawcb *skip)
371 struct netmsg_packet *pmsg;
376 port = cpu0_soport(NULL, NULL, NULL); /* same as for routing socket */
377 pmsg = &m->m_hdr.mh_netmsg;
378 netmsg_init(&pmsg->nm_netmsg, NULL, &netisr_apanic_rport,
379 0, rts_input_handler);
381 pmsg->nm_netmsg.nm_lmsg.u.ms_result = family;
382 m->m_pkthdr.header = skip; /* XXX steal field in pkthdr */
383 lwkt_sendmsg(port, &pmsg->nm_netmsg.nm_lmsg);
387 rts_input(struct mbuf *m, sa_family_t family)
389 rts_input_skip(m, family, NULL);
393 reallocbuf_nofree(void *ptr, size_t len, size_t olen)
397 newptr = kmalloc(len, M_RTABLE, M_INTWAIT | M_NULLOK);
400 bcopy(ptr, newptr, olen);
405 * Internal helper routine for route_output().
408 _fillrtmsg(struct rt_msghdr **prtm, struct rtentry *rt,
409 struct rt_addrinfo *rtinfo)
412 struct rt_msghdr *rtm = *prtm;
414 /* Fill in rt_addrinfo for call to rt_msg_buffer(). */
415 rtinfo->rti_dst = rt_key(rt);
416 rtinfo->rti_gateway = rt->rt_gateway;
417 rtinfo->rti_netmask = rt_mask(rt); /* might be NULL */
418 rtinfo->rti_genmask = rt->rt_genmask; /* might be NULL */
419 if (rtm->rtm_addrs & (RTA_IFP | RTA_IFA)) {
420 if (rt->rt_ifp != NULL) {
421 rtinfo->rti_ifpaddr =
422 TAILQ_FIRST(&rt->rt_ifp->if_addrheads[mycpuid])
424 rtinfo->rti_ifaaddr = rt->rt_ifa->ifa_addr;
425 if (rt->rt_ifp->if_flags & IFF_POINTOPOINT)
426 rtinfo->rti_bcastaddr = rt->rt_ifa->ifa_dstaddr;
427 rtm->rtm_index = rt->rt_ifp->if_index;
429 rtinfo->rti_ifpaddr = NULL;
430 rtinfo->rti_ifaaddr = NULL;
432 } else if (rt->rt_ifp != NULL) {
433 rtm->rtm_index = rt->rt_ifp->if_index;
436 msglen = rt_msgsize(rtm->rtm_type, rtinfo);
437 if (rtm->rtm_msglen < msglen) {
438 /* NOTE: Caller will free the old rtm accordingly */
439 rtm = reallocbuf_nofree(rtm, msglen, rtm->rtm_msglen);
444 rt_msg_buffer(rtm->rtm_type, rtinfo, rtm, msglen);
446 rtm->rtm_flags = rt->rt_flags;
447 rtm->rtm_rmx = rt->rt_rmx;
448 rtm->rtm_addrs = rtinfo->rti_addrs;
454 struct rt_msghdr *bak_rtm;
455 struct rt_msghdr *new_rtm;
459 fillrtmsg(struct rtm_arg *arg, struct rtentry *rt,
460 struct rt_addrinfo *rtinfo)
462 struct rt_msghdr *rtm = arg->new_rtm;
465 error = _fillrtmsg(&rtm, rt, rtinfo);
467 if (arg->new_rtm != rtm) {
469 * _fillrtmsg() just allocated a new rtm;
470 * if the previously allocated rtm is not
471 * the backing rtm, it should be freed.
473 if (arg->new_rtm != arg->bak_rtm)
474 kfree(arg->new_rtm, M_RTABLE);
481 static void route_output_add_callback(int, int, struct rt_addrinfo *,
482 struct rtentry *, void *);
483 static void route_output_delete_callback(int, int, struct rt_addrinfo *,
484 struct rtentry *, void *);
485 static int route_output_get_callback(int, struct rt_addrinfo *,
486 struct rtentry *, void *, int);
487 static int route_output_change_callback(int, struct rt_addrinfo *,
488 struct rtentry *, void *, int);
489 static int route_output_lock_callback(int, struct rt_addrinfo *,
490 struct rtentry *, void *, int);
494 route_output(struct mbuf *m, struct socket *so, ...)
497 struct rt_msghdr *rtm = NULL;
498 struct rawcb *rp = NULL;
499 struct pr_output_info *oi;
500 struct rt_addrinfo rtinfo;
508 oi = __va_arg(ap, struct pr_output_info *);
511 family = familyof(NULL);
513 #define gotoerr(e) { error = e; goto flush;}
516 (m->m_len < sizeof(long) &&
517 (m = m_pullup(m, sizeof(long))) == NULL))
519 len = m->m_pkthdr.len;
520 if (len < sizeof(struct rt_msghdr) ||
521 len != mtod(m, struct rt_msghdr *)->rtm_msglen)
524 rtm = kmalloc(len, M_RTABLE, M_INTWAIT | M_NULLOK);
528 m_copydata(m, 0, len, (caddr_t)rtm);
529 if (rtm->rtm_version != RTM_VERSION)
530 gotoerr(EPROTONOSUPPORT);
532 rtm->rtm_pid = oi->p_pid;
533 bzero(&rtinfo, sizeof(struct rt_addrinfo));
534 rtinfo.rti_addrs = rtm->rtm_addrs;
535 if (rt_xaddrs((char *)(rtm + 1), (char *)rtm + len, &rtinfo) != 0)
538 rtinfo.rti_flags = rtm->rtm_flags;
539 if (rtinfo.rti_dst == NULL || rtinfo.rti_dst->sa_family >= AF_MAX ||
540 (rtinfo.rti_gateway && rtinfo.rti_gateway->sa_family >= AF_MAX))
543 family = familyof(rtinfo.rti_dst);
545 if (rtinfo.rti_genmask != NULL) {
546 error = rtmask_add_global(rtinfo.rti_genmask);
552 * Verify that the caller has the appropriate privilege; RTM_GET
553 * is the only operation the non-superuser is allowed.
555 if (rtm->rtm_type != RTM_GET &&
556 priv_check_cred(so->so_cred, PRIV_ROOT, 0) != 0)
559 switch (rtm->rtm_type) {
561 if (rtinfo.rti_gateway == NULL) {
564 error = rtrequest1_global(RTM_ADD, &rtinfo,
565 route_output_add_callback, rtm);
570 * Backing rtm (bak_rtm) could _not_ be freed during
571 * rtrequest1_global or rtsearch_global, even if the
572 * callback reallocates the rtm due to its size changes,
573 * since rtinfo points to the backing rtm's memory area.
574 * After rtrequest1_global or rtsearch_global returns,
575 * it is safe to free the backing rtm, since rtinfo will
576 * not be used anymore.
578 * new_rtm will be used to save the new rtm allocated
579 * by rtrequest1_global or rtsearch_global.
583 error = rtrequest1_global(RTM_DELETE, &rtinfo,
584 route_output_delete_callback, &arg);
586 if (rtm != arg.bak_rtm)
587 kfree(arg.bak_rtm, M_RTABLE);
590 /* See the comment in RTM_DELETE */
593 error = rtsearch_global(RTM_GET, &rtinfo,
594 route_output_get_callback, &arg,
597 if (rtm != arg.bak_rtm)
598 kfree(arg.bak_rtm, M_RTABLE);
601 error = rtsearch_global(RTM_CHANGE, &rtinfo,
602 route_output_change_callback, rtm,
606 error = rtsearch_global(RTM_LOCK, &rtinfo,
607 route_output_lock_callback, rtm,
617 rtm->rtm_errno = error;
619 rtm->rtm_flags |= RTF_DONE;
623 * Check to see if we don't want our own messages.
625 if (!(so->so_options & SO_USELOOPBACK)) {
626 if (route_cb.any_count <= 1) {
628 kfree(rtm, M_RTABLE);
632 /* There is another listener, so construct message */
636 m_copyback(m, 0, rtm->rtm_msglen, (caddr_t)rtm);
637 if (m->m_pkthdr.len < rtm->rtm_msglen) {
640 } else if (m->m_pkthdr.len > rtm->rtm_msglen)
641 m_adj(m, rtm->rtm_msglen - m->m_pkthdr.len);
642 kfree(rtm, M_RTABLE);
645 rts_input_skip(m, family, rp);
650 route_output_add_callback(int cmd, int error, struct rt_addrinfo *rtinfo,
651 struct rtentry *rt, void *arg)
653 struct rt_msghdr *rtm = arg;
655 if (error == 0 && rt != NULL) {
656 rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx,
658 rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits);
659 rt->rt_rmx.rmx_locks |=
660 (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks);
661 if (rtinfo->rti_genmask != NULL) {
662 rt->rt_genmask = rtmask_purelookup(rtinfo->rti_genmask);
663 if (rt->rt_genmask == NULL) {
665 * This should not happen, since we
666 * have already installed genmask
667 * on each CPU before we reach here.
669 panic("genmask is gone!?");
672 rt->rt_genmask = NULL;
674 rtm->rtm_index = rt->rt_ifp->if_index;
679 route_output_delete_callback(int cmd, int error, struct rt_addrinfo *rtinfo,
680 struct rtentry *rt, void *arg)
682 if (error == 0 && rt) {
684 if (fillrtmsg(arg, rt, rtinfo) != 0) {
686 /* XXX no way to return the error */
690 if (rt && rt->rt_refcnt == 0) {
697 route_output_get_callback(int cmd, struct rt_addrinfo *rtinfo,
698 struct rtentry *rt, void *arg, int found_cnt)
700 int error, found = 0;
702 if (((rtinfo->rti_flags ^ rt->rt_flags) & RTF_HOST) == 0)
705 error = fillrtmsg(arg, rt, rtinfo);
706 if (!error && found) {
707 /* Got the exact match, we could return now! */
714 route_output_change_callback(int cmd, struct rt_addrinfo *rtinfo,
715 struct rtentry *rt, void *arg, int found_cnt)
717 struct rt_msghdr *rtm = arg;
722 * new gateway could require new ifaddr, ifp;
723 * flags may also be different; ifp may be specified
724 * by ll sockaddr when protocol address is ambiguous
726 if (((rt->rt_flags & RTF_GATEWAY) && rtinfo->rti_gateway != NULL) ||
727 rtinfo->rti_ifpaddr != NULL ||
728 (rtinfo->rti_ifaaddr != NULL &&
729 !sa_equal(rtinfo->rti_ifaaddr, rt->rt_ifa->ifa_addr))) {
730 error = rt_getifa(rtinfo);
734 if (rtinfo->rti_gateway != NULL) {
736 * We only need to generate rtmsg upon the
737 * first route to be changed.
739 error = rt_setgate(rt, rt_key(rt), rtinfo->rti_gateway,
740 found_cnt == 1 ? RTL_REPORTMSG : RTL_DONTREPORT);
744 if ((ifa = rtinfo->rti_ifa) != NULL) {
745 struct ifaddr *oifa = rt->rt_ifa;
748 if (oifa && oifa->ifa_rtrequest)
749 oifa->ifa_rtrequest(RTM_DELETE, rt, rtinfo);
753 rt->rt_ifp = rtinfo->rti_ifp;
756 rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx, &rt->rt_rmx);
757 if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest)
758 rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, rtinfo);
759 if (rtinfo->rti_genmask != NULL) {
760 rt->rt_genmask = rtmask_purelookup(rtinfo->rti_genmask);
761 if (rt->rt_genmask == NULL) {
763 * This should not happen, since we
764 * have already installed genmask
765 * on each CPU before we reach here.
767 panic("genmask is gone!?\n");
770 rtm->rtm_index = rt->rt_ifp->if_index;
776 route_output_lock_callback(int cmd, struct rt_addrinfo *rtinfo,
777 struct rtentry *rt, void *arg,
778 int found_cnt __unused)
780 struct rt_msghdr *rtm = arg;
782 rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits);
783 rt->rt_rmx.rmx_locks |=
784 (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks);
789 rt_setmetrics(u_long which, struct rt_metrics *in, struct rt_metrics *out)
791 #define setmetric(flag, elt) if (which & (flag)) out->elt = in->elt;
792 setmetric(RTV_RPIPE, rmx_recvpipe);
793 setmetric(RTV_SPIPE, rmx_sendpipe);
794 setmetric(RTV_SSTHRESH, rmx_ssthresh);
795 setmetric(RTV_RTT, rmx_rtt);
796 setmetric(RTV_RTTVAR, rmx_rttvar);
797 setmetric(RTV_HOPCOUNT, rmx_hopcount);
798 setmetric(RTV_MTU, rmx_mtu);
799 setmetric(RTV_EXPIRE, rmx_expire);
804 ((a) > 0 ? (1 + (((a) - 1) | (sizeof(long) - 1))) : sizeof(long))
807 * Extract the addresses of the passed sockaddrs.
808 * Do a little sanity checking so as to avoid bad memory references.
809 * This data is derived straight from userland.
812 rt_xaddrs(char *cp, char *cplim, struct rt_addrinfo *rtinfo)
817 for (i = 0; (i < RTAX_MAX) && (cp < cplim); i++) {
818 if ((rtinfo->rti_addrs & (1 << i)) == 0)
820 sa = (struct sockaddr *)cp;
824 if ((cp + sa->sa_len) > cplim) {
829 * There are no more... Quit now.
830 * If there are more bits, they are in error.
831 * I've seen this. route(1) can evidently generate these.
832 * This causes kernel to core dump.
833 * For compatibility, if we see this, point to a safe address.
835 if (sa->sa_len == 0) {
836 static struct sockaddr sa_zero = {
837 sizeof sa_zero, AF_INET,
840 rtinfo->rti_info[i] = &sa_zero;
841 kprintf("rtsock: received more addr bits than sockaddrs.\n");
842 return (0); /* should be EINVAL but for compat */
845 /* Accept the sockaddr. */
846 rtinfo->rti_info[i] = sa;
847 cp += ROUNDUP(sa->sa_len);
853 rt_msghdrsize(int type)
858 return sizeof(struct ifa_msghdr);
861 return sizeof(struct ifma_msghdr);
863 return sizeof(struct if_msghdr);
866 return sizeof(struct if_announcemsghdr);
868 return sizeof(struct rt_msghdr);
873 rt_msgsize(int type, struct rt_addrinfo *rtinfo)
877 len = rt_msghdrsize(type);
878 for (i = 0; i < RTAX_MAX; i++) {
879 if (rtinfo->rti_info[i] != NULL)
880 len += ROUNDUP(rtinfo->rti_info[i]->sa_len);
887 * Build a routing message in a buffer.
888 * Copy the addresses in the rtinfo->rti_info[] sockaddr array
889 * to the end of the buffer after the message header.
891 * Set the rtinfo->rti_addrs bitmask of addresses present in rtinfo->rti_info[].
892 * This side-effect can be avoided if we reorder the addrs bitmask field in all
893 * the route messages to line up so we can set it here instead of back in the
897 rt_msg_buffer(int type, struct rt_addrinfo *rtinfo, void *buf, int msglen)
899 struct rt_msghdr *rtm;
903 rtm = (struct rt_msghdr *) buf;
904 rtm->rtm_version = RTM_VERSION;
905 rtm->rtm_type = type;
906 rtm->rtm_msglen = msglen;
908 cp = (char *)buf + rt_msghdrsize(type);
909 rtinfo->rti_addrs = 0;
910 for (i = 0; i < RTAX_MAX; i++) {
913 if ((sa = rtinfo->rti_info[i]) == NULL)
915 rtinfo->rti_addrs |= (1 << i);
916 dlen = ROUNDUP(sa->sa_len);
923 * Build a routing message in a mbuf chain.
924 * Copy the addresses in the rtinfo->rti_info[] sockaddr array
925 * to the end of the mbuf after the message header.
927 * Set the rtinfo->rti_addrs bitmask of addresses present in rtinfo->rti_info[].
928 * This side-effect can be avoided if we reorder the addrs bitmask field in all
929 * the route messages to line up so we can set it here instead of back in the
933 rt_msg_mbuf(int type, struct rt_addrinfo *rtinfo)
936 struct rt_msghdr *rtm;
940 hlen = rt_msghdrsize(type);
941 KASSERT(hlen <= MCLBYTES, ("rt_msg_mbuf: hlen %d doesn't fit", hlen));
943 m = m_getl(hlen, MB_DONTWAIT, MT_DATA, M_PKTHDR, NULL);
947 m->m_pkthdr.len = m->m_len = hlen;
948 m->m_pkthdr.rcvif = NULL;
949 rtinfo->rti_addrs = 0;
951 for (i = 0; i < RTAX_MAX; i++) {
955 if ((sa = rtinfo->rti_info[i]) == NULL)
957 rtinfo->rti_addrs |= (1 << i);
958 dlen = ROUNDUP(sa->sa_len);
959 m_copyback(m, len, dlen, (caddr_t)sa); /* can grow mbuf chain */
962 if (m->m_pkthdr.len != len) { /* one of the m_copyback() calls failed */
966 rtm = mtod(m, struct rt_msghdr *);
968 rtm->rtm_msglen = len;
969 rtm->rtm_version = RTM_VERSION;
970 rtm->rtm_type = type;
975 * This routine is called to generate a message from the routing
976 * socket indicating that a redirect has occurred, a routing lookup
977 * has failed, or that a protocol has detected timeouts to a particular
981 rt_missmsg(int type, struct rt_addrinfo *rtinfo, int flags, int error)
983 struct sockaddr *dst = rtinfo->rti_info[RTAX_DST];
984 struct rt_msghdr *rtm;
987 if (route_cb.any_count == 0)
989 m = rt_msg_mbuf(type, rtinfo);
992 rtm = mtod(m, struct rt_msghdr *);
993 rtm->rtm_flags = RTF_DONE | flags;
994 rtm->rtm_errno = error;
995 rtm->rtm_addrs = rtinfo->rti_addrs;
996 rts_input(m, familyof(dst));
1000 rt_dstmsg(int type, struct sockaddr *dst, int error)
1002 struct rt_msghdr *rtm;
1003 struct rt_addrinfo addrs;
1006 if (route_cb.any_count == 0)
1008 bzero(&addrs, sizeof(struct rt_addrinfo));
1009 addrs.rti_info[RTAX_DST] = dst;
1010 m = rt_msg_mbuf(type, &addrs);
1013 rtm = mtod(m, struct rt_msghdr *);
1014 rtm->rtm_flags = RTF_DONE;
1015 rtm->rtm_errno = error;
1016 rtm->rtm_addrs = addrs.rti_addrs;
1017 rts_input(m, familyof(dst));
1021 * This routine is called to generate a message from the routing
1022 * socket indicating that the status of a network interface has changed.
1025 rt_ifmsg(struct ifnet *ifp)
1027 struct if_msghdr *ifm;
1029 struct rt_addrinfo rtinfo;
1031 if (route_cb.any_count == 0)
1033 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1034 m = rt_msg_mbuf(RTM_IFINFO, &rtinfo);
1037 ifm = mtod(m, struct if_msghdr *);
1038 ifm->ifm_index = ifp->if_index;
1039 ifm->ifm_flags = ifp->if_flags;
1040 ifm->ifm_data = ifp->if_data;
1046 rt_ifamsg(int cmd, struct ifaddr *ifa)
1048 struct ifa_msghdr *ifam;
1049 struct rt_addrinfo rtinfo;
1051 struct ifnet *ifp = ifa->ifa_ifp;
1053 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1054 rtinfo.rti_ifaaddr = ifa->ifa_addr;
1055 rtinfo.rti_ifpaddr =
1056 TAILQ_FIRST(&ifp->if_addrheads[mycpuid])->ifa->ifa_addr;
1057 rtinfo.rti_netmask = ifa->ifa_netmask;
1058 rtinfo.rti_bcastaddr = ifa->ifa_dstaddr;
1060 m = rt_msg_mbuf(cmd, &rtinfo);
1064 ifam = mtod(m, struct ifa_msghdr *);
1065 ifam->ifam_index = ifp->if_index;
1066 ifam->ifam_metric = ifa->ifa_metric;
1067 ifam->ifam_flags = ifa->ifa_flags;
1068 ifam->ifam_addrs = rtinfo.rti_addrs;
1070 rts_input(m, familyof(ifa->ifa_addr));
1074 rt_rtmsg(int cmd, struct rtentry *rt, struct ifnet *ifp, int error)
1076 struct rt_msghdr *rtm;
1077 struct rt_addrinfo rtinfo;
1079 struct sockaddr *dst;
1084 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1085 rtinfo.rti_dst = dst = rt_key(rt);
1086 rtinfo.rti_gateway = rt->rt_gateway;
1087 rtinfo.rti_netmask = rt_mask(rt);
1089 rtinfo.rti_ifpaddr =
1090 TAILQ_FIRST(&ifp->if_addrheads[mycpuid])->ifa->ifa_addr;
1092 rtinfo.rti_ifaaddr = rt->rt_ifa->ifa_addr;
1094 m = rt_msg_mbuf(cmd, &rtinfo);
1098 rtm = mtod(m, struct rt_msghdr *);
1100 rtm->rtm_index = ifp->if_index;
1101 rtm->rtm_flags |= rt->rt_flags;
1102 rtm->rtm_errno = error;
1103 rtm->rtm_addrs = rtinfo.rti_addrs;
1105 rts_input(m, familyof(dst));
1109 * This is called to generate messages from the routing socket
1110 * indicating a network interface has had addresses associated with it.
1111 * if we ever reverse the logic and replace messages TO the routing
1112 * socket indicate a request to configure interfaces, then it will
1113 * be unnecessary as the routing socket will automatically generate
1117 rt_newaddrmsg(int cmd, struct ifaddr *ifa, int error, struct rtentry *rt)
1121 * notify the SCTP stack
1122 * this will only get called when an address is added/deleted
1123 * XXX pass the ifaddr struct instead if ifa->ifa_addr...
1126 sctp_add_ip_address(ifa);
1127 else if (cmd == RTM_DELETE)
1128 sctp_delete_ip_address(ifa);
1131 if (route_cb.any_count == 0)
1134 if (cmd == RTM_ADD) {
1135 rt_ifamsg(RTM_NEWADDR, ifa);
1136 rt_rtmsg(RTM_ADD, rt, ifa->ifa_ifp, error);
1138 KASSERT((cmd == RTM_DELETE), ("unknown cmd %d", cmd));
1139 rt_rtmsg(RTM_DELETE, rt, ifa->ifa_ifp, error);
1140 rt_ifamsg(RTM_DELADDR, ifa);
1145 * This is the analogue to the rt_newaddrmsg which performs the same
1146 * function but for multicast group memberhips. This is easier since
1147 * there is no route state to worry about.
1150 rt_newmaddrmsg(int cmd, struct ifmultiaddr *ifma)
1152 struct rt_addrinfo rtinfo;
1153 struct mbuf *m = NULL;
1154 struct ifnet *ifp = ifma->ifma_ifp;
1155 struct ifma_msghdr *ifmam;
1157 if (route_cb.any_count == 0)
1160 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1161 rtinfo.rti_ifaaddr = ifma->ifma_addr;
1162 if (ifp != NULL && !TAILQ_EMPTY(&ifp->if_addrheads[mycpuid])) {
1163 rtinfo.rti_ifpaddr =
1164 TAILQ_FIRST(&ifp->if_addrheads[mycpuid])->ifa->ifa_addr;
1167 * If a link-layer address is present, present it as a ``gateway''
1168 * (similarly to how ARP entries, e.g., are presented).
1170 rtinfo.rti_gateway = ifma->ifma_lladdr;
1172 m = rt_msg_mbuf(cmd, &rtinfo);
1176 ifmam = mtod(m, struct ifma_msghdr *);
1177 ifmam->ifmam_index = ifp->if_index;
1178 ifmam->ifmam_addrs = rtinfo.rti_addrs;
1180 rts_input(m, familyof(ifma->ifma_addr));
1183 static struct mbuf *
1184 rt_makeifannouncemsg(struct ifnet *ifp, int type, int what,
1185 struct rt_addrinfo *info)
1187 struct if_announcemsghdr *ifan;
1190 if (route_cb.any_count == 0)
1193 bzero(info, sizeof(*info));
1194 m = rt_msg_mbuf(type, info);
1198 ifan = mtod(m, struct if_announcemsghdr *);
1199 ifan->ifan_index = ifp->if_index;
1200 strlcpy(ifan->ifan_name, ifp->if_xname, sizeof ifan->ifan_name);
1201 ifan->ifan_what = what;
1206 * This is called to generate routing socket messages indicating
1207 * IEEE80211 wireless events.
1208 * XXX we piggyback on the RTM_IFANNOUNCE msg format in a clumsy way.
1211 rt_ieee80211msg(struct ifnet *ifp, int what, void *data, size_t data_len)
1213 struct rt_addrinfo info;
1216 m = rt_makeifannouncemsg(ifp, RTM_IEEE80211, what, &info);
1221 * Append the ieee80211 data. Try to stick it in the
1222 * mbuf containing the ifannounce msg; otherwise allocate
1223 * a new mbuf and append.
1225 * NB: we assume m is a single mbuf.
1227 if (data_len > M_TRAILINGSPACE(m)) {
1228 struct mbuf *n = m_get(MB_DONTWAIT, MT_DATA);
1233 bcopy(data, mtod(n, void *), data_len);
1234 n->m_len = data_len;
1236 } else if (data_len > 0) {
1237 bcopy(data, mtod(m, u_int8_t *) + m->m_len, data_len);
1238 m->m_len += data_len;
1241 if (m->m_flags & M_PKTHDR)
1242 m->m_pkthdr.len += data_len;
1243 mtod(m, struct if_announcemsghdr *)->ifan_msglen += data_len;
1248 * This is called to generate routing socket messages indicating
1249 * network interface arrival and departure.
1252 rt_ifannouncemsg(struct ifnet *ifp, int what)
1254 struct rt_addrinfo addrinfo;
1257 m = rt_makeifannouncemsg(ifp, RTM_IFANNOUNCE, what, &addrinfo);
1263 resizewalkarg(struct walkarg *w, int len)
1267 newptr = kmalloc(len, M_RTABLE, M_INTWAIT | M_NULLOK);
1270 if (w->w_tmem != NULL)
1271 kfree(w->w_tmem, M_RTABLE);
1273 w->w_tmemsize = len;
1278 * This is used in dumping the kernel table via sysctl().
1281 sysctl_dumpentry(struct radix_node *rn, void *vw)
1283 struct walkarg *w = vw;
1284 struct rtentry *rt = (struct rtentry *)rn;
1285 struct rt_addrinfo rtinfo;
1288 if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg))
1291 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1292 rtinfo.rti_dst = rt_key(rt);
1293 rtinfo.rti_gateway = rt->rt_gateway;
1294 rtinfo.rti_netmask = rt_mask(rt);
1295 rtinfo.rti_genmask = rt->rt_genmask;
1296 if (rt->rt_ifp != NULL) {
1297 rtinfo.rti_ifpaddr =
1298 TAILQ_FIRST(&rt->rt_ifp->if_addrheads[mycpuid])->ifa->ifa_addr;
1299 rtinfo.rti_ifaaddr = rt->rt_ifa->ifa_addr;
1300 if (rt->rt_ifp->if_flags & IFF_POINTOPOINT)
1301 rtinfo.rti_bcastaddr = rt->rt_ifa->ifa_dstaddr;
1303 msglen = rt_msgsize(RTM_GET, &rtinfo);
1304 if (w->w_tmemsize < msglen && resizewalkarg(w, msglen) != 0)
1306 rt_msg_buffer(RTM_GET, &rtinfo, w->w_tmem, msglen);
1307 if (w->w_req != NULL) {
1308 struct rt_msghdr *rtm = w->w_tmem;
1310 rtm->rtm_flags = rt->rt_flags;
1311 rtm->rtm_use = rt->rt_use;
1312 rtm->rtm_rmx = rt->rt_rmx;
1313 rtm->rtm_index = rt->rt_ifp->if_index;
1314 rtm->rtm_errno = rtm->rtm_pid = rtm->rtm_seq = 0;
1315 rtm->rtm_addrs = rtinfo.rti_addrs;
1316 error = SYSCTL_OUT(w->w_req, rtm, msglen);
1323 sysctl_iflist(int af, struct walkarg *w)
1326 struct rt_addrinfo rtinfo;
1329 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1330 TAILQ_FOREACH(ifp, &ifnet, if_link) {
1331 struct ifaddr_container *ifac;
1334 if (w->w_arg && w->w_arg != ifp->if_index)
1336 ifac = TAILQ_FIRST(&ifp->if_addrheads[mycpuid]);
1338 rtinfo.rti_ifpaddr = ifa->ifa_addr;
1339 msglen = rt_msgsize(RTM_IFINFO, &rtinfo);
1340 if (w->w_tmemsize < msglen && resizewalkarg(w, msglen) != 0)
1342 rt_msg_buffer(RTM_IFINFO, &rtinfo, w->w_tmem, msglen);
1343 rtinfo.rti_ifpaddr = NULL;
1344 if (w->w_req != NULL && w->w_tmem != NULL) {
1345 struct if_msghdr *ifm = w->w_tmem;
1347 ifm->ifm_index = ifp->if_index;
1348 ifm->ifm_flags = ifp->if_flags;
1349 ifm->ifm_data = ifp->if_data;
1350 ifm->ifm_addrs = rtinfo.rti_addrs;
1351 error = SYSCTL_OUT(w->w_req, ifm, msglen);
1355 while ((ifac = TAILQ_NEXT(ifac, ifa_link)) != NULL) {
1358 if (af && af != ifa->ifa_addr->sa_family)
1360 if (curproc->p_ucred->cr_prison &&
1361 prison_if(curproc->p_ucred, ifa->ifa_addr))
1363 rtinfo.rti_ifaaddr = ifa->ifa_addr;
1364 rtinfo.rti_netmask = ifa->ifa_netmask;
1365 rtinfo.rti_bcastaddr = ifa->ifa_dstaddr;
1366 msglen = rt_msgsize(RTM_NEWADDR, &rtinfo);
1367 if (w->w_tmemsize < msglen &&
1368 resizewalkarg(w, msglen) != 0)
1370 rt_msg_buffer(RTM_NEWADDR, &rtinfo, w->w_tmem, msglen);
1371 if (w->w_req != NULL) {
1372 struct ifa_msghdr *ifam = w->w_tmem;
1374 ifam->ifam_index = ifa->ifa_ifp->if_index;
1375 ifam->ifam_flags = ifa->ifa_flags;
1376 ifam->ifam_metric = ifa->ifa_metric;
1377 ifam->ifam_addrs = rtinfo.rti_addrs;
1378 error = SYSCTL_OUT(w->w_req, w->w_tmem, msglen);
1383 rtinfo.rti_netmask = NULL;
1384 rtinfo.rti_ifaaddr = NULL;
1385 rtinfo.rti_bcastaddr = NULL;
1391 sysctl_rtsock(SYSCTL_HANDLER_ARGS)
1393 int *name = (int *)arg1;
1394 u_int namelen = arg2;
1395 struct radix_node_head *rnh;
1396 int i, error = EINVAL;
1405 if (namelen != 3 && namelen != 4)
1408 bzero(&w, sizeof w);
1414 * Optional third argument specifies cpu, used primarily for
1415 * debugging the route table.
1418 if (name[3] < 0 || name[3] >= ncpus)
1421 lwkt_migratecpu(name[3]);
1429 for (i = 1; i <= AF_MAX; i++)
1430 if ((rnh = rt_tables[mycpuid][i]) &&
1431 (af == 0 || af == i) &&
1432 (error = rnh->rnh_walktree(rnh,
1433 sysctl_dumpentry, &w)))
1438 error = sysctl_iflist(af, &w);
1441 if (w.w_tmem != NULL)
1442 kfree(w.w_tmem, M_RTABLE);
1444 lwkt_migratecpu(origcpu);
1448 SYSCTL_NODE(_net, PF_ROUTE, routetable, CTLFLAG_RD, sysctl_rtsock, "");
1451 * Definitions of protocols supported in the ROUTE domain.
1454 static struct domain routedomain; /* or at least forward */
1456 static struct protosw routesw[] = {
1457 { SOCK_RAW, &routedomain, 0, PR_ATOMIC|PR_ADDR,
1458 0, route_output, raw_ctlinput, 0,
1459 cpu0_soport, cpu0_ctlport,
1465 static struct domain routedomain = {
1466 PF_ROUTE, "route", NULL, NULL, NULL,
1467 routesw, &routesw[(sizeof routesw)/(sizeof routesw[0])],