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>
85 #include <sys/thread2.h>
86 #include <sys/socketvar2.h>
89 #include <net/route.h>
90 #include <net/raw_cb.h>
91 #include <net/netmsg2.h>
94 extern void sctp_add_ip_address(struct ifaddr *ifa);
95 extern void sctp_delete_ip_address(struct ifaddr *ifa);
98 MALLOC_DEFINE(M_RTABLE, "routetbl", "routing tables");
100 static struct route_cb {
108 static const struct sockaddr route_src = { 2, PF_ROUTE, };
114 struct sysctl_req *w_req;
118 rt_msg_mbuf (int, struct rt_addrinfo *);
119 static void rt_msg_buffer (int, struct rt_addrinfo *, void *buf, int len);
120 static int rt_msgsize (int type, struct rt_addrinfo *rtinfo);
121 static int rt_xaddrs (char *, char *, struct rt_addrinfo *);
122 static int sysctl_dumpentry (struct radix_node *rn, void *vw);
123 static int sysctl_iflist (int af, struct walkarg *w);
124 static int route_output(struct mbuf *, struct socket *, ...);
125 static void rt_setmetrics (u_long, struct rt_metrics *,
126 struct rt_metrics *);
129 * It really doesn't make any sense at all for this code to share much
130 * with raw_usrreq.c, since its functionality is so restricted. XXX
133 rts_abort(netmsg_t msg)
136 raw_usrreqs.pru_abort(msg);
137 /* msg invalid now */
141 /* pru_accept is EOPNOTSUPP */
144 rts_attach(netmsg_t msg)
146 struct socket *so = msg->base.nm_so;
147 struct pru_attach_info *ai = msg->attach.nm_ai;
149 int proto = msg->attach.nm_proto;
153 if (sotorawcb(so) != NULL) {
158 rp = kmalloc(sizeof *rp, M_PCB, M_WAITOK | M_ZERO);
161 * The critical section is necessary to block protocols from sending
162 * error notifications (like RTM_REDIRECT or RTM_LOSING) while
163 * this PCB is extant but incompletely initialized.
164 * Probably we should try to do more of this work beforehand and
165 * eliminate the critical section.
168 soreference(so); /* so_pcb assignment */
169 error = raw_attach(so, proto, ai->sb_rlimit);
175 switch(rp->rcb_proto.sp_protocol) {
180 route_cb.ip6_count++;
183 route_cb.ipx_count++;
189 rp->rcb_faddr = &route_src;
190 route_cb.any_count++;
192 so->so_options |= SO_USELOOPBACK;
196 lwkt_replymsg(&msg->lmsg, error);
200 rts_bind(netmsg_t msg)
203 raw_usrreqs.pru_bind(msg); /* xxx just EINVAL */
204 /* msg invalid now */
209 rts_connect(netmsg_t msg)
212 raw_usrreqs.pru_connect(msg); /* XXX just EINVAL */
213 /* msg invalid now */
217 /* pru_connect2 is EOPNOTSUPP */
218 /* pru_control is EOPNOTSUPP */
221 rts_detach(netmsg_t msg)
223 struct socket *so = msg->base.nm_so;
224 struct rawcb *rp = sotorawcb(so);
228 switch(rp->rcb_proto.sp_protocol) {
233 route_cb.ip6_count--;
236 route_cb.ipx_count--;
242 route_cb.any_count--;
244 raw_usrreqs.pru_detach(msg);
245 /* msg invalid now */
250 rts_disconnect(netmsg_t msg)
253 raw_usrreqs.pru_disconnect(msg);
254 /* msg invalid now */
258 /* pru_listen is EOPNOTSUPP */
261 rts_peeraddr(netmsg_t msg)
264 raw_usrreqs.pru_peeraddr(msg);
265 /* msg invalid now */
269 /* pru_rcvd is EOPNOTSUPP */
270 /* pru_rcvoob is EOPNOTSUPP */
273 rts_send(netmsg_t msg)
276 raw_usrreqs.pru_send(msg);
277 /* msg invalid now */
281 /* pru_sense is null */
284 rts_shutdown(netmsg_t msg)
287 raw_usrreqs.pru_shutdown(msg);
288 /* msg invalid now */
293 rts_sockaddr(netmsg_t msg)
296 raw_usrreqs.pru_sockaddr(msg);
297 /* msg invalid now */
301 static struct pr_usrreqs route_usrreqs = {
302 .pru_abort = rts_abort,
303 .pru_accept = pr_generic_notsupp,
304 .pru_attach = rts_attach,
305 .pru_bind = rts_bind,
306 .pru_connect = rts_connect,
307 .pru_connect2 = pr_generic_notsupp,
308 .pru_control = pr_generic_notsupp,
309 .pru_detach = rts_detach,
310 .pru_disconnect = rts_disconnect,
311 .pru_listen = pr_generic_notsupp,
312 .pru_peeraddr = rts_peeraddr,
313 .pru_rcvd = pr_generic_notsupp,
314 .pru_rcvoob = pr_generic_notsupp,
315 .pru_send = rts_send,
316 .pru_sense = pru_sense_null,
317 .pru_shutdown = rts_shutdown,
318 .pru_sockaddr = rts_sockaddr,
319 .pru_sosend = sosend,
320 .pru_soreceive = soreceive
323 static __inline sa_family_t
324 familyof(struct sockaddr *sa)
326 return (sa != NULL ? sa->sa_family : 0);
330 * Routing socket input function. The packet must be serialized onto cpu 0.
331 * We use the cpu0_soport() netisr processing loop to handle it.
333 * This looks messy but it means that anyone, including interrupt code,
334 * can send a message to the routing socket.
337 rts_input_handler(netmsg_t msg)
339 static const struct sockaddr route_dst = { 2, PF_ROUTE, };
340 struct sockproto route_proto;
341 struct netmsg_packet *pmsg = &msg->packet;
346 family = pmsg->base.lmsg.u.ms_result;
347 route_proto.sp_family = PF_ROUTE;
348 route_proto.sp_protocol = family;
353 skip = m->m_pkthdr.header;
354 m->m_pkthdr.header = NULL;
356 raw_input(m, &route_proto, &route_src, &route_dst, skip);
360 rts_input_skip(struct mbuf *m, sa_family_t family, struct rawcb *skip)
362 struct netmsg_packet *pmsg;
367 port = cpu_portfn(0); /* XXX same as for routing socket */
368 pmsg = &m->m_hdr.mh_netmsg;
369 netmsg_init(&pmsg->base, NULL, &netisr_apanic_rport,
370 0, rts_input_handler);
372 pmsg->base.lmsg.u.ms_result = family;
373 m->m_pkthdr.header = skip; /* XXX steal field in pkthdr */
374 lwkt_sendmsg(port, &pmsg->base.lmsg);
378 rts_input(struct mbuf *m, sa_family_t family)
380 rts_input_skip(m, family, NULL);
384 reallocbuf_nofree(void *ptr, size_t len, size_t olen)
388 newptr = kmalloc(len, M_RTABLE, M_INTWAIT | M_NULLOK);
391 bcopy(ptr, newptr, olen);
396 * Internal helper routine for route_output().
399 _fillrtmsg(struct rt_msghdr **prtm, struct rtentry *rt,
400 struct rt_addrinfo *rtinfo)
403 struct rt_msghdr *rtm = *prtm;
405 /* Fill in rt_addrinfo for call to rt_msg_buffer(). */
406 rtinfo->rti_dst = rt_key(rt);
407 rtinfo->rti_gateway = rt->rt_gateway;
408 rtinfo->rti_netmask = rt_mask(rt); /* might be NULL */
409 rtinfo->rti_genmask = rt->rt_genmask; /* might be NULL */
410 if (rtm->rtm_addrs & (RTA_IFP | RTA_IFA)) {
411 if (rt->rt_ifp != NULL) {
412 rtinfo->rti_ifpaddr =
413 TAILQ_FIRST(&rt->rt_ifp->if_addrheads[mycpuid])
415 rtinfo->rti_ifaaddr = rt->rt_ifa->ifa_addr;
416 if (rt->rt_ifp->if_flags & IFF_POINTOPOINT)
417 rtinfo->rti_bcastaddr = rt->rt_ifa->ifa_dstaddr;
418 rtm->rtm_index = rt->rt_ifp->if_index;
420 rtinfo->rti_ifpaddr = NULL;
421 rtinfo->rti_ifaaddr = NULL;
423 } else if (rt->rt_ifp != NULL) {
424 rtm->rtm_index = rt->rt_ifp->if_index;
427 msglen = rt_msgsize(rtm->rtm_type, rtinfo);
428 if (rtm->rtm_msglen < msglen) {
429 /* NOTE: Caller will free the old rtm accordingly */
430 rtm = reallocbuf_nofree(rtm, msglen, rtm->rtm_msglen);
435 rt_msg_buffer(rtm->rtm_type, rtinfo, rtm, msglen);
437 rtm->rtm_flags = rt->rt_flags;
438 rtm->rtm_rmx = rt->rt_rmx;
439 rtm->rtm_addrs = rtinfo->rti_addrs;
445 struct rt_msghdr *bak_rtm;
446 struct rt_msghdr *new_rtm;
450 fillrtmsg(struct rtm_arg *arg, struct rtentry *rt,
451 struct rt_addrinfo *rtinfo)
453 struct rt_msghdr *rtm = arg->new_rtm;
456 error = _fillrtmsg(&rtm, rt, rtinfo);
458 if (arg->new_rtm != rtm) {
460 * _fillrtmsg() just allocated a new rtm;
461 * if the previously allocated rtm is not
462 * the backing rtm, it should be freed.
464 if (arg->new_rtm != arg->bak_rtm)
465 kfree(arg->new_rtm, M_RTABLE);
472 static void route_output_add_callback(int, int, struct rt_addrinfo *,
473 struct rtentry *, void *);
474 static void route_output_delete_callback(int, int, struct rt_addrinfo *,
475 struct rtentry *, void *);
476 static int route_output_get_callback(int, struct rt_addrinfo *,
477 struct rtentry *, void *, int);
478 static int route_output_change_callback(int, struct rt_addrinfo *,
479 struct rtentry *, void *, int);
480 static int route_output_lock_callback(int, struct rt_addrinfo *,
481 struct rtentry *, void *, int);
485 route_output(struct mbuf *m, struct socket *so, ...)
488 struct rt_msghdr *rtm = NULL;
489 struct rawcb *rp = NULL;
490 struct pr_output_info *oi;
491 struct rt_addrinfo rtinfo;
499 oi = __va_arg(ap, struct pr_output_info *);
502 family = familyof(NULL);
504 #define gotoerr(e) { error = e; goto flush;}
507 (m->m_len < sizeof(long) &&
508 (m = m_pullup(m, sizeof(long))) == NULL))
510 len = m->m_pkthdr.len;
511 if (len < sizeof(struct rt_msghdr) ||
512 len != mtod(m, struct rt_msghdr *)->rtm_msglen)
515 rtm = kmalloc(len, M_RTABLE, M_INTWAIT | M_NULLOK);
519 m_copydata(m, 0, len, (caddr_t)rtm);
520 if (rtm->rtm_version != RTM_VERSION)
521 gotoerr(EPROTONOSUPPORT);
523 rtm->rtm_pid = oi->p_pid;
524 bzero(&rtinfo, sizeof(struct rt_addrinfo));
525 rtinfo.rti_addrs = rtm->rtm_addrs;
526 if (rt_xaddrs((char *)(rtm + 1), (char *)rtm + len, &rtinfo) != 0)
529 rtinfo.rti_flags = rtm->rtm_flags;
530 if (rtinfo.rti_dst == NULL || rtinfo.rti_dst->sa_family >= AF_MAX ||
531 (rtinfo.rti_gateway && rtinfo.rti_gateway->sa_family >= AF_MAX))
534 family = familyof(rtinfo.rti_dst);
536 if (rtinfo.rti_genmask != NULL) {
537 error = rtmask_add_global(rtinfo.rti_genmask);
543 * Verify that the caller has the appropriate privilege; RTM_GET
544 * is the only operation the non-superuser is allowed.
546 if (rtm->rtm_type != RTM_GET &&
547 priv_check_cred(so->so_cred, PRIV_ROOT, 0) != 0)
550 switch (rtm->rtm_type) {
552 if (rtinfo.rti_gateway == NULL) {
555 error = rtrequest1_global(RTM_ADD, &rtinfo,
556 route_output_add_callback, rtm);
561 * Backing rtm (bak_rtm) could _not_ be freed during
562 * rtrequest1_global or rtsearch_global, even if the
563 * callback reallocates the rtm due to its size changes,
564 * since rtinfo points to the backing rtm's memory area.
565 * After rtrequest1_global or rtsearch_global returns,
566 * it is safe to free the backing rtm, since rtinfo will
567 * not be used anymore.
569 * new_rtm will be used to save the new rtm allocated
570 * by rtrequest1_global or rtsearch_global.
574 error = rtrequest1_global(RTM_DELETE, &rtinfo,
575 route_output_delete_callback, &arg);
577 if (rtm != arg.bak_rtm)
578 kfree(arg.bak_rtm, M_RTABLE);
581 /* See the comment in RTM_DELETE */
584 error = rtsearch_global(RTM_GET, &rtinfo,
585 route_output_get_callback, &arg,
588 if (rtm != arg.bak_rtm)
589 kfree(arg.bak_rtm, M_RTABLE);
592 error = rtsearch_global(RTM_CHANGE, &rtinfo,
593 route_output_change_callback, rtm,
597 error = rtsearch_global(RTM_LOCK, &rtinfo,
598 route_output_lock_callback, rtm,
608 rtm->rtm_errno = error;
610 rtm->rtm_flags |= RTF_DONE;
614 * Check to see if we don't want our own messages.
616 if (!(so->so_options & SO_USELOOPBACK)) {
617 if (route_cb.any_count <= 1) {
619 kfree(rtm, M_RTABLE);
623 /* There is another listener, so construct message */
627 m_copyback(m, 0, rtm->rtm_msglen, (caddr_t)rtm);
628 if (m->m_pkthdr.len < rtm->rtm_msglen) {
631 } else if (m->m_pkthdr.len > rtm->rtm_msglen)
632 m_adj(m, rtm->rtm_msglen - m->m_pkthdr.len);
633 kfree(rtm, M_RTABLE);
636 rts_input_skip(m, family, rp);
641 route_output_add_callback(int cmd, int error, struct rt_addrinfo *rtinfo,
642 struct rtentry *rt, void *arg)
644 struct rt_msghdr *rtm = arg;
646 if (error == 0 && rt != NULL) {
647 rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx,
649 rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits);
650 rt->rt_rmx.rmx_locks |=
651 (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks);
652 if (rtinfo->rti_genmask != NULL) {
653 rt->rt_genmask = rtmask_purelookup(rtinfo->rti_genmask);
654 if (rt->rt_genmask == NULL) {
656 * This should not happen, since we
657 * have already installed genmask
658 * on each CPU before we reach here.
660 panic("genmask is gone!?");
663 rt->rt_genmask = NULL;
665 rtm->rtm_index = rt->rt_ifp->if_index;
670 route_output_delete_callback(int cmd, int error, struct rt_addrinfo *rtinfo,
671 struct rtentry *rt, void *arg)
673 if (error == 0 && rt) {
675 if (fillrtmsg(arg, rt, rtinfo) != 0) {
677 /* XXX no way to return the error */
681 if (rt && rt->rt_refcnt == 0) {
688 route_output_get_callback(int cmd, struct rt_addrinfo *rtinfo,
689 struct rtentry *rt, void *arg, int found_cnt)
691 int error, found = 0;
693 if (((rtinfo->rti_flags ^ rt->rt_flags) & RTF_HOST) == 0)
696 error = fillrtmsg(arg, rt, rtinfo);
697 if (!error && found) {
698 /* Got the exact match, we could return now! */
705 route_output_change_callback(int cmd, struct rt_addrinfo *rtinfo,
706 struct rtentry *rt, void *arg, int found_cnt)
708 struct rt_msghdr *rtm = arg;
713 * new gateway could require new ifaddr, ifp;
714 * flags may also be different; ifp may be specified
715 * by ll sockaddr when protocol address is ambiguous
717 if (((rt->rt_flags & RTF_GATEWAY) && rtinfo->rti_gateway != NULL) ||
718 rtinfo->rti_ifpaddr != NULL ||
719 (rtinfo->rti_ifaaddr != NULL &&
720 !sa_equal(rtinfo->rti_ifaaddr, rt->rt_ifa->ifa_addr))) {
721 error = rt_getifa(rtinfo);
725 if (rtinfo->rti_gateway != NULL) {
727 * We only need to generate rtmsg upon the
728 * first route to be changed.
730 error = rt_setgate(rt, rt_key(rt), rtinfo->rti_gateway,
731 found_cnt == 1 ? RTL_REPORTMSG : RTL_DONTREPORT);
735 if ((ifa = rtinfo->rti_ifa) != NULL) {
736 struct ifaddr *oifa = rt->rt_ifa;
739 if (oifa && oifa->ifa_rtrequest)
740 oifa->ifa_rtrequest(RTM_DELETE, rt, rtinfo);
744 rt->rt_ifp = rtinfo->rti_ifp;
747 rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx, &rt->rt_rmx);
748 if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest)
749 rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, rtinfo);
750 if (rtinfo->rti_genmask != NULL) {
751 rt->rt_genmask = rtmask_purelookup(rtinfo->rti_genmask);
752 if (rt->rt_genmask == NULL) {
754 * This should not happen, since we
755 * have already installed genmask
756 * on each CPU before we reach here.
758 panic("genmask is gone!?\n");
761 rtm->rtm_index = rt->rt_ifp->if_index;
767 route_output_lock_callback(int cmd, struct rt_addrinfo *rtinfo,
768 struct rtentry *rt, void *arg,
769 int found_cnt __unused)
771 struct rt_msghdr *rtm = arg;
773 rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits);
774 rt->rt_rmx.rmx_locks |=
775 (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks);
780 rt_setmetrics(u_long which, struct rt_metrics *in, struct rt_metrics *out)
782 #define setmetric(flag, elt) if (which & (flag)) out->elt = in->elt;
783 setmetric(RTV_RPIPE, rmx_recvpipe);
784 setmetric(RTV_SPIPE, rmx_sendpipe);
785 setmetric(RTV_SSTHRESH, rmx_ssthresh);
786 setmetric(RTV_RTT, rmx_rtt);
787 setmetric(RTV_RTTVAR, rmx_rttvar);
788 setmetric(RTV_HOPCOUNT, rmx_hopcount);
789 setmetric(RTV_MTU, rmx_mtu);
790 setmetric(RTV_EXPIRE, rmx_expire);
791 setmetric(RTV_MSL, rmx_msl);
796 ((a) > 0 ? (1 + (((a) - 1) | (sizeof(long) - 1))) : sizeof(long))
799 * Extract the addresses of the passed sockaddrs.
800 * Do a little sanity checking so as to avoid bad memory references.
801 * This data is derived straight from userland.
804 rt_xaddrs(char *cp, char *cplim, struct rt_addrinfo *rtinfo)
809 for (i = 0; (i < RTAX_MAX) && (cp < cplim); i++) {
810 if ((rtinfo->rti_addrs & (1 << i)) == 0)
812 sa = (struct sockaddr *)cp;
816 if ((cp + sa->sa_len) > cplim) {
821 * There are no more... Quit now.
822 * If there are more bits, they are in error.
823 * I've seen this. route(1) can evidently generate these.
824 * This causes kernel to core dump.
825 * For compatibility, if we see this, point to a safe address.
827 if (sa->sa_len == 0) {
828 static struct sockaddr sa_zero = {
829 sizeof sa_zero, AF_INET,
832 rtinfo->rti_info[i] = &sa_zero;
833 kprintf("rtsock: received more addr bits than sockaddrs.\n");
834 return (0); /* should be EINVAL but for compat */
837 /* Accept the sockaddr. */
838 rtinfo->rti_info[i] = sa;
839 cp += ROUNDUP(sa->sa_len);
845 rt_msghdrsize(int type)
850 return sizeof(struct ifa_msghdr);
853 return sizeof(struct ifma_msghdr);
855 return sizeof(struct if_msghdr);
858 return sizeof(struct if_announcemsghdr);
860 return sizeof(struct rt_msghdr);
865 rt_msgsize(int type, struct rt_addrinfo *rtinfo)
869 len = rt_msghdrsize(type);
870 for (i = 0; i < RTAX_MAX; i++) {
871 if (rtinfo->rti_info[i] != NULL)
872 len += ROUNDUP(rtinfo->rti_info[i]->sa_len);
879 * Build a routing message in a buffer.
880 * Copy the addresses in the rtinfo->rti_info[] sockaddr array
881 * to the end of the buffer after the message header.
883 * Set the rtinfo->rti_addrs bitmask of addresses present in rtinfo->rti_info[].
884 * This side-effect can be avoided if we reorder the addrs bitmask field in all
885 * the route messages to line up so we can set it here instead of back in the
889 rt_msg_buffer(int type, struct rt_addrinfo *rtinfo, void *buf, int msglen)
891 struct rt_msghdr *rtm;
895 rtm = (struct rt_msghdr *) buf;
896 rtm->rtm_version = RTM_VERSION;
897 rtm->rtm_type = type;
898 rtm->rtm_msglen = msglen;
900 cp = (char *)buf + rt_msghdrsize(type);
901 rtinfo->rti_addrs = 0;
902 for (i = 0; i < RTAX_MAX; i++) {
905 if ((sa = rtinfo->rti_info[i]) == NULL)
907 rtinfo->rti_addrs |= (1 << i);
908 dlen = ROUNDUP(sa->sa_len);
915 * Build a routing message in a mbuf chain.
916 * Copy the addresses in the rtinfo->rti_info[] sockaddr array
917 * to the end of the mbuf after the message header.
919 * Set the rtinfo->rti_addrs bitmask of addresses present in rtinfo->rti_info[].
920 * This side-effect can be avoided if we reorder the addrs bitmask field in all
921 * the route messages to line up so we can set it here instead of back in the
925 rt_msg_mbuf(int type, struct rt_addrinfo *rtinfo)
928 struct rt_msghdr *rtm;
932 hlen = rt_msghdrsize(type);
933 KASSERT(hlen <= MCLBYTES, ("rt_msg_mbuf: hlen %d doesn't fit", hlen));
935 m = m_getl(hlen, MB_DONTWAIT, MT_DATA, M_PKTHDR, NULL);
939 m->m_pkthdr.len = m->m_len = hlen;
940 m->m_pkthdr.rcvif = NULL;
941 rtinfo->rti_addrs = 0;
943 for (i = 0; i < RTAX_MAX; i++) {
947 if ((sa = rtinfo->rti_info[i]) == NULL)
949 rtinfo->rti_addrs |= (1 << i);
950 dlen = ROUNDUP(sa->sa_len);
951 m_copyback(m, len, dlen, (caddr_t)sa); /* can grow mbuf chain */
954 if (m->m_pkthdr.len != len) { /* one of the m_copyback() calls failed */
958 rtm = mtod(m, struct rt_msghdr *);
960 rtm->rtm_msglen = len;
961 rtm->rtm_version = RTM_VERSION;
962 rtm->rtm_type = type;
967 * This routine is called to generate a message from the routing
968 * socket indicating that a redirect has occurred, a routing lookup
969 * has failed, or that a protocol has detected timeouts to a particular
973 rt_missmsg(int type, struct rt_addrinfo *rtinfo, int flags, int error)
975 struct sockaddr *dst = rtinfo->rti_info[RTAX_DST];
976 struct rt_msghdr *rtm;
979 if (route_cb.any_count == 0)
981 m = rt_msg_mbuf(type, rtinfo);
984 rtm = mtod(m, struct rt_msghdr *);
985 rtm->rtm_flags = RTF_DONE | flags;
986 rtm->rtm_errno = error;
987 rtm->rtm_addrs = rtinfo->rti_addrs;
988 rts_input(m, familyof(dst));
992 rt_dstmsg(int type, struct sockaddr *dst, int error)
994 struct rt_msghdr *rtm;
995 struct rt_addrinfo addrs;
998 if (route_cb.any_count == 0)
1000 bzero(&addrs, sizeof(struct rt_addrinfo));
1001 addrs.rti_info[RTAX_DST] = dst;
1002 m = rt_msg_mbuf(type, &addrs);
1005 rtm = mtod(m, struct rt_msghdr *);
1006 rtm->rtm_flags = RTF_DONE;
1007 rtm->rtm_errno = error;
1008 rtm->rtm_addrs = addrs.rti_addrs;
1009 rts_input(m, familyof(dst));
1013 * This routine is called to generate a message from the routing
1014 * socket indicating that the status of a network interface has changed.
1017 rt_ifmsg(struct ifnet *ifp)
1019 struct if_msghdr *ifm;
1021 struct rt_addrinfo rtinfo;
1023 if (route_cb.any_count == 0)
1025 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1026 m = rt_msg_mbuf(RTM_IFINFO, &rtinfo);
1029 ifm = mtod(m, struct if_msghdr *);
1030 ifm->ifm_index = ifp->if_index;
1031 ifm->ifm_flags = ifp->if_flags;
1032 ifm->ifm_data = ifp->if_data;
1038 rt_ifamsg(int cmd, struct ifaddr *ifa)
1040 struct ifa_msghdr *ifam;
1041 struct rt_addrinfo rtinfo;
1043 struct ifnet *ifp = ifa->ifa_ifp;
1045 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1046 rtinfo.rti_ifaaddr = ifa->ifa_addr;
1047 rtinfo.rti_ifpaddr =
1048 TAILQ_FIRST(&ifp->if_addrheads[mycpuid])->ifa->ifa_addr;
1049 rtinfo.rti_netmask = ifa->ifa_netmask;
1050 rtinfo.rti_bcastaddr = ifa->ifa_dstaddr;
1052 m = rt_msg_mbuf(cmd, &rtinfo);
1056 ifam = mtod(m, struct ifa_msghdr *);
1057 ifam->ifam_index = ifp->if_index;
1058 ifam->ifam_metric = ifa->ifa_metric;
1059 ifam->ifam_flags = ifa->ifa_flags;
1060 ifam->ifam_addrs = rtinfo.rti_addrs;
1062 rts_input(m, familyof(ifa->ifa_addr));
1066 rt_rtmsg(int cmd, struct rtentry *rt, struct ifnet *ifp, int error)
1068 struct rt_msghdr *rtm;
1069 struct rt_addrinfo rtinfo;
1071 struct sockaddr *dst;
1076 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1077 rtinfo.rti_dst = dst = rt_key(rt);
1078 rtinfo.rti_gateway = rt->rt_gateway;
1079 rtinfo.rti_netmask = rt_mask(rt);
1081 rtinfo.rti_ifpaddr =
1082 TAILQ_FIRST(&ifp->if_addrheads[mycpuid])->ifa->ifa_addr;
1084 rtinfo.rti_ifaaddr = rt->rt_ifa->ifa_addr;
1086 m = rt_msg_mbuf(cmd, &rtinfo);
1090 rtm = mtod(m, struct rt_msghdr *);
1092 rtm->rtm_index = ifp->if_index;
1093 rtm->rtm_flags |= rt->rt_flags;
1094 rtm->rtm_errno = error;
1095 rtm->rtm_addrs = rtinfo.rti_addrs;
1097 rts_input(m, familyof(dst));
1101 * This is called to generate messages from the routing socket
1102 * indicating a network interface has had addresses associated with it.
1103 * if we ever reverse the logic and replace messages TO the routing
1104 * socket indicate a request to configure interfaces, then it will
1105 * be unnecessary as the routing socket will automatically generate
1109 rt_newaddrmsg(int cmd, struct ifaddr *ifa, int error, struct rtentry *rt)
1113 * notify the SCTP stack
1114 * this will only get called when an address is added/deleted
1115 * XXX pass the ifaddr struct instead if ifa->ifa_addr...
1118 sctp_add_ip_address(ifa);
1119 else if (cmd == RTM_DELETE)
1120 sctp_delete_ip_address(ifa);
1123 if (route_cb.any_count == 0)
1126 if (cmd == RTM_ADD) {
1127 rt_ifamsg(RTM_NEWADDR, ifa);
1128 rt_rtmsg(RTM_ADD, rt, ifa->ifa_ifp, error);
1130 KASSERT((cmd == RTM_DELETE), ("unknown cmd %d", cmd));
1131 rt_rtmsg(RTM_DELETE, rt, ifa->ifa_ifp, error);
1132 rt_ifamsg(RTM_DELADDR, ifa);
1137 * This is the analogue to the rt_newaddrmsg which performs the same
1138 * function but for multicast group memberhips. This is easier since
1139 * there is no route state to worry about.
1142 rt_newmaddrmsg(int cmd, struct ifmultiaddr *ifma)
1144 struct rt_addrinfo rtinfo;
1145 struct mbuf *m = NULL;
1146 struct ifnet *ifp = ifma->ifma_ifp;
1147 struct ifma_msghdr *ifmam;
1149 if (route_cb.any_count == 0)
1152 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1153 rtinfo.rti_ifaaddr = ifma->ifma_addr;
1154 if (ifp != NULL && !TAILQ_EMPTY(&ifp->if_addrheads[mycpuid])) {
1155 rtinfo.rti_ifpaddr =
1156 TAILQ_FIRST(&ifp->if_addrheads[mycpuid])->ifa->ifa_addr;
1159 * If a link-layer address is present, present it as a ``gateway''
1160 * (similarly to how ARP entries, e.g., are presented).
1162 rtinfo.rti_gateway = ifma->ifma_lladdr;
1164 m = rt_msg_mbuf(cmd, &rtinfo);
1168 ifmam = mtod(m, struct ifma_msghdr *);
1169 ifmam->ifmam_index = ifp->if_index;
1170 ifmam->ifmam_addrs = rtinfo.rti_addrs;
1172 rts_input(m, familyof(ifma->ifma_addr));
1175 static struct mbuf *
1176 rt_makeifannouncemsg(struct ifnet *ifp, int type, int what,
1177 struct rt_addrinfo *info)
1179 struct if_announcemsghdr *ifan;
1182 if (route_cb.any_count == 0)
1185 bzero(info, sizeof(*info));
1186 m = rt_msg_mbuf(type, info);
1190 ifan = mtod(m, struct if_announcemsghdr *);
1191 ifan->ifan_index = ifp->if_index;
1192 strlcpy(ifan->ifan_name, ifp->if_xname, sizeof ifan->ifan_name);
1193 ifan->ifan_what = what;
1198 * This is called to generate routing socket messages indicating
1199 * IEEE80211 wireless events.
1200 * XXX we piggyback on the RTM_IFANNOUNCE msg format in a clumsy way.
1203 rt_ieee80211msg(struct ifnet *ifp, int what, void *data, size_t data_len)
1205 struct rt_addrinfo info;
1208 m = rt_makeifannouncemsg(ifp, RTM_IEEE80211, what, &info);
1213 * Append the ieee80211 data. Try to stick it in the
1214 * mbuf containing the ifannounce msg; otherwise allocate
1215 * a new mbuf and append.
1217 * NB: we assume m is a single mbuf.
1219 if (data_len > M_TRAILINGSPACE(m)) {
1220 /* XXX use m_getb(data_len, MB_DONTWAIT, MT_DATA, 0); */
1221 struct mbuf *n = m_get(MB_DONTWAIT, MT_DATA);
1226 KKASSERT(data_len <= M_TRAILINGSPACE(n));
1227 bcopy(data, mtod(n, void *), data_len);
1228 n->m_len = data_len;
1230 } else if (data_len > 0) {
1231 bcopy(data, mtod(m, u_int8_t *) + m->m_len, data_len);
1232 m->m_len += data_len;
1235 if (m->m_flags & M_PKTHDR)
1236 m->m_pkthdr.len += data_len;
1237 mtod(m, struct if_announcemsghdr *)->ifan_msglen += data_len;
1242 * This is called to generate routing socket messages indicating
1243 * network interface arrival and departure.
1246 rt_ifannouncemsg(struct ifnet *ifp, int what)
1248 struct rt_addrinfo addrinfo;
1251 m = rt_makeifannouncemsg(ifp, RTM_IFANNOUNCE, what, &addrinfo);
1257 resizewalkarg(struct walkarg *w, int len)
1261 newptr = kmalloc(len, M_RTABLE, M_INTWAIT | M_NULLOK);
1264 if (w->w_tmem != NULL)
1265 kfree(w->w_tmem, M_RTABLE);
1267 w->w_tmemsize = len;
1272 * This is used in dumping the kernel table via sysctl().
1275 sysctl_dumpentry(struct radix_node *rn, void *vw)
1277 struct walkarg *w = vw;
1278 struct rtentry *rt = (struct rtentry *)rn;
1279 struct rt_addrinfo rtinfo;
1282 if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg))
1285 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1286 rtinfo.rti_dst = rt_key(rt);
1287 rtinfo.rti_gateway = rt->rt_gateway;
1288 rtinfo.rti_netmask = rt_mask(rt);
1289 rtinfo.rti_genmask = rt->rt_genmask;
1290 if (rt->rt_ifp != NULL) {
1291 rtinfo.rti_ifpaddr =
1292 TAILQ_FIRST(&rt->rt_ifp->if_addrheads[mycpuid])->ifa->ifa_addr;
1293 rtinfo.rti_ifaaddr = rt->rt_ifa->ifa_addr;
1294 if (rt->rt_ifp->if_flags & IFF_POINTOPOINT)
1295 rtinfo.rti_bcastaddr = rt->rt_ifa->ifa_dstaddr;
1297 msglen = rt_msgsize(RTM_GET, &rtinfo);
1298 if (w->w_tmemsize < msglen && resizewalkarg(w, msglen) != 0)
1300 rt_msg_buffer(RTM_GET, &rtinfo, w->w_tmem, msglen);
1301 if (w->w_req != NULL) {
1302 struct rt_msghdr *rtm = w->w_tmem;
1304 rtm->rtm_flags = rt->rt_flags;
1305 rtm->rtm_use = rt->rt_use;
1306 rtm->rtm_rmx = rt->rt_rmx;
1307 rtm->rtm_index = rt->rt_ifp->if_index;
1308 rtm->rtm_errno = rtm->rtm_pid = rtm->rtm_seq = 0;
1309 rtm->rtm_addrs = rtinfo.rti_addrs;
1310 error = SYSCTL_OUT(w->w_req, rtm, msglen);
1317 sysctl_iflist(int af, struct walkarg *w)
1320 struct rt_addrinfo rtinfo;
1323 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1324 TAILQ_FOREACH(ifp, &ifnet, if_link) {
1325 struct ifaddr_container *ifac;
1328 if (w->w_arg && w->w_arg != ifp->if_index)
1330 ifac = TAILQ_FIRST(&ifp->if_addrheads[mycpuid]);
1332 rtinfo.rti_ifpaddr = ifa->ifa_addr;
1333 msglen = rt_msgsize(RTM_IFINFO, &rtinfo);
1334 if (w->w_tmemsize < msglen && resizewalkarg(w, msglen) != 0)
1336 rt_msg_buffer(RTM_IFINFO, &rtinfo, w->w_tmem, msglen);
1337 rtinfo.rti_ifpaddr = NULL;
1338 if (w->w_req != NULL && w->w_tmem != NULL) {
1339 struct if_msghdr *ifm = w->w_tmem;
1341 ifm->ifm_index = ifp->if_index;
1342 ifm->ifm_flags = ifp->if_flags;
1343 ifm->ifm_data = ifp->if_data;
1344 ifm->ifm_addrs = rtinfo.rti_addrs;
1345 error = SYSCTL_OUT(w->w_req, ifm, msglen);
1349 while ((ifac = TAILQ_NEXT(ifac, ifa_link)) != NULL) {
1352 if (af && af != ifa->ifa_addr->sa_family)
1354 if (curproc->p_ucred->cr_prison &&
1355 prison_if(curproc->p_ucred, ifa->ifa_addr))
1357 rtinfo.rti_ifaaddr = ifa->ifa_addr;
1358 rtinfo.rti_netmask = ifa->ifa_netmask;
1359 rtinfo.rti_bcastaddr = ifa->ifa_dstaddr;
1360 msglen = rt_msgsize(RTM_NEWADDR, &rtinfo);
1361 if (w->w_tmemsize < msglen &&
1362 resizewalkarg(w, msglen) != 0)
1364 rt_msg_buffer(RTM_NEWADDR, &rtinfo, w->w_tmem, msglen);
1365 if (w->w_req != NULL) {
1366 struct ifa_msghdr *ifam = w->w_tmem;
1368 ifam->ifam_index = ifa->ifa_ifp->if_index;
1369 ifam->ifam_flags = ifa->ifa_flags;
1370 ifam->ifam_metric = ifa->ifa_metric;
1371 ifam->ifam_addrs = rtinfo.rti_addrs;
1372 error = SYSCTL_OUT(w->w_req, w->w_tmem, msglen);
1377 rtinfo.rti_netmask = NULL;
1378 rtinfo.rti_ifaaddr = NULL;
1379 rtinfo.rti_bcastaddr = NULL;
1385 sysctl_rtsock(SYSCTL_HANDLER_ARGS)
1387 int *name = (int *)arg1;
1388 u_int namelen = arg2;
1389 struct radix_node_head *rnh;
1390 int i, error = EINVAL;
1399 if (namelen != 3 && namelen != 4)
1402 bzero(&w, sizeof w);
1408 * Optional third argument specifies cpu, used primarily for
1409 * debugging the route table.
1412 if (name[3] < 0 || name[3] >= ncpus)
1415 lwkt_migratecpu(name[3]);
1423 for (i = 1; i <= AF_MAX; i++)
1424 if ((rnh = rt_tables[mycpuid][i]) &&
1425 (af == 0 || af == i) &&
1426 (error = rnh->rnh_walktree(rnh,
1427 sysctl_dumpentry, &w)))
1432 error = sysctl_iflist(af, &w);
1435 if (w.w_tmem != NULL)
1436 kfree(w.w_tmem, M_RTABLE);
1438 lwkt_migratecpu(origcpu);
1442 SYSCTL_NODE(_net, PF_ROUTE, routetable, CTLFLAG_RD, sysctl_rtsock, "");
1445 * Definitions of protocols supported in the ROUTE domain.
1448 static struct domain routedomain; /* or at least forward */
1450 static struct protosw routesw[] = {
1452 .pr_type = SOCK_RAW,
1453 .pr_domain = &routedomain,
1455 .pr_flags = PR_ATOMIC|PR_ADDR,
1457 .pr_output = route_output,
1458 .pr_ctlinput = raw_ctlinput,
1459 .pr_ctloutput = NULL,
1460 .pr_ctlport = cpu0_ctlport,
1462 .pr_init = raw_init,
1463 .pr_usrreqs = &route_usrreqs
1467 static struct domain routedomain = {
1468 PF_ROUTE, "route", NULL, NULL, NULL,
1469 routesw, &routesw[(sizeof routesw)/(sizeof routesw[0])],