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 $
71 #include <sys/param.h>
72 #include <sys/systm.h>
73 #include <sys/kernel.h>
74 #include <sys/sysctl.h>
77 #include <sys/malloc.h>
79 #include <sys/protosw.h>
80 #include <sys/socket.h>
81 #include <sys/socketvar.h>
82 #include <sys/domain.h>
84 #include <sys/thread2.h>
85 #include <sys/socketvar2.h>
88 #include <net/route.h>
89 #include <net/raw_cb.h>
90 #include <net/netmsg2.h>
91 #include <net/netisr2.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 = netisr_cpuport(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!?");
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);
792 setmetric(RTV_IWMAXSEGS, rmx_iwmaxsegs);
793 setmetric(RTV_IWCAPSEGS, rmx_iwcapsegs);
798 ((a) > 0 ? (1 + (((a) - 1) | (sizeof(long) - 1))) : sizeof(long))
801 * Extract the addresses of the passed sockaddrs.
802 * Do a little sanity checking so as to avoid bad memory references.
803 * This data is derived straight from userland.
806 rt_xaddrs(char *cp, char *cplim, struct rt_addrinfo *rtinfo)
811 for (i = 0; (i < RTAX_MAX) && (cp < cplim); i++) {
812 if ((rtinfo->rti_addrs & (1 << i)) == 0)
814 sa = (struct sockaddr *)cp;
818 if ((cp + sa->sa_len) > cplim) {
823 * There are no more... Quit now.
824 * If there are more bits, they are in error.
825 * I've seen this. route(1) can evidently generate these.
826 * This causes kernel to core dump.
827 * For compatibility, if we see this, point to a safe address.
829 if (sa->sa_len == 0) {
830 static struct sockaddr sa_zero = {
831 sizeof sa_zero, AF_INET,
834 rtinfo->rti_info[i] = &sa_zero;
835 kprintf("rtsock: received more addr bits than sockaddrs.\n");
836 return (0); /* should be EINVAL but for compat */
839 /* Accept the sockaddr. */
840 rtinfo->rti_info[i] = sa;
841 cp += ROUNDUP(sa->sa_len);
847 rt_msghdrsize(int type)
852 return sizeof(struct ifa_msghdr);
855 return sizeof(struct ifma_msghdr);
857 return sizeof(struct if_msghdr);
860 return sizeof(struct if_announcemsghdr);
862 return sizeof(struct rt_msghdr);
867 rt_msgsize(int type, struct rt_addrinfo *rtinfo)
871 len = rt_msghdrsize(type);
872 for (i = 0; i < RTAX_MAX; i++) {
873 if (rtinfo->rti_info[i] != NULL)
874 len += ROUNDUP(rtinfo->rti_info[i]->sa_len);
881 * Build a routing message in a buffer.
882 * Copy the addresses in the rtinfo->rti_info[] sockaddr array
883 * to the end of the buffer after the message header.
885 * Set the rtinfo->rti_addrs bitmask of addresses present in rtinfo->rti_info[].
886 * This side-effect can be avoided if we reorder the addrs bitmask field in all
887 * the route messages to line up so we can set it here instead of back in the
891 rt_msg_buffer(int type, struct rt_addrinfo *rtinfo, void *buf, int msglen)
893 struct rt_msghdr *rtm;
897 rtm = (struct rt_msghdr *) buf;
898 rtm->rtm_version = RTM_VERSION;
899 rtm->rtm_type = type;
900 rtm->rtm_msglen = msglen;
902 cp = (char *)buf + rt_msghdrsize(type);
903 rtinfo->rti_addrs = 0;
904 for (i = 0; i < RTAX_MAX; i++) {
907 if ((sa = rtinfo->rti_info[i]) == NULL)
909 rtinfo->rti_addrs |= (1 << i);
910 dlen = ROUNDUP(sa->sa_len);
917 * Build a routing message in a mbuf chain.
918 * Copy the addresses in the rtinfo->rti_info[] sockaddr array
919 * to the end of the mbuf after the message header.
921 * Set the rtinfo->rti_addrs bitmask of addresses present in rtinfo->rti_info[].
922 * This side-effect can be avoided if we reorder the addrs bitmask field in all
923 * the route messages to line up so we can set it here instead of back in the
927 rt_msg_mbuf(int type, struct rt_addrinfo *rtinfo)
930 struct rt_msghdr *rtm;
934 hlen = rt_msghdrsize(type);
935 KASSERT(hlen <= MCLBYTES, ("rt_msg_mbuf: hlen %d doesn't fit", hlen));
937 m = m_getl(hlen, MB_DONTWAIT, MT_DATA, M_PKTHDR, NULL);
941 m->m_pkthdr.len = m->m_len = hlen;
942 m->m_pkthdr.rcvif = NULL;
943 rtinfo->rti_addrs = 0;
945 for (i = 0; i < RTAX_MAX; i++) {
949 if ((sa = rtinfo->rti_info[i]) == NULL)
951 rtinfo->rti_addrs |= (1 << i);
952 dlen = ROUNDUP(sa->sa_len);
953 m_copyback(m, len, dlen, (caddr_t)sa); /* can grow mbuf chain */
956 if (m->m_pkthdr.len != len) { /* one of the m_copyback() calls failed */
960 rtm = mtod(m, struct rt_msghdr *);
962 rtm->rtm_msglen = len;
963 rtm->rtm_version = RTM_VERSION;
964 rtm->rtm_type = type;
969 * This routine is called to generate a message from the routing
970 * socket indicating that a redirect has occurred, a routing lookup
971 * has failed, or that a protocol has detected timeouts to a particular
975 rt_missmsg(int type, struct rt_addrinfo *rtinfo, int flags, int error)
977 struct sockaddr *dst = rtinfo->rti_info[RTAX_DST];
978 struct rt_msghdr *rtm;
981 if (route_cb.any_count == 0)
983 m = rt_msg_mbuf(type, rtinfo);
986 rtm = mtod(m, struct rt_msghdr *);
987 rtm->rtm_flags = RTF_DONE | flags;
988 rtm->rtm_errno = error;
989 rtm->rtm_addrs = rtinfo->rti_addrs;
990 rts_input(m, familyof(dst));
994 rt_dstmsg(int type, struct sockaddr *dst, int error)
996 struct rt_msghdr *rtm;
997 struct rt_addrinfo addrs;
1000 if (route_cb.any_count == 0)
1002 bzero(&addrs, sizeof(struct rt_addrinfo));
1003 addrs.rti_info[RTAX_DST] = dst;
1004 m = rt_msg_mbuf(type, &addrs);
1007 rtm = mtod(m, struct rt_msghdr *);
1008 rtm->rtm_flags = RTF_DONE;
1009 rtm->rtm_errno = error;
1010 rtm->rtm_addrs = addrs.rti_addrs;
1011 rts_input(m, familyof(dst));
1015 * This routine is called to generate a message from the routing
1016 * socket indicating that the status of a network interface has changed.
1019 rt_ifmsg(struct ifnet *ifp)
1021 struct if_msghdr *ifm;
1023 struct rt_addrinfo rtinfo;
1025 if (route_cb.any_count == 0)
1027 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1028 m = rt_msg_mbuf(RTM_IFINFO, &rtinfo);
1031 ifm = mtod(m, struct if_msghdr *);
1032 ifm->ifm_index = ifp->if_index;
1033 ifm->ifm_flags = ifp->if_flags;
1034 ifm->ifm_data = ifp->if_data;
1040 rt_ifamsg(int cmd, struct ifaddr *ifa)
1042 struct ifa_msghdr *ifam;
1043 struct rt_addrinfo rtinfo;
1045 struct ifnet *ifp = ifa->ifa_ifp;
1047 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1048 rtinfo.rti_ifaaddr = ifa->ifa_addr;
1049 rtinfo.rti_ifpaddr =
1050 TAILQ_FIRST(&ifp->if_addrheads[mycpuid])->ifa->ifa_addr;
1051 rtinfo.rti_netmask = ifa->ifa_netmask;
1052 rtinfo.rti_bcastaddr = ifa->ifa_dstaddr;
1054 m = rt_msg_mbuf(cmd, &rtinfo);
1058 ifam = mtod(m, struct ifa_msghdr *);
1059 ifam->ifam_index = ifp->if_index;
1060 ifam->ifam_metric = ifa->ifa_metric;
1061 ifam->ifam_flags = ifa->ifa_flags;
1062 ifam->ifam_addrs = rtinfo.rti_addrs;
1064 rts_input(m, familyof(ifa->ifa_addr));
1068 rt_rtmsg(int cmd, struct rtentry *rt, struct ifnet *ifp, int error)
1070 struct rt_msghdr *rtm;
1071 struct rt_addrinfo rtinfo;
1073 struct sockaddr *dst;
1078 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1079 rtinfo.rti_dst = dst = rt_key(rt);
1080 rtinfo.rti_gateway = rt->rt_gateway;
1081 rtinfo.rti_netmask = rt_mask(rt);
1083 rtinfo.rti_ifpaddr =
1084 TAILQ_FIRST(&ifp->if_addrheads[mycpuid])->ifa->ifa_addr;
1086 rtinfo.rti_ifaaddr = rt->rt_ifa->ifa_addr;
1088 m = rt_msg_mbuf(cmd, &rtinfo);
1092 rtm = mtod(m, struct rt_msghdr *);
1094 rtm->rtm_index = ifp->if_index;
1095 rtm->rtm_flags |= rt->rt_flags;
1096 rtm->rtm_errno = error;
1097 rtm->rtm_addrs = rtinfo.rti_addrs;
1099 rts_input(m, familyof(dst));
1103 * This is called to generate messages from the routing socket
1104 * indicating a network interface has had addresses associated with it.
1105 * if we ever reverse the logic and replace messages TO the routing
1106 * socket indicate a request to configure interfaces, then it will
1107 * be unnecessary as the routing socket will automatically generate
1111 rt_newaddrmsg(int cmd, struct ifaddr *ifa, int error, struct rtentry *rt)
1115 * notify the SCTP stack
1116 * this will only get called when an address is added/deleted
1117 * XXX pass the ifaddr struct instead if ifa->ifa_addr...
1120 sctp_add_ip_address(ifa);
1121 else if (cmd == RTM_DELETE)
1122 sctp_delete_ip_address(ifa);
1125 if (route_cb.any_count == 0)
1128 if (cmd == RTM_ADD) {
1129 rt_ifamsg(RTM_NEWADDR, ifa);
1130 rt_rtmsg(RTM_ADD, rt, ifa->ifa_ifp, error);
1132 KASSERT((cmd == RTM_DELETE), ("unknown cmd %d", cmd));
1133 rt_rtmsg(RTM_DELETE, rt, ifa->ifa_ifp, error);
1134 rt_ifamsg(RTM_DELADDR, ifa);
1139 * This is the analogue to the rt_newaddrmsg which performs the same
1140 * function but for multicast group memberhips. This is easier since
1141 * there is no route state to worry about.
1144 rt_newmaddrmsg(int cmd, struct ifmultiaddr *ifma)
1146 struct rt_addrinfo rtinfo;
1147 struct mbuf *m = NULL;
1148 struct ifnet *ifp = ifma->ifma_ifp;
1149 struct ifma_msghdr *ifmam;
1151 if (route_cb.any_count == 0)
1154 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1155 rtinfo.rti_ifaaddr = ifma->ifma_addr;
1156 if (ifp != NULL && !TAILQ_EMPTY(&ifp->if_addrheads[mycpuid])) {
1157 rtinfo.rti_ifpaddr =
1158 TAILQ_FIRST(&ifp->if_addrheads[mycpuid])->ifa->ifa_addr;
1161 * If a link-layer address is present, present it as a ``gateway''
1162 * (similarly to how ARP entries, e.g., are presented).
1164 rtinfo.rti_gateway = ifma->ifma_lladdr;
1166 m = rt_msg_mbuf(cmd, &rtinfo);
1170 ifmam = mtod(m, struct ifma_msghdr *);
1171 ifmam->ifmam_index = ifp->if_index;
1172 ifmam->ifmam_addrs = rtinfo.rti_addrs;
1174 rts_input(m, familyof(ifma->ifma_addr));
1177 static struct mbuf *
1178 rt_makeifannouncemsg(struct ifnet *ifp, int type, int what,
1179 struct rt_addrinfo *info)
1181 struct if_announcemsghdr *ifan;
1184 if (route_cb.any_count == 0)
1187 bzero(info, sizeof(*info));
1188 m = rt_msg_mbuf(type, info);
1192 ifan = mtod(m, struct if_announcemsghdr *);
1193 ifan->ifan_index = ifp->if_index;
1194 strlcpy(ifan->ifan_name, ifp->if_xname, sizeof ifan->ifan_name);
1195 ifan->ifan_what = what;
1200 * This is called to generate routing socket messages indicating
1201 * IEEE80211 wireless events.
1202 * XXX we piggyback on the RTM_IFANNOUNCE msg format in a clumsy way.
1205 rt_ieee80211msg(struct ifnet *ifp, int what, void *data, size_t data_len)
1207 struct rt_addrinfo info;
1210 m = rt_makeifannouncemsg(ifp, RTM_IEEE80211, what, &info);
1215 * Append the ieee80211 data. Try to stick it in the
1216 * mbuf containing the ifannounce msg; otherwise allocate
1217 * a new mbuf and append.
1219 * NB: we assume m is a single mbuf.
1221 if (data_len > M_TRAILINGSPACE(m)) {
1222 /* XXX use m_getb(data_len, MB_DONTWAIT, MT_DATA, 0); */
1223 struct mbuf *n = m_get(MB_DONTWAIT, MT_DATA);
1228 KKASSERT(data_len <= M_TRAILINGSPACE(n));
1229 bcopy(data, mtod(n, void *), data_len);
1230 n->m_len = data_len;
1232 } else if (data_len > 0) {
1233 bcopy(data, mtod(m, u_int8_t *) + m->m_len, data_len);
1234 m->m_len += data_len;
1237 if (m->m_flags & M_PKTHDR)
1238 m->m_pkthdr.len += data_len;
1239 mtod(m, struct if_announcemsghdr *)->ifan_msglen += data_len;
1244 * This is called to generate routing socket messages indicating
1245 * network interface arrival and departure.
1248 rt_ifannouncemsg(struct ifnet *ifp, int what)
1250 struct rt_addrinfo addrinfo;
1253 m = rt_makeifannouncemsg(ifp, RTM_IFANNOUNCE, what, &addrinfo);
1259 resizewalkarg(struct walkarg *w, int len)
1263 newptr = kmalloc(len, M_RTABLE, M_INTWAIT | M_NULLOK);
1266 if (w->w_tmem != NULL)
1267 kfree(w->w_tmem, M_RTABLE);
1269 w->w_tmemsize = len;
1274 * This is used in dumping the kernel table via sysctl().
1277 sysctl_dumpentry(struct radix_node *rn, void *vw)
1279 struct walkarg *w = vw;
1280 struct rtentry *rt = (struct rtentry *)rn;
1281 struct rt_addrinfo rtinfo;
1284 if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg))
1287 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1288 rtinfo.rti_dst = rt_key(rt);
1289 rtinfo.rti_gateway = rt->rt_gateway;
1290 rtinfo.rti_netmask = rt_mask(rt);
1291 rtinfo.rti_genmask = rt->rt_genmask;
1292 if (rt->rt_ifp != NULL) {
1293 rtinfo.rti_ifpaddr =
1294 TAILQ_FIRST(&rt->rt_ifp->if_addrheads[mycpuid])->ifa->ifa_addr;
1295 rtinfo.rti_ifaaddr = rt->rt_ifa->ifa_addr;
1296 if (rt->rt_ifp->if_flags & IFF_POINTOPOINT)
1297 rtinfo.rti_bcastaddr = rt->rt_ifa->ifa_dstaddr;
1299 msglen = rt_msgsize(RTM_GET, &rtinfo);
1300 if (w->w_tmemsize < msglen && resizewalkarg(w, msglen) != 0)
1302 rt_msg_buffer(RTM_GET, &rtinfo, w->w_tmem, msglen);
1303 if (w->w_req != NULL) {
1304 struct rt_msghdr *rtm = w->w_tmem;
1306 rtm->rtm_flags = rt->rt_flags;
1307 rtm->rtm_use = rt->rt_use;
1308 rtm->rtm_rmx = rt->rt_rmx;
1309 rtm->rtm_index = rt->rt_ifp->if_index;
1310 rtm->rtm_errno = rtm->rtm_pid = rtm->rtm_seq = 0;
1311 rtm->rtm_addrs = rtinfo.rti_addrs;
1312 error = SYSCTL_OUT(w->w_req, rtm, msglen);
1319 sysctl_iflist(int af, struct walkarg *w)
1322 struct rt_addrinfo rtinfo;
1325 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1326 TAILQ_FOREACH(ifp, &ifnet, if_link) {
1327 struct ifaddr_container *ifac;
1330 if (w->w_arg && w->w_arg != ifp->if_index)
1332 ifac = TAILQ_FIRST(&ifp->if_addrheads[mycpuid]);
1334 rtinfo.rti_ifpaddr = ifa->ifa_addr;
1335 msglen = rt_msgsize(RTM_IFINFO, &rtinfo);
1336 if (w->w_tmemsize < msglen && resizewalkarg(w, msglen) != 0)
1338 rt_msg_buffer(RTM_IFINFO, &rtinfo, w->w_tmem, msglen);
1339 rtinfo.rti_ifpaddr = NULL;
1340 if (w->w_req != NULL && w->w_tmem != NULL) {
1341 struct if_msghdr *ifm = w->w_tmem;
1343 ifm->ifm_index = ifp->if_index;
1344 ifm->ifm_flags = ifp->if_flags;
1345 ifm->ifm_data = ifp->if_data;
1346 ifm->ifm_addrs = rtinfo.rti_addrs;
1347 error = SYSCTL_OUT(w->w_req, ifm, msglen);
1351 while ((ifac = TAILQ_NEXT(ifac, ifa_link)) != NULL) {
1354 if (af && af != ifa->ifa_addr->sa_family)
1356 if (curproc->p_ucred->cr_prison &&
1357 prison_if(curproc->p_ucred, ifa->ifa_addr))
1359 rtinfo.rti_ifaaddr = ifa->ifa_addr;
1360 rtinfo.rti_netmask = ifa->ifa_netmask;
1361 rtinfo.rti_bcastaddr = ifa->ifa_dstaddr;
1362 msglen = rt_msgsize(RTM_NEWADDR, &rtinfo);
1363 if (w->w_tmemsize < msglen &&
1364 resizewalkarg(w, msglen) != 0)
1366 rt_msg_buffer(RTM_NEWADDR, &rtinfo, w->w_tmem, msglen);
1367 if (w->w_req != NULL) {
1368 struct ifa_msghdr *ifam = w->w_tmem;
1370 ifam->ifam_index = ifa->ifa_ifp->if_index;
1371 ifam->ifam_flags = ifa->ifa_flags;
1372 ifam->ifam_metric = ifa->ifa_metric;
1373 ifam->ifam_addrs = rtinfo.rti_addrs;
1374 error = SYSCTL_OUT(w->w_req, w->w_tmem, msglen);
1379 rtinfo.rti_netmask = NULL;
1380 rtinfo.rti_ifaaddr = NULL;
1381 rtinfo.rti_bcastaddr = NULL;
1387 sysctl_rtsock(SYSCTL_HANDLER_ARGS)
1389 int *name = (int *)arg1;
1390 u_int namelen = arg2;
1391 struct radix_node_head *rnh;
1392 int i, error = EINVAL;
1401 if (namelen != 3 && namelen != 4)
1404 bzero(&w, sizeof w);
1410 * Optional third argument specifies cpu, used primarily for
1411 * debugging the route table.
1414 if (name[3] < 0 || name[3] >= ncpus)
1417 lwkt_migratecpu(name[3]);
1425 for (i = 1; i <= AF_MAX; i++)
1426 if ((rnh = rt_tables[mycpuid][i]) &&
1427 (af == 0 || af == i) &&
1428 (error = rnh->rnh_walktree(rnh,
1429 sysctl_dumpentry, &w)))
1434 error = sysctl_iflist(af, &w);
1437 if (w.w_tmem != NULL)
1438 kfree(w.w_tmem, M_RTABLE);
1440 lwkt_migratecpu(origcpu);
1444 SYSCTL_NODE(_net, PF_ROUTE, routetable, CTLFLAG_RD, sysctl_rtsock, "");
1447 * Definitions of protocols supported in the ROUTE domain.
1450 static struct domain routedomain; /* or at least forward */
1452 static struct protosw routesw[] = {
1454 .pr_type = SOCK_RAW,
1455 .pr_domain = &routedomain,
1457 .pr_flags = PR_ATOMIC|PR_ADDR,
1459 .pr_output = route_output,
1460 .pr_ctlinput = raw_ctlinput,
1461 .pr_ctloutput = NULL,
1462 .pr_ctlport = cpu0_ctlport,
1464 .pr_init = raw_init,
1465 .pr_usrreqs = &route_usrreqs
1469 static struct domain routedomain = {
1470 PF_ROUTE, "route", NULL, NULL, NULL,
1471 routesw, &routesw[(sizeof routesw)/(sizeof routesw[0])],