rt_metrics: Add rmx_iw* to record user-set IW parameters
[dragonfly.git] / sys / net / rtsock.c
CommitLineData
f3ed2586
JH
1/*
2 * Copyright (c) 2004, 2005 The DragonFly Project. All rights reserved.
3 *
4 * This code is derived from software contributed to The DragonFly Project
5 * by Jeffrey M. Hsu.
6 *
7 * Redistribution and use in source and binary forms, with or without
8 * modification, are permitted provided that the following conditions
9 * are met:
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.
18 *
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
30 * SUCH DAMAGE.
31 */
32
984263bc
MD
33/*
34 * Copyright (c) 1988, 1991, 1993
35 * The Regents of the University of California. All rights reserved.
36 *
37 * Redistribution and use in source and binary forms, with or without
38 * modification, are permitted provided that the following conditions
39 * are met:
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.
52 *
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
63 * SUCH DAMAGE.
64 *
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 $
e3873585 67 * $DragonFly: src/sys/net/rtsock.c,v 1.45 2008/10/27 02:56:30 sephe Exp $
984263bc
MD
68 */
69
78812139
EN
70#include "opt_sctp.h"
71
984263bc
MD
72#include <sys/param.h>
73#include <sys/systm.h>
74#include <sys/kernel.h>
75#include <sys/sysctl.h>
76#include <sys/proc.h>
895c1f85 77#include <sys/priv.h>
984263bc
MD
78#include <sys/malloc.h>
79#include <sys/mbuf.h>
47e78ae8 80#include <sys/protosw.h>
984263bc
MD
81#include <sys/socket.h>
82#include <sys/socketvar.h>
83#include <sys/domain.h>
6cef7136 84
a49aa710 85#include <sys/thread2.h>
6cef7136 86#include <sys/socketvar2.h>
deffea2e 87
984263bc
MD
88#include <net/if.h>
89#include <net/route.h>
90#include <net/raw_cb.h>
47db7f9b 91#include <net/netmsg2.h>
984263bc 92
78812139
EN
93#ifdef SCTP
94extern void sctp_add_ip_address(struct ifaddr *ifa);
95extern void sctp_delete_ip_address(struct ifaddr *ifa);
96#endif /* SCTP */
97
984263bc
MD
98MALLOC_DEFINE(M_RTABLE, "routetbl", "routing tables");
99
f23061d4
JH
100static struct route_cb {
101 int ip_count;
102 int ip6_count;
103 int ipx_count;
104 int ns_count;
105 int any_count;
106} route_cb;
107
590b8cd4 108static const struct sockaddr route_src = { 2, PF_ROUTE, };
984263bc
MD
109
110struct walkarg {
111 int w_tmemsize;
112 int w_op, w_arg;
8e63efec 113 void *w_tmem;
984263bc
MD
114 struct sysctl_req *w_req;
115};
116
117static struct mbuf *
590b8cd4
JH
118 rt_msg_mbuf (int, struct rt_addrinfo *);
119static void rt_msg_buffer (int, struct rt_addrinfo *, void *buf, int len);
120static int rt_msgsize (int type, struct rt_addrinfo *rtinfo);
ef87f48d 121static int rt_xaddrs (char *, char *, struct rt_addrinfo *);
158abb01
RG
122static int sysctl_dumpentry (struct radix_node *rn, void *vw);
123static int sysctl_iflist (int af, struct walkarg *w);
deffea2e 124static int route_output(struct mbuf *, struct socket *, ...);
47e78ae8 125static void rt_setmetrics (u_long, struct rt_metrics *,
f23061d4 126 struct rt_metrics *);
984263bc 127
984263bc
MD
128/*
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
131 */
002c1265
MD
132static void
133rts_abort(netmsg_t msg)
984263bc 134{
a49aa710 135 crit_enter();
002c1265
MD
136 raw_usrreqs.pru_abort(msg);
137 /* msg invalid now */
a49aa710 138 crit_exit();
984263bc
MD
139}
140
141/* pru_accept is EOPNOTSUPP */
142
002c1265
MD
143static void
144rts_attach(netmsg_t msg)
984263bc 145{
002c1265
MD
146 struct socket *so = msg->base.nm_so;
147 struct pru_attach_info *ai = msg->attach.nm_ai;
984263bc 148 struct rawcb *rp;
002c1265 149 int proto = msg->attach.nm_proto;
a49aa710 150 int error;
984263bc 151
002c1265
MD
152 crit_enter();
153 if (sotorawcb(so) != NULL) {
154 error = EISCONN;
155 goto done;
156 }
f23061d4 157
efda3bd0 158 rp = kmalloc(sizeof *rp, M_PCB, M_WAITOK | M_ZERO);
984263bc
MD
159
160 /*
a49aa710 161 * The critical section is necessary to block protocols from sending
984263bc
MD
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
a49aa710 165 * eliminate the critical section.
984263bc 166 */
ef87f48d 167 so->so_pcb = rp;
6cef7136 168 soreference(so); /* so_pcb assignment */
e4700d00 169 error = raw_attach(so, proto, ai->sb_rlimit);
984263bc
MD
170 rp = sotorawcb(so);
171 if (error) {
efda3bd0 172 kfree(rp, M_PCB);
002c1265 173 goto done;
984263bc
MD
174 }
175 switch(rp->rcb_proto.sp_protocol) {
176 case AF_INET:
177 route_cb.ip_count++;
178 break;
179 case AF_INET6:
180 route_cb.ip6_count++;
181 break;
182 case AF_IPX:
183 route_cb.ipx_count++;
184 break;
185 case AF_NS:
186 route_cb.ns_count++;
187 break;
188 }
189 rp->rcb_faddr = &route_src;
190 route_cb.any_count++;
191 soisconnected(so);
192 so->so_options |= SO_USELOOPBACK;
002c1265
MD
193 error = 0;
194done:
a49aa710 195 crit_exit();
002c1265 196 lwkt_replymsg(&msg->lmsg, error);
984263bc
MD
197}
198
002c1265
MD
199static void
200rts_bind(netmsg_t msg)
984263bc 201{
a49aa710 202 crit_enter();
002c1265
MD
203 raw_usrreqs.pru_bind(msg); /* xxx just EINVAL */
204 /* msg invalid now */
a49aa710 205 crit_exit();
984263bc
MD
206}
207
002c1265
MD
208static void
209rts_connect(netmsg_t msg)
984263bc 210{
a49aa710 211 crit_enter();
002c1265
MD
212 raw_usrreqs.pru_connect(msg); /* XXX just EINVAL */
213 /* msg invalid now */
a49aa710 214 crit_exit();
984263bc
MD
215}
216
217/* pru_connect2 is EOPNOTSUPP */
218/* pru_control is EOPNOTSUPP */
219
002c1265
MD
220static void
221rts_detach(netmsg_t msg)
984263bc 222{
002c1265 223 struct socket *so = msg->base.nm_so;
984263bc 224 struct rawcb *rp = sotorawcb(so);
984263bc 225
a49aa710 226 crit_enter();
ef87f48d 227 if (rp != NULL) {
984263bc
MD
228 switch(rp->rcb_proto.sp_protocol) {
229 case AF_INET:
230 route_cb.ip_count--;
231 break;
232 case AF_INET6:
233 route_cb.ip6_count--;
234 break;
235 case AF_IPX:
236 route_cb.ipx_count--;
237 break;
238 case AF_NS:
239 route_cb.ns_count--;
240 break;
241 }
242 route_cb.any_count--;
243 }
002c1265
MD
244 raw_usrreqs.pru_detach(msg);
245 /* msg invalid now */
a49aa710 246 crit_exit();
984263bc
MD
247}
248
002c1265
MD
249static void
250rts_disconnect(netmsg_t msg)
984263bc 251{
a49aa710 252 crit_enter();
002c1265
MD
253 raw_usrreqs.pru_disconnect(msg);
254 /* msg invalid now */
a49aa710 255 crit_exit();
984263bc
MD
256}
257
258/* pru_listen is EOPNOTSUPP */
259
002c1265
MD
260static void
261rts_peeraddr(netmsg_t msg)
984263bc 262{
a49aa710 263 crit_enter();
002c1265
MD
264 raw_usrreqs.pru_peeraddr(msg);
265 /* msg invalid now */
a49aa710 266 crit_exit();
984263bc
MD
267}
268
269/* pru_rcvd is EOPNOTSUPP */
270/* pru_rcvoob is EOPNOTSUPP */
271
002c1265
MD
272static void
273rts_send(netmsg_t msg)
984263bc 274{
a49aa710 275 crit_enter();
002c1265
MD
276 raw_usrreqs.pru_send(msg);
277 /* msg invalid now */
a49aa710 278 crit_exit();
984263bc
MD
279}
280
281/* pru_sense is null */
282
002c1265
MD
283static void
284rts_shutdown(netmsg_t msg)
984263bc 285{
a49aa710 286 crit_enter();
002c1265
MD
287 raw_usrreqs.pru_shutdown(msg);
288 /* msg invalid now */
a49aa710 289 crit_exit();
984263bc
MD
290}
291
002c1265
MD
292static void
293rts_sockaddr(netmsg_t msg)
984263bc 294{
a49aa710 295 crit_enter();
002c1265
MD
296 raw_usrreqs.pru_sockaddr(msg);
297 /* msg invalid now */
a49aa710 298 crit_exit();
984263bc
MD
299}
300
301static struct pr_usrreqs route_usrreqs = {
fa5e758c 302 .pru_abort = rts_abort,
002c1265 303 .pru_accept = pr_generic_notsupp,
fa5e758c
MD
304 .pru_attach = rts_attach,
305 .pru_bind = rts_bind,
306 .pru_connect = rts_connect,
002c1265
MD
307 .pru_connect2 = pr_generic_notsupp,
308 .pru_control = pr_generic_notsupp,
fa5e758c
MD
309 .pru_detach = rts_detach,
310 .pru_disconnect = rts_disconnect,
002c1265 311 .pru_listen = pr_generic_notsupp,
fa5e758c 312 .pru_peeraddr = rts_peeraddr,
002c1265
MD
313 .pru_rcvd = pr_generic_notsupp,
314 .pru_rcvoob = pr_generic_notsupp,
fa5e758c
MD
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,
8b5c39bb 320 .pru_soreceive = soreceive
984263bc
MD
321};
322
590b8cd4
JH
323static __inline sa_family_t
324familyof(struct sockaddr *sa)
325{
326 return (sa != NULL ? sa->sa_family : 0);
327}
328
47db7f9b
MD
329/*
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.
332 *
333 * This looks messy but it means that anyone, including interrupt code,
334 * can send a message to the routing socket.
335 */
590b8cd4 336static void
002c1265 337rts_input_handler(netmsg_t msg)
590b8cd4
JH
338{
339 static const struct sockaddr route_dst = { 2, PF_ROUTE, };
47db7f9b 340 struct sockproto route_proto;
002c1265 341 struct netmsg_packet *pmsg = &msg->packet;
47db7f9b
MD
342 struct mbuf *m;
343 sa_family_t family;
d16fa201 344 struct rawcb *skip;
47db7f9b 345
002c1265 346 family = pmsg->base.lmsg.u.ms_result;
47db7f9b
MD
347 route_proto.sp_family = PF_ROUTE;
348 route_proto.sp_protocol = family;
590b8cd4 349
d16fa201
SZ
350 m = pmsg->nm_packet;
351 M_ASSERTPKTHDR(m);
352
353 skip = m->m_pkthdr.header;
354 m->m_pkthdr.header = NULL;
355
356 raw_input(m, &route_proto, &route_src, &route_dst, skip);
590b8cd4
JH
357}
358
47db7f9b 359static void
d16fa201 360rts_input_skip(struct mbuf *m, sa_family_t family, struct rawcb *skip)
47db7f9b
MD
361{
362 struct netmsg_packet *pmsg;
363 lwkt_port_t port;
364
d16fa201
SZ
365 M_ASSERTPKTHDR(m);
366
002c1265 367 port = cpu_portfn(0); /* XXX same as for routing socket */
47db7f9b 368 pmsg = &m->m_hdr.mh_netmsg;
002c1265 369 netmsg_init(&pmsg->base, NULL, &netisr_apanic_rport,
47db7f9b
MD
370 0, rts_input_handler);
371 pmsg->nm_packet = m;
002c1265 372 pmsg->base.lmsg.u.ms_result = family;
d16fa201 373 m->m_pkthdr.header = skip; /* XXX steal field in pkthdr */
002c1265 374 lwkt_sendmsg(port, &pmsg->base.lmsg);
47db7f9b
MD
375}
376
d16fa201
SZ
377static __inline void
378rts_input(struct mbuf *m, sa_family_t family)
379{
380 rts_input_skip(m, family, NULL);
381}
382
590b8cd4 383static void *
6e6b3e50 384reallocbuf_nofree(void *ptr, size_t len, size_t olen)
590b8cd4
JH
385{
386 void *newptr;
387
efda3bd0 388 newptr = kmalloc(len, M_RTABLE, M_INTWAIT | M_NULLOK);
590b8cd4
JH
389 if (newptr == NULL)
390 return NULL;
391 bcopy(ptr, newptr, olen);
590b8cd4
JH
392 return (newptr);
393}
394
8e63efec
JH
395/*
396 * Internal helper routine for route_output().
397 */
590b8cd4 398static int
6e6b3e50
SZ
399_fillrtmsg(struct rt_msghdr **prtm, struct rtentry *rt,
400 struct rt_addrinfo *rtinfo)
590b8cd4
JH
401{
402 int msglen;
403 struct rt_msghdr *rtm = *prtm;
404
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 =
b2632176
SZ
413 TAILQ_FIRST(&rt->rt_ifp->if_addrheads[mycpuid])
414 ->ifa->ifa_addr;
590b8cd4
JH
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;
419 } else {
420 rtinfo->rti_ifpaddr = NULL;
421 rtinfo->rti_ifaaddr = NULL;
1730f0e1 422 }
9d1f1679
SZ
423 } else if (rt->rt_ifp != NULL) {
424 rtm->rtm_index = rt->rt_ifp->if_index;
590b8cd4
JH
425 }
426
427 msglen = rt_msgsize(rtm->rtm_type, rtinfo);
428 if (rtm->rtm_msglen < msglen) {
6e6b3e50
SZ
429 /* NOTE: Caller will free the old rtm accordingly */
430 rtm = reallocbuf_nofree(rtm, msglen, rtm->rtm_msglen);
590b8cd4
JH
431 if (rtm == NULL)
432 return (ENOBUFS);
433 *prtm = rtm;
434 }
435 rt_msg_buffer(rtm->rtm_type, rtinfo, rtm, msglen);
436
437 rtm->rtm_flags = rt->rt_flags;
438 rtm->rtm_rmx = rt->rt_rmx;
439 rtm->rtm_addrs = rtinfo->rti_addrs;
440
441 return (0);
442}
443
6e6b3e50
SZ
444struct rtm_arg {
445 struct rt_msghdr *bak_rtm;
446 struct rt_msghdr *new_rtm;
447};
448
449static int
450fillrtmsg(struct rtm_arg *arg, struct rtentry *rt,
451 struct rt_addrinfo *rtinfo)
452{
453 struct rt_msghdr *rtm = arg->new_rtm;
454 int error;
455
456 error = _fillrtmsg(&rtm, rt, rtinfo);
457 if (!error) {
458 if (arg->new_rtm != rtm) {
459 /*
460 * _fillrtmsg() just allocated a new rtm;
461 * if the previously allocated rtm is not
462 * the backing rtm, it should be freed.
463 */
464 if (arg->new_rtm != arg->bak_rtm)
465 kfree(arg->new_rtm, M_RTABLE);
466 arg->new_rtm = rtm;
467 }
468 }
469 return error;
470}
471
ecdefdda
MD
472static void route_output_add_callback(int, int, struct rt_addrinfo *,
473 struct rtentry *, void *);
474static void route_output_delete_callback(int, int, struct rt_addrinfo *,
475 struct rtentry *, void *);
00104f14 476static int route_output_get_callback(int, struct rt_addrinfo *,
b5e01dd3 477 struct rtentry *, void *, int);
2c62ae3b 478static int route_output_change_callback(int, struct rt_addrinfo *,
b5e01dd3 479 struct rtentry *, void *, int);
ae2a5ccd 480static int route_output_lock_callback(int, struct rt_addrinfo *,
b5e01dd3 481 struct rtentry *, void *, int);
ecdefdda 482
984263bc
MD
483/*ARGSUSED*/
484static int
deffea2e 485route_output(struct mbuf *m, struct socket *so, ...)
984263bc 486{
6e6b3e50 487 struct rtm_arg arg;
2e9572df 488 struct rt_msghdr *rtm = NULL;
f23061d4 489 struct rawcb *rp = NULL;
deffea2e 490 struct pr_output_info *oi;
590b8cd4 491 struct rt_addrinfo rtinfo;
a8c218b2 492 sa_family_t family;
ef87f48d 493 int len, error = 0;
deffea2e
JS
494 __va_list ap;
495
1cfcc9b6
SZ
496 M_ASSERTPKTHDR(m);
497
deffea2e
JS
498 __va_start(ap, so);
499 oi = __va_arg(ap, struct pr_output_info *);
500 __va_end(ap);
984263bc 501
a8c218b2
SZ
502 family = familyof(NULL);
503
ef87f48d 504#define gotoerr(e) { error = e; goto flush;}
590b8cd4
JH
505
506 if (m == NULL ||
507 (m->m_len < sizeof(long) &&
508 (m = m_pullup(m, sizeof(long))) == NULL))
984263bc 509 return (ENOBUFS);
984263bc 510 len = m->m_pkthdr.len;
590b8cd4 511 if (len < sizeof(struct rt_msghdr) ||
a8c218b2 512 len != mtod(m, struct rt_msghdr *)->rtm_msglen)
ef87f48d 513 gotoerr(EINVAL);
a8c218b2 514
efda3bd0 515 rtm = kmalloc(len, M_RTABLE, M_INTWAIT | M_NULLOK);
a8c218b2 516 if (rtm == NULL)
ef87f48d 517 gotoerr(ENOBUFS);
a8c218b2 518
984263bc 519 m_copydata(m, 0, len, (caddr_t)rtm);
a8c218b2 520 if (rtm->rtm_version != RTM_VERSION)
ef87f48d 521 gotoerr(EPROTONOSUPPORT);
a8c218b2 522
47e78ae8 523 rtm->rtm_pid = oi->p_pid;
590b8cd4
JH
524 bzero(&rtinfo, sizeof(struct rt_addrinfo));
525 rtinfo.rti_addrs = rtm->rtm_addrs;
a8c218b2 526 if (rt_xaddrs((char *)(rtm + 1), (char *)rtm + len, &rtinfo) != 0)
ef87f48d 527 gotoerr(EINVAL);
a8c218b2 528
590b8cd4
JH
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))
ef87f48d
JH
532 gotoerr(EINVAL);
533
a8c218b2
SZ
534 family = familyof(rtinfo.rti_dst);
535
590b8cd4 536 if (rtinfo.rti_genmask != NULL) {
ac0f13c5
SZ
537 error = rtmask_add_global(rtinfo.rti_genmask);
538 if (error)
539 goto flush;
984263bc
MD
540 }
541
542 /*
543 * Verify that the caller has the appropriate privilege; RTM_GET
544 * is the only operation the non-superuser is allowed.
545 */
02650706
SZ
546 if (rtm->rtm_type != RTM_GET &&
547 priv_check_cred(so->so_cred, PRIV_ROOT, 0) != 0)
ef87f48d 548 gotoerr(EPERM);
984263bc
MD
549
550 switch (rtm->rtm_type) {
984263bc 551 case RTM_ADD:
ecdefdda
MD
552 if (rtinfo.rti_gateway == NULL) {
553 error = EINVAL;
554 } else {
555 error = rtrequest1_global(RTM_ADD, &rtinfo,
556 route_output_add_callback, rtm);
984263bc
MD
557 }
558 break;
984263bc 559 case RTM_DELETE:
ecdefdda 560 /*
6e6b3e50
SZ
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.
568 *
569 * new_rtm will be used to save the new rtm allocated
570 * by rtrequest1_global or rtsearch_global.
ecdefdda 571 */
6e6b3e50
SZ
572 arg.bak_rtm = rtm;
573 arg.new_rtm = rtm;
ecdefdda 574 error = rtrequest1_global(RTM_DELETE, &rtinfo,
6e6b3e50
SZ
575 route_output_delete_callback, &arg);
576 rtm = arg.new_rtm;
577 if (rtm != arg.bak_rtm)
578 kfree(arg.bak_rtm, M_RTABLE);
984263bc 579 break;
984263bc 580 case RTM_GET:
6e6b3e50
SZ
581 /* See the comment in RTM_DELETE */
582 arg.bak_rtm = rtm;
583 arg.new_rtm = rtm;
00104f14 584 error = rtsearch_global(RTM_GET, &rtinfo,
6e6b3e50 585 route_output_get_callback, &arg,
00104f14 586 RTS_NOEXACTMATCH);
6e6b3e50
SZ
587 rtm = arg.new_rtm;
588 if (rtm != arg.bak_rtm)
589 kfree(arg.bak_rtm, M_RTABLE);
ecdefdda
MD
590 break;
591 case RTM_CHANGE:
2c62ae3b
SZ
592 error = rtsearch_global(RTM_CHANGE, &rtinfo,
593 route_output_change_callback, rtm,
594 RTS_EXACTMATCH);
ecdefdda
MD
595 break;
596 case RTM_LOCK:
ae2a5ccd
SZ
597 error = rtsearch_global(RTM_LOCK, &rtinfo,
598 route_output_lock_callback, rtm,
599 RTS_EXACTMATCH);
590b8cd4 600 break;
984263bc 601 default:
ecdefdda
MD
602 error = EOPNOTSUPP;
603 break;
984263bc 604 }
984263bc 605flush:
590b8cd4
JH
606 if (rtm != NULL) {
607 if (error != 0)
984263bc
MD
608 rtm->rtm_errno = error;
609 else
610 rtm->rtm_flags |= RTF_DONE;
611 }
ecdefdda 612
984263bc
MD
613 /*
614 * Check to see if we don't want our own messages.
615 */
ef87f48d 616 if (!(so->so_options & SO_USELOOPBACK)) {
984263bc 617 if (route_cb.any_count <= 1) {
590b8cd4 618 if (rtm != NULL)
efda3bd0 619 kfree(rtm, M_RTABLE);
984263bc
MD
620 m_freem(m);
621 return (error);
622 }
623 /* There is another listener, so construct message */
624 rp = sotorawcb(so);
625 }
590b8cd4 626 if (rtm != NULL) {
984263bc
MD
627 m_copyback(m, 0, rtm->rtm_msglen, (caddr_t)rtm);
628 if (m->m_pkthdr.len < rtm->rtm_msglen) {
629 m_freem(m);
630 m = NULL;
631 } else if (m->m_pkthdr.len > rtm->rtm_msglen)
632 m_adj(m, rtm->rtm_msglen - m->m_pkthdr.len);
efda3bd0 633 kfree(rtm, M_RTABLE);
984263bc 634 }
ef87f48d 635 if (m != NULL)
a8c218b2 636 rts_input_skip(m, family, rp);
984263bc
MD
637 return (error);
638}
639
ecdefdda
MD
640static void
641route_output_add_callback(int cmd, int error, struct rt_addrinfo *rtinfo,
642 struct rtentry *rt, void *arg)
643{
644 struct rt_msghdr *rtm = arg;
645
646 if (error == 0 && rt != NULL) {
647 rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx,
648 &rt->rt_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);
ac0f13c5
SZ
652 if (rtinfo->rti_genmask != NULL) {
653 rt->rt_genmask = rtmask_purelookup(rtinfo->rti_genmask);
654 if (rt->rt_genmask == NULL) {
655 /*
656 * This should not happen, since we
657 * have already installed genmask
658 * on each CPU before we reach here.
659 */
660 panic("genmask is gone!?");
661 }
662 } else {
663 rt->rt_genmask = NULL;
664 }
31f89d97 665 rtm->rtm_index = rt->rt_ifp->if_index;
ecdefdda
MD
666 }
667}
668
669static void
670route_output_delete_callback(int cmd, int error, struct rt_addrinfo *rtinfo,
671 struct rtentry *rt, void *arg)
672{
ecdefdda
MD
673 if (error == 0 && rt) {
674 ++rt->rt_refcnt;
6e6b3e50 675 if (fillrtmsg(arg, rt, rtinfo) != 0) {
ecdefdda
MD
676 error = ENOBUFS;
677 /* XXX no way to return the error */
678 }
679 --rt->rt_refcnt;
680 }
6e6b3e50
SZ
681 if (rt && rt->rt_refcnt == 0) {
682 ++rt->rt_refcnt;
683 rtfree(rt);
684 }
ecdefdda
MD
685}
686
00104f14
SZ
687static int
688route_output_get_callback(int cmd, struct rt_addrinfo *rtinfo,
b5e01dd3 689 struct rtentry *rt, void *arg, int found_cnt)
00104f14 690{
00104f14
SZ
691 int error, found = 0;
692
693 if (((rtinfo->rti_flags ^ rt->rt_flags) & RTF_HOST) == 0)
694 found = 1;
695
6e6b3e50 696 error = fillrtmsg(arg, rt, rtinfo);
00104f14
SZ
697 if (!error && found) {
698 /* Got the exact match, we could return now! */
699 error = EJUSTRETURN;
700 }
701 return error;
702}
703
2c62ae3b
SZ
704static int
705route_output_change_callback(int cmd, struct rt_addrinfo *rtinfo,
b5e01dd3 706 struct rtentry *rt, void *arg, int found_cnt)
ecdefdda
MD
707{
708 struct rt_msghdr *rtm = arg;
709 struct ifaddr *ifa;
2c62ae3b 710 int error = 0;
ecdefdda
MD
711
712 /*
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
716 */
717 if (((rt->rt_flags & RTF_GATEWAY) && rtinfo->rti_gateway != NULL) ||
fb41af81
SZ
718 rtinfo->rti_ifpaddr != NULL ||
719 (rtinfo->rti_ifaaddr != NULL &&
720 !sa_equal(rtinfo->rti_ifaaddr, rt->rt_ifa->ifa_addr))) {
ecdefdda
MD
721 error = rt_getifa(rtinfo);
722 if (error != 0)
723 goto done;
724 }
725 if (rtinfo->rti_gateway != NULL) {
b5e01dd3
SZ
726 /*
727 * We only need to generate rtmsg upon the
728 * first route to be changed.
729 */
afe44b68 730 error = rt_setgate(rt, rt_key(rt), rtinfo->rti_gateway,
b5e01dd3 731 found_cnt == 1 ? RTL_REPORTMSG : RTL_DONTREPORT);
ecdefdda
MD
732 if (error != 0)
733 goto done;
734 }
735 if ((ifa = rtinfo->rti_ifa) != NULL) {
736 struct ifaddr *oifa = rt->rt_ifa;
737
738 if (oifa != ifa) {
739 if (oifa && oifa->ifa_rtrequest)
740 oifa->ifa_rtrequest(RTM_DELETE, rt, rtinfo);
741 IFAFREE(rt->rt_ifa);
742 IFAREF(ifa);
743 rt->rt_ifa = ifa;
744 rt->rt_ifp = rtinfo->rti_ifp;
745 }
746 }
747 rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx, &rt->rt_rmx);
748 if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest)
2c62ae3b 749 rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, rtinfo);
ac0f13c5
SZ
750 if (rtinfo->rti_genmask != NULL) {
751 rt->rt_genmask = rtmask_purelookup(rtinfo->rti_genmask);
752 if (rt->rt_genmask == NULL) {
753 /*
754 * This should not happen, since we
755 * have already installed genmask
756 * on each CPU before we reach here.
757 */
758 panic("genmask is gone!?\n");
759 }
760 }
31f89d97 761 rtm->rtm_index = rt->rt_ifp->if_index;
ecdefdda 762done:
2c62ae3b 763 return error;
ecdefdda
MD
764}
765
ae2a5ccd
SZ
766static int
767route_output_lock_callback(int cmd, struct rt_addrinfo *rtinfo,
b5e01dd3
SZ
768 struct rtentry *rt, void *arg,
769 int found_cnt __unused)
ecdefdda
MD
770{
771 struct rt_msghdr *rtm = arg;
772
773 rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits);
774 rt->rt_rmx.rmx_locks |=
775 (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks);
ae2a5ccd 776 return 0;
ecdefdda
MD
777}
778
984263bc 779static void
ef87f48d 780rt_setmetrics(u_long which, struct rt_metrics *in, struct rt_metrics *out)
984263bc 781{
2e9572df
JH
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);
8ec89e60 791 setmetric(RTV_MSL, rmx_msl);
bf522a91
SZ
792 setmetric(RTV_IWMAXSEGS, rmx_iwmaxsegs);
793 setmetric(RTV_IWCAPSEGS, rmx_iwcapsegs);
2e9572df 794#undef setmetric
984263bc
MD
795}
796
797#define ROUNDUP(a) \
798 ((a) > 0 ? (1 + (((a) - 1) | (sizeof(long) - 1))) : sizeof(long))
984263bc 799
984263bc
MD
800/*
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.
804 */
805static int
ef87f48d 806rt_xaddrs(char *cp, char *cplim, struct rt_addrinfo *rtinfo)
984263bc 807{
82ed7fc2
RG
808 struct sockaddr *sa;
809 int i;
984263bc
MD
810
811 for (i = 0; (i < RTAX_MAX) && (cp < cplim); i++) {
812 if ((rtinfo->rti_addrs & (1 << i)) == 0)
813 continue;
814 sa = (struct sockaddr *)cp;
815 /*
816 * It won't fit.
817 */
0c3c561c 818 if ((cp + sa->sa_len) > cplim) {
984263bc
MD
819 return (EINVAL);
820 }
821
822 /*
ef87f48d 823 * There are no more... Quit now.
984263bc 824 * If there are more bits, they are in error.
ef87f48d 825 * I've seen this. route(1) can evidently generate these.
984263bc 826 * This causes kernel to core dump.
ef87f48d 827 * For compatibility, if we see this, point to a safe address.
984263bc
MD
828 */
829 if (sa->sa_len == 0) {
590b8cd4
JH
830 static struct sockaddr sa_zero = {
831 sizeof sa_zero, AF_INET,
832 };
833
984263bc 834 rtinfo->rti_info[i] = &sa_zero;
9b42cabe 835 kprintf("rtsock: received more addr bits than sockaddrs.\n");
984263bc
MD
836 return (0); /* should be EINVAL but for compat */
837 }
838
ef87f48d 839 /* Accept the sockaddr. */
984263bc 840 rtinfo->rti_info[i] = sa;
590b8cd4 841 cp += ROUNDUP(sa->sa_len);
984263bc
MD
842 }
843 return (0);
844}
845
590b8cd4
JH
846static int
847rt_msghdrsize(int type)
984263bc 848{
984263bc 849 switch (type) {
984263bc
MD
850 case RTM_DELADDR:
851 case RTM_NEWADDR:
590b8cd4 852 return sizeof(struct ifa_msghdr);
984263bc
MD
853 case RTM_DELMADDR:
854 case RTM_NEWMADDR:
590b8cd4 855 return sizeof(struct ifma_msghdr);
984263bc 856 case RTM_IFINFO:
590b8cd4 857 return sizeof(struct if_msghdr);
984263bc 858 case RTM_IFANNOUNCE:
841ab66c 859 case RTM_IEEE80211:
590b8cd4 860 return sizeof(struct if_announcemsghdr);
984263bc 861 default:
590b8cd4 862 return sizeof(struct rt_msghdr);
984263bc 863 }
590b8cd4
JH
864}
865
866static int
867rt_msgsize(int type, struct rt_addrinfo *rtinfo)
868{
869 int len, i;
870
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);
984263bc 875 }
590b8cd4
JH
876 len = ALIGN(len);
877 return len;
878}
879
880/*
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.
884 *
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
888 * calling routine.
889 */
890static void
891rt_msg_buffer(int type, struct rt_addrinfo *rtinfo, void *buf, int msglen)
892{
893 struct rt_msghdr *rtm;
894 char *cp;
895 int dlen, i;
896
897 rtm = (struct rt_msghdr *) buf;
898 rtm->rtm_version = RTM_VERSION;
899 rtm->rtm_type = type;
900 rtm->rtm_msglen = msglen;
901
902 cp = (char *)buf + rt_msghdrsize(type);
903 rtinfo->rti_addrs = 0;
984263bc 904 for (i = 0; i < RTAX_MAX; i++) {
590b8cd4
JH
905 struct sockaddr *sa;
906
984263bc
MD
907 if ((sa = rtinfo->rti_info[i]) == NULL)
908 continue;
909 rtinfo->rti_addrs |= (1 << i);
910 dlen = ROUNDUP(sa->sa_len);
590b8cd4
JH
911 bcopy(sa, cp, dlen);
912 cp += dlen;
984263bc 913 }
984263bc
MD
914}
915
590b8cd4
JH
916/*
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.
920 *
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
924 * calling routine.
925 */
926static struct mbuf *
927rt_msg_mbuf(int type, struct rt_addrinfo *rtinfo)
984263bc 928{
590b8cd4
JH
929 struct mbuf *m;
930 struct rt_msghdr *rtm;
931 int hlen, len;
82ed7fc2 932 int i;
984263bc 933
590b8cd4
JH
934 hlen = rt_msghdrsize(type);
935 KASSERT(hlen <= MCLBYTES, ("rt_msg_mbuf: hlen %d doesn't fit", hlen));
984263bc 936
50503f0f 937 m = m_getl(hlen, MB_DONTWAIT, MT_DATA, M_PKTHDR, NULL);
590b8cd4
JH
938 if (m == NULL)
939 return (NULL);
e9fa4b60 940 mbuftrackid(m, 32);
590b8cd4
JH
941 m->m_pkthdr.len = m->m_len = hlen;
942 m->m_pkthdr.rcvif = NULL;
943 rtinfo->rti_addrs = 0;
944 len = hlen;
984263bc 945 for (i = 0; i < RTAX_MAX; i++) {
82ed7fc2 946 struct sockaddr *sa;
590b8cd4 947 int dlen;
984263bc 948
ef87f48d 949 if ((sa = rtinfo->rti_info[i]) == NULL)
984263bc
MD
950 continue;
951 rtinfo->rti_addrs |= (1 << i);
952 dlen = ROUNDUP(sa->sa_len);
590b8cd4 953 m_copyback(m, len, dlen, (caddr_t)sa); /* can grow mbuf chain */
984263bc
MD
954 len += dlen;
955 }
590b8cd4
JH
956 if (m->m_pkthdr.len != len) { /* one of the m_copyback() calls failed */
957 m_freem(m);
958 return (NULL);
984263bc 959 }
590b8cd4
JH
960 rtm = mtod(m, struct rt_msghdr *);
961 bzero(rtm, hlen);
962 rtm->rtm_msglen = len;
963 rtm->rtm_version = RTM_VERSION;
964 rtm->rtm_type = type;
965 return (m);
984263bc
MD
966}
967
968/*
969 * This routine is called to generate a message from the routing
f23061d4 970 * socket indicating that a redirect has occurred, a routing lookup
984263bc
MD
971 * has failed, or that a protocol has detected timeouts to a particular
972 * destination.
973 */
974void
f23061d4 975rt_missmsg(int type, struct rt_addrinfo *rtinfo, int flags, int error)
984263bc 976{
0c3c561c 977 struct sockaddr *dst = rtinfo->rti_info[RTAX_DST];
82ed7fc2
RG
978 struct rt_msghdr *rtm;
979 struct mbuf *m;
984263bc
MD
980
981 if (route_cb.any_count == 0)
982 return;
590b8cd4 983 m = rt_msg_mbuf(type, rtinfo);
ef87f48d 984 if (m == NULL)
984263bc
MD
985 return;
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;
590b8cd4 990 rts_input(m, familyof(dst));
984263bc
MD
991}
992
f3ed2586
JH
993void
994rt_dstmsg(int type, struct sockaddr *dst, int error)
995{
996 struct rt_msghdr *rtm;
997 struct rt_addrinfo addrs;
998 struct mbuf *m;
999
1000 if (route_cb.any_count == 0)
1001 return;
1002 bzero(&addrs, sizeof(struct rt_addrinfo));
1003 addrs.rti_info[RTAX_DST] = dst;
590b8cd4 1004 m = rt_msg_mbuf(type, &addrs);
f3ed2586
JH
1005 if (m == NULL)
1006 return;
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;
590b8cd4 1011 rts_input(m, familyof(dst));
f3ed2586
JH
1012}
1013
984263bc
MD
1014/*
1015 * This routine is called to generate a message from the routing
1016 * socket indicating that the status of a network interface has changed.
1017 */
1018void
0c3c561c 1019rt_ifmsg(struct ifnet *ifp)
984263bc 1020{
82ed7fc2 1021 struct if_msghdr *ifm;
984263bc 1022 struct mbuf *m;
590b8cd4 1023 struct rt_addrinfo rtinfo;
984263bc
MD
1024
1025 if (route_cb.any_count == 0)
1026 return;
590b8cd4
JH
1027 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1028 m = rt_msg_mbuf(RTM_IFINFO, &rtinfo);
ef87f48d 1029 if (m == NULL)
984263bc
MD
1030 return;
1031 ifm = mtod(m, struct if_msghdr *);
1032 ifm->ifm_index = ifp->if_index;
9c095379 1033 ifm->ifm_flags = ifp->if_flags;
984263bc 1034 ifm->ifm_data = ifp->if_data;
590b8cd4
JH
1035 ifm->ifm_addrs = 0;
1036 rts_input(m, 0);
984263bc
MD
1037}
1038
372316d9
JH
1039static void
1040rt_ifamsg(int cmd, struct ifaddr *ifa)
1041{
1042 struct ifa_msghdr *ifam;
590b8cd4 1043 struct rt_addrinfo rtinfo;
372316d9 1044 struct mbuf *m;
372316d9
JH
1045 struct ifnet *ifp = ifa->ifa_ifp;
1046
590b8cd4
JH
1047 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1048 rtinfo.rti_ifaaddr = ifa->ifa_addr;
b2632176
SZ
1049 rtinfo.rti_ifpaddr =
1050 TAILQ_FIRST(&ifp->if_addrheads[mycpuid])->ifa->ifa_addr;
590b8cd4
JH
1051 rtinfo.rti_netmask = ifa->ifa_netmask;
1052 rtinfo.rti_bcastaddr = ifa->ifa_dstaddr;
f23061d4 1053
590b8cd4 1054 m = rt_msg_mbuf(cmd, &rtinfo);
f23061d4 1055 if (m == NULL)
372316d9 1056 return;
f23061d4 1057
372316d9
JH
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;
590b8cd4 1062 ifam->ifam_addrs = rtinfo.rti_addrs;
f23061d4 1063
590b8cd4 1064 rts_input(m, familyof(ifa->ifa_addr));
372316d9
JH
1065}
1066
f3ed2586
JH
1067void
1068rt_rtmsg(int cmd, struct rtentry *rt, struct ifnet *ifp, int error)
372316d9
JH
1069{
1070 struct rt_msghdr *rtm;
590b8cd4 1071 struct rt_addrinfo rtinfo;
372316d9 1072 struct mbuf *m;
f3ed2586 1073 struct sockaddr *dst;
372316d9
JH
1074
1075 if (rt == NULL)
1076 return;
f23061d4 1077
590b8cd4
JH
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);
b2632176
SZ
1082 if (ifp != NULL) {
1083 rtinfo.rti_ifpaddr =
1084 TAILQ_FIRST(&ifp->if_addrheads[mycpuid])->ifa->ifa_addr;
1085 }
590b8cd4 1086 rtinfo.rti_ifaaddr = rt->rt_ifa->ifa_addr;
f23061d4 1087
590b8cd4 1088 m = rt_msg_mbuf(cmd, &rtinfo);
f23061d4 1089 if (m == NULL)
372316d9 1090 return;
f23061d4 1091
372316d9 1092 rtm = mtod(m, struct rt_msghdr *);
f3ed2586
JH
1093 if (ifp != NULL)
1094 rtm->rtm_index = ifp->if_index;
372316d9
JH
1095 rtm->rtm_flags |= rt->rt_flags;
1096 rtm->rtm_errno = error;
590b8cd4 1097 rtm->rtm_addrs = rtinfo.rti_addrs;
f23061d4 1098
590b8cd4 1099 rts_input(m, familyof(dst));
372316d9
JH
1100}
1101
984263bc
MD
1102/*
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
1108 * copies of it.
1109 */
1110void
0c3c561c 1111rt_newaddrmsg(int cmd, struct ifaddr *ifa, int error, struct rtentry *rt)
984263bc 1112{
78812139
EN
1113#ifdef SCTP
1114 /*
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...
1118 */
1119 if (cmd == RTM_ADD)
1120 sctp_add_ip_address(ifa);
1121 else if (cmd == RTM_DELETE)
1122 sctp_delete_ip_address(ifa);
1123#endif /* SCTP */
1124
984263bc
MD
1125 if (route_cb.any_count == 0)
1126 return;
984263bc 1127
372316d9
JH
1128 if (cmd == RTM_ADD) {
1129 rt_ifamsg(RTM_NEWADDR, ifa);
f3ed2586 1130 rt_rtmsg(RTM_ADD, rt, ifa->ifa_ifp, error);
372316d9
JH
1131 } else {
1132 KASSERT((cmd == RTM_DELETE), ("unknown cmd %d", cmd));
f3ed2586 1133 rt_rtmsg(RTM_DELETE, rt, ifa->ifa_ifp, error);
372316d9 1134 rt_ifamsg(RTM_DELADDR, ifa);
984263bc
MD
1135 }
1136}
1137
1138/*
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.
1142 */
1143void
0c3c561c 1144rt_newmaddrmsg(int cmd, struct ifmultiaddr *ifma)
984263bc 1145{
590b8cd4 1146 struct rt_addrinfo rtinfo;
ef87f48d 1147 struct mbuf *m = NULL;
984263bc
MD
1148 struct ifnet *ifp = ifma->ifma_ifp;
1149 struct ifma_msghdr *ifmam;
1150
1151 if (route_cb.any_count == 0)
1152 return;
1153
590b8cd4
JH
1154 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1155 rtinfo.rti_ifaaddr = ifma->ifma_addr;
b2632176
SZ
1156 if (ifp != NULL && !TAILQ_EMPTY(&ifp->if_addrheads[mycpuid])) {
1157 rtinfo.rti_ifpaddr =
1158 TAILQ_FIRST(&ifp->if_addrheads[mycpuid])->ifa->ifa_addr;
1159 }
984263bc
MD
1160 /*
1161 * If a link-layer address is present, present it as a ``gateway''
1162 * (similarly to how ARP entries, e.g., are presented).
1163 */
590b8cd4 1164 rtinfo.rti_gateway = ifma->ifma_lladdr;
f23061d4 1165
590b8cd4 1166 m = rt_msg_mbuf(cmd, &rtinfo);
f23061d4 1167 if (m == NULL)
984263bc 1168 return;
f23061d4 1169
984263bc
MD
1170 ifmam = mtod(m, struct ifma_msghdr *);
1171 ifmam->ifmam_index = ifp->if_index;
590b8cd4 1172 ifmam->ifmam_addrs = rtinfo.rti_addrs;
f23061d4 1173
590b8cd4 1174 rts_input(m, familyof(ifma->ifma_addr));
984263bc
MD
1175}
1176
841ab66c
SZ
1177static struct mbuf *
1178rt_makeifannouncemsg(struct ifnet *ifp, int type, int what,
1179 struct rt_addrinfo *info)
984263bc 1180{
590b8cd4 1181 struct if_announcemsghdr *ifan;
841ab66c 1182 struct mbuf *m;
984263bc
MD
1183
1184 if (route_cb.any_count == 0)
841ab66c 1185 return NULL;
f23061d4 1186
841ab66c
SZ
1187 bzero(info, sizeof(*info));
1188 m = rt_msg_mbuf(type, info);
984263bc 1189 if (m == NULL)
841ab66c 1190 return NULL;
f23061d4 1191
984263bc
MD
1192 ifan = mtod(m, struct if_announcemsghdr *);
1193 ifan->ifan_index = ifp->if_index;
5fe66e68 1194 strlcpy(ifan->ifan_name, ifp->if_xname, sizeof ifan->ifan_name);
984263bc 1195 ifan->ifan_what = what;
841ab66c
SZ
1196 return m;
1197}
1198
1199/*
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.
1203 */
1204void
1205rt_ieee80211msg(struct ifnet *ifp, int what, void *data, size_t data_len)
1206{
1207 struct rt_addrinfo info;
1208 struct mbuf *m;
f23061d4 1209
841ab66c
SZ
1210 m = rt_makeifannouncemsg(ifp, RTM_IEEE80211, what, &info);
1211 if (m == NULL)
1212 return;
1213
1214 /*
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.
1218 *
1219 * NB: we assume m is a single mbuf.
1220 */
1221 if (data_len > M_TRAILINGSPACE(m)) {
e28d8186 1222 /* XXX use m_getb(data_len, MB_DONTWAIT, MT_DATA, 0); */
b11e1686 1223 struct mbuf *n = m_get(MB_DONTWAIT, MT_DATA);
841ab66c
SZ
1224 if (n == NULL) {
1225 m_freem(m);
1226 return;
1227 }
e28d8186 1228 KKASSERT(data_len <= M_TRAILINGSPACE(n));
841ab66c
SZ
1229 bcopy(data, mtod(n, void *), data_len);
1230 n->m_len = data_len;
1231 m->m_next = n;
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;
1235 }
e9fa4b60 1236 mbuftrackid(m, 33);
841ab66c
SZ
1237 if (m->m_flags & M_PKTHDR)
1238 m->m_pkthdr.len += data_len;
1239 mtod(m, struct if_announcemsghdr *)->ifan_msglen += data_len;
590b8cd4 1240 rts_input(m, 0);
841ab66c
SZ
1241}
1242
1243/*
1244 * This is called to generate routing socket messages indicating
1245 * network interface arrival and departure.
1246 */
1247void
1248rt_ifannouncemsg(struct ifnet *ifp, int what)
1249{
1250 struct rt_addrinfo addrinfo;
1251 struct mbuf *m;
1252
1253 m = rt_makeifannouncemsg(ifp, RTM_IFANNOUNCE, what, &addrinfo);
1254 if (m != NULL)
1255 rts_input(m, 0);
590b8cd4 1256}
f23061d4 1257
590b8cd4
JH
1258static int
1259resizewalkarg(struct walkarg *w, int len)
1260{
1261 void *newptr;
1262
efda3bd0 1263 newptr = kmalloc(len, M_RTABLE, M_INTWAIT | M_NULLOK);
590b8cd4
JH
1264 if (newptr == NULL)
1265 return (ENOMEM);
1266 if (w->w_tmem != NULL)
efda3bd0 1267 kfree(w->w_tmem, M_RTABLE);
590b8cd4
JH
1268 w->w_tmem = newptr;
1269 w->w_tmemsize = len;
1270 return (0);
1271}
984263bc
MD
1272
1273/*
1274 * This is used in dumping the kernel table via sysctl().
1275 */
1276int
590b8cd4 1277sysctl_dumpentry(struct radix_node *rn, void *vw)
984263bc 1278{
82ed7fc2
RG
1279 struct walkarg *w = vw;
1280 struct rtentry *rt = (struct rtentry *)rn;
590b8cd4
JH
1281 struct rt_addrinfo rtinfo;
1282 int error, msglen;
984263bc
MD
1283
1284 if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg))
1285 return 0;
f23061d4 1286
590b8cd4
JH
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;
ef87f48d 1292 if (rt->rt_ifp != NULL) {
590b8cd4 1293 rtinfo.rti_ifpaddr =
b2632176 1294 TAILQ_FIRST(&rt->rt_ifp->if_addrheads[mycpuid])->ifa->ifa_addr;
590b8cd4 1295 rtinfo.rti_ifaaddr = rt->rt_ifa->ifa_addr;
984263bc 1296 if (rt->rt_ifp->if_flags & IFF_POINTOPOINT)
590b8cd4 1297 rtinfo.rti_bcastaddr = rt->rt_ifa->ifa_dstaddr;
984263bc 1298 }
590b8cd4
JH
1299 msglen = rt_msgsize(RTM_GET, &rtinfo);
1300 if (w->w_tmemsize < msglen && resizewalkarg(w, msglen) != 0)
1301 return (ENOMEM);
1302 rt_msg_buffer(RTM_GET, &rtinfo, w->w_tmem, msglen);
1303 if (w->w_req != NULL) {
8e63efec 1304 struct rt_msghdr *rtm = w->w_tmem;
984263bc
MD
1305
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;
590b8cd4
JH
1311 rtm->rtm_addrs = rtinfo.rti_addrs;
1312 error = SYSCTL_OUT(w->w_req, rtm, msglen);
984263bc
MD
1313 return (error);
1314 }
590b8cd4 1315 return (0);
984263bc
MD
1316}
1317
590b8cd4
JH
1318static int
1319sysctl_iflist(int af, struct walkarg *w)
984263bc 1320{
82ed7fc2 1321 struct ifnet *ifp;
590b8cd4
JH
1322 struct rt_addrinfo rtinfo;
1323 int msglen, error;
984263bc 1324
590b8cd4 1325 bzero(&rtinfo, sizeof(struct rt_addrinfo));
984263bc 1326 TAILQ_FOREACH(ifp, &ifnet, if_link) {
b2632176
SZ
1327 struct ifaddr_container *ifac;
1328 struct ifaddr *ifa;
1329
984263bc
MD
1330 if (w->w_arg && w->w_arg != ifp->if_index)
1331 continue;
b2632176
SZ
1332 ifac = TAILQ_FIRST(&ifp->if_addrheads[mycpuid]);
1333 ifa = ifac->ifa;
590b8cd4
JH
1334 rtinfo.rti_ifpaddr = ifa->ifa_addr;
1335 msglen = rt_msgsize(RTM_IFINFO, &rtinfo);
1336 if (w->w_tmemsize < msglen && resizewalkarg(w, msglen) != 0)
1337 return (ENOMEM);
1338 rt_msg_buffer(RTM_IFINFO, &rtinfo, w->w_tmem, msglen);
1339 rtinfo.rti_ifpaddr = NULL;
ef87f48d 1340 if (w->w_req != NULL && w->w_tmem != NULL) {
8e63efec 1341 struct if_msghdr *ifm = w->w_tmem;
984263bc 1342
984263bc 1343 ifm->ifm_index = ifp->if_index;
9c095379 1344 ifm->ifm_flags = ifp->if_flags;
984263bc 1345 ifm->ifm_data = ifp->if_data;
590b8cd4
JH
1346 ifm->ifm_addrs = rtinfo.rti_addrs;
1347 error = SYSCTL_OUT(w->w_req, ifm, msglen);
984263bc
MD
1348 if (error)
1349 return (error);
1350 }
b2632176
SZ
1351 while ((ifac = TAILQ_NEXT(ifac, ifa_link)) != NULL) {
1352 ifa = ifac->ifa;
1353
984263bc
MD
1354 if (af && af != ifa->ifa_addr->sa_family)
1355 continue;
590b8cd4 1356 if (curproc->p_ucred->cr_prison &&
87de5057 1357 prison_if(curproc->p_ucred, ifa->ifa_addr))
984263bc 1358 continue;
590b8cd4
JH
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)
1365 return (ENOMEM);
1366 rt_msg_buffer(RTM_NEWADDR, &rtinfo, w->w_tmem, msglen);
1367 if (w->w_req != NULL) {
8e63efec 1368 struct ifa_msghdr *ifam = w->w_tmem;
984263bc 1369
984263bc
MD
1370 ifam->ifam_index = ifa->ifa_ifp->if_index;
1371 ifam->ifam_flags = ifa->ifa_flags;
1372 ifam->ifam_metric = ifa->ifa_metric;
590b8cd4
JH
1373 ifam->ifam_addrs = rtinfo.rti_addrs;
1374 error = SYSCTL_OUT(w->w_req, w->w_tmem, msglen);
984263bc
MD
1375 if (error)
1376 return (error);
1377 }
1378 }
590b8cd4
JH
1379 rtinfo.rti_netmask = NULL;
1380 rtinfo.rti_ifaaddr = NULL;
1381 rtinfo.rti_bcastaddr = NULL;
984263bc
MD
1382 }
1383 return (0);
1384}
1385
1386static int
1387sysctl_rtsock(SYSCTL_HANDLER_ARGS)
1388{
1389 int *name = (int *)arg1;
1390 u_int namelen = arg2;
82ed7fc2 1391 struct radix_node_head *rnh;
a49aa710 1392 int i, error = EINVAL;
ecdefdda 1393 int origcpu;
984263bc
MD
1394 u_char af;
1395 struct walkarg w;
1396
1397 name ++;
1398 namelen--;
1399 if (req->newptr)
1400 return (EPERM);
ecdefdda 1401 if (namelen != 3 && namelen != 4)
984263bc
MD
1402 return (EINVAL);
1403 af = name[0];
5fe66e68 1404 bzero(&w, sizeof w);
984263bc
MD
1405 w.w_op = name[1];
1406 w.w_arg = name[2];
1407 w.w_req = req;
1408
ecdefdda
MD
1409 /*
1410 * Optional third argument specifies cpu, used primarily for
1411 * debugging the route table.
1412 */
1413 if (namelen == 4) {
1414 if (name[3] < 0 || name[3] >= ncpus)
1415 return (EINVAL);
1416 origcpu = mycpuid;
1417 lwkt_migratecpu(name[3]);
1418 } else {
1419 origcpu = -1;
1420 }
a49aa710 1421 crit_enter();
984263bc 1422 switch (w.w_op) {
984263bc
MD
1423 case NET_RT_DUMP:
1424 case NET_RT_FLAGS:
1425 for (i = 1; i <= AF_MAX; i++)
ecdefdda
MD
1426 if ((rnh = rt_tables[mycpuid][i]) &&
1427 (af == 0 || af == i) &&
984263bc 1428 (error = rnh->rnh_walktree(rnh,
f23061d4 1429 sysctl_dumpentry, &w)))
984263bc
MD
1430 break;
1431 break;
1432
1433 case NET_RT_IFLIST:
1434 error = sysctl_iflist(af, &w);
1435 }
a49aa710 1436 crit_exit();
590b8cd4 1437 if (w.w_tmem != NULL)
efda3bd0 1438 kfree(w.w_tmem, M_RTABLE);
ecdefdda
MD
1439 if (origcpu >= 0)
1440 lwkt_migratecpu(origcpu);
984263bc
MD
1441 return (error);
1442}
1443
1444SYSCTL_NODE(_net, PF_ROUTE, routetable, CTLFLAG_RD, sysctl_rtsock, "");
1445
1446/*
1447 * Definitions of protocols supported in the ROUTE domain.
1448 */
1449
ba4257b8 1450static struct domain routedomain; /* or at least forward */
984263bc
MD
1451
1452static struct protosw routesw[] = {
002c1265
MD
1453 {
1454 .pr_type = SOCK_RAW,
1455 .pr_domain = &routedomain,
1456 .pr_protocol = 0,
1457 .pr_flags = PR_ATOMIC|PR_ADDR,
1458 .pr_input = NULL,
1459 .pr_output = route_output,
1460 .pr_ctlinput = raw_ctlinput,
1461 .pr_ctloutput = NULL,
1462 .pr_ctlport = cpu0_ctlport,
1463
1464 .pr_init = raw_init,
1465 .pr_usrreqs = &route_usrreqs
1466 }
984263bc
MD
1467};
1468
ba4257b8 1469static struct domain routedomain = {
9c70fe43
JH
1470 PF_ROUTE, "route", NULL, NULL, NULL,
1471 routesw, &routesw[(sizeof routesw)/(sizeof routesw[0])],
1472};
984263bc
MD
1473
1474DOMAIN_SET(route);
47db7f9b 1475