nataraid(4): Add devstat support.
[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);
2e9572df 792#undef setmetric
984263bc
MD
793}
794
795#define ROUNDUP(a) \
796 ((a) > 0 ? (1 + (((a) - 1) | (sizeof(long) - 1))) : sizeof(long))
984263bc 797
984263bc
MD
798/*
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.
802 */
803static int
ef87f48d 804rt_xaddrs(char *cp, char *cplim, struct rt_addrinfo *rtinfo)
984263bc 805{
82ed7fc2
RG
806 struct sockaddr *sa;
807 int i;
984263bc
MD
808
809 for (i = 0; (i < RTAX_MAX) && (cp < cplim); i++) {
810 if ((rtinfo->rti_addrs & (1 << i)) == 0)
811 continue;
812 sa = (struct sockaddr *)cp;
813 /*
814 * It won't fit.
815 */
0c3c561c 816 if ((cp + sa->sa_len) > cplim) {
984263bc
MD
817 return (EINVAL);
818 }
819
820 /*
ef87f48d 821 * There are no more... Quit now.
984263bc 822 * If there are more bits, they are in error.
ef87f48d 823 * I've seen this. route(1) can evidently generate these.
984263bc 824 * This causes kernel to core dump.
ef87f48d 825 * For compatibility, if we see this, point to a safe address.
984263bc
MD
826 */
827 if (sa->sa_len == 0) {
590b8cd4
JH
828 static struct sockaddr sa_zero = {
829 sizeof sa_zero, AF_INET,
830 };
831
984263bc 832 rtinfo->rti_info[i] = &sa_zero;
9b42cabe 833 kprintf("rtsock: received more addr bits than sockaddrs.\n");
984263bc
MD
834 return (0); /* should be EINVAL but for compat */
835 }
836
ef87f48d 837 /* Accept the sockaddr. */
984263bc 838 rtinfo->rti_info[i] = sa;
590b8cd4 839 cp += ROUNDUP(sa->sa_len);
984263bc
MD
840 }
841 return (0);
842}
843
590b8cd4
JH
844static int
845rt_msghdrsize(int type)
984263bc 846{
984263bc 847 switch (type) {
984263bc
MD
848 case RTM_DELADDR:
849 case RTM_NEWADDR:
590b8cd4 850 return sizeof(struct ifa_msghdr);
984263bc
MD
851 case RTM_DELMADDR:
852 case RTM_NEWMADDR:
590b8cd4 853 return sizeof(struct ifma_msghdr);
984263bc 854 case RTM_IFINFO:
590b8cd4 855 return sizeof(struct if_msghdr);
984263bc 856 case RTM_IFANNOUNCE:
841ab66c 857 case RTM_IEEE80211:
590b8cd4 858 return sizeof(struct if_announcemsghdr);
984263bc 859 default:
590b8cd4 860 return sizeof(struct rt_msghdr);
984263bc 861 }
590b8cd4
JH
862}
863
864static int
865rt_msgsize(int type, struct rt_addrinfo *rtinfo)
866{
867 int len, i;
868
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);
984263bc 873 }
590b8cd4
JH
874 len = ALIGN(len);
875 return len;
876}
877
878/*
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.
882 *
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
886 * calling routine.
887 */
888static void
889rt_msg_buffer(int type, struct rt_addrinfo *rtinfo, void *buf, int msglen)
890{
891 struct rt_msghdr *rtm;
892 char *cp;
893 int dlen, i;
894
895 rtm = (struct rt_msghdr *) buf;
896 rtm->rtm_version = RTM_VERSION;
897 rtm->rtm_type = type;
898 rtm->rtm_msglen = msglen;
899
900 cp = (char *)buf + rt_msghdrsize(type);
901 rtinfo->rti_addrs = 0;
984263bc 902 for (i = 0; i < RTAX_MAX; i++) {
590b8cd4
JH
903 struct sockaddr *sa;
904
984263bc
MD
905 if ((sa = rtinfo->rti_info[i]) == NULL)
906 continue;
907 rtinfo->rti_addrs |= (1 << i);
908 dlen = ROUNDUP(sa->sa_len);
590b8cd4
JH
909 bcopy(sa, cp, dlen);
910 cp += dlen;
984263bc 911 }
984263bc
MD
912}
913
590b8cd4
JH
914/*
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.
918 *
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
922 * calling routine.
923 */
924static struct mbuf *
925rt_msg_mbuf(int type, struct rt_addrinfo *rtinfo)
984263bc 926{
590b8cd4
JH
927 struct mbuf *m;
928 struct rt_msghdr *rtm;
929 int hlen, len;
82ed7fc2 930 int i;
984263bc 931
590b8cd4
JH
932 hlen = rt_msghdrsize(type);
933 KASSERT(hlen <= MCLBYTES, ("rt_msg_mbuf: hlen %d doesn't fit", hlen));
984263bc 934
50503f0f 935 m = m_getl(hlen, MB_DONTWAIT, MT_DATA, M_PKTHDR, NULL);
590b8cd4
JH
936 if (m == NULL)
937 return (NULL);
e9fa4b60 938 mbuftrackid(m, 32);
590b8cd4
JH
939 m->m_pkthdr.len = m->m_len = hlen;
940 m->m_pkthdr.rcvif = NULL;
941 rtinfo->rti_addrs = 0;
942 len = hlen;
984263bc 943 for (i = 0; i < RTAX_MAX; i++) {
82ed7fc2 944 struct sockaddr *sa;
590b8cd4 945 int dlen;
984263bc 946
ef87f48d 947 if ((sa = rtinfo->rti_info[i]) == NULL)
984263bc
MD
948 continue;
949 rtinfo->rti_addrs |= (1 << i);
950 dlen = ROUNDUP(sa->sa_len);
590b8cd4 951 m_copyback(m, len, dlen, (caddr_t)sa); /* can grow mbuf chain */
984263bc
MD
952 len += dlen;
953 }
590b8cd4
JH
954 if (m->m_pkthdr.len != len) { /* one of the m_copyback() calls failed */
955 m_freem(m);
956 return (NULL);
984263bc 957 }
590b8cd4
JH
958 rtm = mtod(m, struct rt_msghdr *);
959 bzero(rtm, hlen);
960 rtm->rtm_msglen = len;
961 rtm->rtm_version = RTM_VERSION;
962 rtm->rtm_type = type;
963 return (m);
984263bc
MD
964}
965
966/*
967 * This routine is called to generate a message from the routing
f23061d4 968 * socket indicating that a redirect has occurred, a routing lookup
984263bc
MD
969 * has failed, or that a protocol has detected timeouts to a particular
970 * destination.
971 */
972void
f23061d4 973rt_missmsg(int type, struct rt_addrinfo *rtinfo, int flags, int error)
984263bc 974{
0c3c561c 975 struct sockaddr *dst = rtinfo->rti_info[RTAX_DST];
82ed7fc2
RG
976 struct rt_msghdr *rtm;
977 struct mbuf *m;
984263bc
MD
978
979 if (route_cb.any_count == 0)
980 return;
590b8cd4 981 m = rt_msg_mbuf(type, rtinfo);
ef87f48d 982 if (m == NULL)
984263bc
MD
983 return;
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;
590b8cd4 988 rts_input(m, familyof(dst));
984263bc
MD
989}
990
f3ed2586
JH
991void
992rt_dstmsg(int type, struct sockaddr *dst, int error)
993{
994 struct rt_msghdr *rtm;
995 struct rt_addrinfo addrs;
996 struct mbuf *m;
997
998 if (route_cb.any_count == 0)
999 return;
1000 bzero(&addrs, sizeof(struct rt_addrinfo));
1001 addrs.rti_info[RTAX_DST] = dst;
590b8cd4 1002 m = rt_msg_mbuf(type, &addrs);
f3ed2586
JH
1003 if (m == NULL)
1004 return;
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;
590b8cd4 1009 rts_input(m, familyof(dst));
f3ed2586
JH
1010}
1011
984263bc
MD
1012/*
1013 * This routine is called to generate a message from the routing
1014 * socket indicating that the status of a network interface has changed.
1015 */
1016void
0c3c561c 1017rt_ifmsg(struct ifnet *ifp)
984263bc 1018{
82ed7fc2 1019 struct if_msghdr *ifm;
984263bc 1020 struct mbuf *m;
590b8cd4 1021 struct rt_addrinfo rtinfo;
984263bc
MD
1022
1023 if (route_cb.any_count == 0)
1024 return;
590b8cd4
JH
1025 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1026 m = rt_msg_mbuf(RTM_IFINFO, &rtinfo);
ef87f48d 1027 if (m == NULL)
984263bc
MD
1028 return;
1029 ifm = mtod(m, struct if_msghdr *);
1030 ifm->ifm_index = ifp->if_index;
9c095379 1031 ifm->ifm_flags = ifp->if_flags;
984263bc 1032 ifm->ifm_data = ifp->if_data;
590b8cd4
JH
1033 ifm->ifm_addrs = 0;
1034 rts_input(m, 0);
984263bc
MD
1035}
1036
372316d9
JH
1037static void
1038rt_ifamsg(int cmd, struct ifaddr *ifa)
1039{
1040 struct ifa_msghdr *ifam;
590b8cd4 1041 struct rt_addrinfo rtinfo;
372316d9 1042 struct mbuf *m;
372316d9
JH
1043 struct ifnet *ifp = ifa->ifa_ifp;
1044
590b8cd4
JH
1045 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1046 rtinfo.rti_ifaaddr = ifa->ifa_addr;
b2632176
SZ
1047 rtinfo.rti_ifpaddr =
1048 TAILQ_FIRST(&ifp->if_addrheads[mycpuid])->ifa->ifa_addr;
590b8cd4
JH
1049 rtinfo.rti_netmask = ifa->ifa_netmask;
1050 rtinfo.rti_bcastaddr = ifa->ifa_dstaddr;
f23061d4 1051
590b8cd4 1052 m = rt_msg_mbuf(cmd, &rtinfo);
f23061d4 1053 if (m == NULL)
372316d9 1054 return;
f23061d4 1055
372316d9
JH
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;
590b8cd4 1060 ifam->ifam_addrs = rtinfo.rti_addrs;
f23061d4 1061
590b8cd4 1062 rts_input(m, familyof(ifa->ifa_addr));
372316d9
JH
1063}
1064
f3ed2586
JH
1065void
1066rt_rtmsg(int cmd, struct rtentry *rt, struct ifnet *ifp, int error)
372316d9
JH
1067{
1068 struct rt_msghdr *rtm;
590b8cd4 1069 struct rt_addrinfo rtinfo;
372316d9 1070 struct mbuf *m;
f3ed2586 1071 struct sockaddr *dst;
372316d9
JH
1072
1073 if (rt == NULL)
1074 return;
f23061d4 1075
590b8cd4
JH
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);
b2632176
SZ
1080 if (ifp != NULL) {
1081 rtinfo.rti_ifpaddr =
1082 TAILQ_FIRST(&ifp->if_addrheads[mycpuid])->ifa->ifa_addr;
1083 }
590b8cd4 1084 rtinfo.rti_ifaaddr = rt->rt_ifa->ifa_addr;
f23061d4 1085
590b8cd4 1086 m = rt_msg_mbuf(cmd, &rtinfo);
f23061d4 1087 if (m == NULL)
372316d9 1088 return;
f23061d4 1089
372316d9 1090 rtm = mtod(m, struct rt_msghdr *);
f3ed2586
JH
1091 if (ifp != NULL)
1092 rtm->rtm_index = ifp->if_index;
372316d9
JH
1093 rtm->rtm_flags |= rt->rt_flags;
1094 rtm->rtm_errno = error;
590b8cd4 1095 rtm->rtm_addrs = rtinfo.rti_addrs;
f23061d4 1096
590b8cd4 1097 rts_input(m, familyof(dst));
372316d9
JH
1098}
1099
984263bc
MD
1100/*
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
1106 * copies of it.
1107 */
1108void
0c3c561c 1109rt_newaddrmsg(int cmd, struct ifaddr *ifa, int error, struct rtentry *rt)
984263bc 1110{
78812139
EN
1111#ifdef SCTP
1112 /*
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...
1116 */
1117 if (cmd == RTM_ADD)
1118 sctp_add_ip_address(ifa);
1119 else if (cmd == RTM_DELETE)
1120 sctp_delete_ip_address(ifa);
1121#endif /* SCTP */
1122
984263bc
MD
1123 if (route_cb.any_count == 0)
1124 return;
984263bc 1125
372316d9
JH
1126 if (cmd == RTM_ADD) {
1127 rt_ifamsg(RTM_NEWADDR, ifa);
f3ed2586 1128 rt_rtmsg(RTM_ADD, rt, ifa->ifa_ifp, error);
372316d9
JH
1129 } else {
1130 KASSERT((cmd == RTM_DELETE), ("unknown cmd %d", cmd));
f3ed2586 1131 rt_rtmsg(RTM_DELETE, rt, ifa->ifa_ifp, error);
372316d9 1132 rt_ifamsg(RTM_DELADDR, ifa);
984263bc
MD
1133 }
1134}
1135
1136/*
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.
1140 */
1141void
0c3c561c 1142rt_newmaddrmsg(int cmd, struct ifmultiaddr *ifma)
984263bc 1143{
590b8cd4 1144 struct rt_addrinfo rtinfo;
ef87f48d 1145 struct mbuf *m = NULL;
984263bc
MD
1146 struct ifnet *ifp = ifma->ifma_ifp;
1147 struct ifma_msghdr *ifmam;
1148
1149 if (route_cb.any_count == 0)
1150 return;
1151
590b8cd4
JH
1152 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1153 rtinfo.rti_ifaaddr = ifma->ifma_addr;
b2632176
SZ
1154 if (ifp != NULL && !TAILQ_EMPTY(&ifp->if_addrheads[mycpuid])) {
1155 rtinfo.rti_ifpaddr =
1156 TAILQ_FIRST(&ifp->if_addrheads[mycpuid])->ifa->ifa_addr;
1157 }
984263bc
MD
1158 /*
1159 * If a link-layer address is present, present it as a ``gateway''
1160 * (similarly to how ARP entries, e.g., are presented).
1161 */
590b8cd4 1162 rtinfo.rti_gateway = ifma->ifma_lladdr;
f23061d4 1163
590b8cd4 1164 m = rt_msg_mbuf(cmd, &rtinfo);
f23061d4 1165 if (m == NULL)
984263bc 1166 return;
f23061d4 1167
984263bc
MD
1168 ifmam = mtod(m, struct ifma_msghdr *);
1169 ifmam->ifmam_index = ifp->if_index;
590b8cd4 1170 ifmam->ifmam_addrs = rtinfo.rti_addrs;
f23061d4 1171
590b8cd4 1172 rts_input(m, familyof(ifma->ifma_addr));
984263bc
MD
1173}
1174
841ab66c
SZ
1175static struct mbuf *
1176rt_makeifannouncemsg(struct ifnet *ifp, int type, int what,
1177 struct rt_addrinfo *info)
984263bc 1178{
590b8cd4 1179 struct if_announcemsghdr *ifan;
841ab66c 1180 struct mbuf *m;
984263bc
MD
1181
1182 if (route_cb.any_count == 0)
841ab66c 1183 return NULL;
f23061d4 1184
841ab66c
SZ
1185 bzero(info, sizeof(*info));
1186 m = rt_msg_mbuf(type, info);
984263bc 1187 if (m == NULL)
841ab66c 1188 return NULL;
f23061d4 1189
984263bc
MD
1190 ifan = mtod(m, struct if_announcemsghdr *);
1191 ifan->ifan_index = ifp->if_index;
5fe66e68 1192 strlcpy(ifan->ifan_name, ifp->if_xname, sizeof ifan->ifan_name);
984263bc 1193 ifan->ifan_what = what;
841ab66c
SZ
1194 return m;
1195}
1196
1197/*
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.
1201 */
1202void
1203rt_ieee80211msg(struct ifnet *ifp, int what, void *data, size_t data_len)
1204{
1205 struct rt_addrinfo info;
1206 struct mbuf *m;
f23061d4 1207
841ab66c
SZ
1208 m = rt_makeifannouncemsg(ifp, RTM_IEEE80211, what, &info);
1209 if (m == NULL)
1210 return;
1211
1212 /*
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.
1216 *
1217 * NB: we assume m is a single mbuf.
1218 */
1219 if (data_len > M_TRAILINGSPACE(m)) {
e28d8186 1220 /* XXX use m_getb(data_len, MB_DONTWAIT, MT_DATA, 0); */
b11e1686 1221 struct mbuf *n = m_get(MB_DONTWAIT, MT_DATA);
841ab66c
SZ
1222 if (n == NULL) {
1223 m_freem(m);
1224 return;
1225 }
e28d8186 1226 KKASSERT(data_len <= M_TRAILINGSPACE(n));
841ab66c
SZ
1227 bcopy(data, mtod(n, void *), data_len);
1228 n->m_len = data_len;
1229 m->m_next = n;
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;
1233 }
e9fa4b60 1234 mbuftrackid(m, 33);
841ab66c
SZ
1235 if (m->m_flags & M_PKTHDR)
1236 m->m_pkthdr.len += data_len;
1237 mtod(m, struct if_announcemsghdr *)->ifan_msglen += data_len;
590b8cd4 1238 rts_input(m, 0);
841ab66c
SZ
1239}
1240
1241/*
1242 * This is called to generate routing socket messages indicating
1243 * network interface arrival and departure.
1244 */
1245void
1246rt_ifannouncemsg(struct ifnet *ifp, int what)
1247{
1248 struct rt_addrinfo addrinfo;
1249 struct mbuf *m;
1250
1251 m = rt_makeifannouncemsg(ifp, RTM_IFANNOUNCE, what, &addrinfo);
1252 if (m != NULL)
1253 rts_input(m, 0);
590b8cd4 1254}
f23061d4 1255
590b8cd4
JH
1256static int
1257resizewalkarg(struct walkarg *w, int len)
1258{
1259 void *newptr;
1260
efda3bd0 1261 newptr = kmalloc(len, M_RTABLE, M_INTWAIT | M_NULLOK);
590b8cd4
JH
1262 if (newptr == NULL)
1263 return (ENOMEM);
1264 if (w->w_tmem != NULL)
efda3bd0 1265 kfree(w->w_tmem, M_RTABLE);
590b8cd4
JH
1266 w->w_tmem = newptr;
1267 w->w_tmemsize = len;
1268 return (0);
1269}
984263bc
MD
1270
1271/*
1272 * This is used in dumping the kernel table via sysctl().
1273 */
1274int
590b8cd4 1275sysctl_dumpentry(struct radix_node *rn, void *vw)
984263bc 1276{
82ed7fc2
RG
1277 struct walkarg *w = vw;
1278 struct rtentry *rt = (struct rtentry *)rn;
590b8cd4
JH
1279 struct rt_addrinfo rtinfo;
1280 int error, msglen;
984263bc
MD
1281
1282 if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg))
1283 return 0;
f23061d4 1284
590b8cd4
JH
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;
ef87f48d 1290 if (rt->rt_ifp != NULL) {
590b8cd4 1291 rtinfo.rti_ifpaddr =
b2632176 1292 TAILQ_FIRST(&rt->rt_ifp->if_addrheads[mycpuid])->ifa->ifa_addr;
590b8cd4 1293 rtinfo.rti_ifaaddr = rt->rt_ifa->ifa_addr;
984263bc 1294 if (rt->rt_ifp->if_flags & IFF_POINTOPOINT)
590b8cd4 1295 rtinfo.rti_bcastaddr = rt->rt_ifa->ifa_dstaddr;
984263bc 1296 }
590b8cd4
JH
1297 msglen = rt_msgsize(RTM_GET, &rtinfo);
1298 if (w->w_tmemsize < msglen && resizewalkarg(w, msglen) != 0)
1299 return (ENOMEM);
1300 rt_msg_buffer(RTM_GET, &rtinfo, w->w_tmem, msglen);
1301 if (w->w_req != NULL) {
8e63efec 1302 struct rt_msghdr *rtm = w->w_tmem;
984263bc
MD
1303
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;
590b8cd4
JH
1309 rtm->rtm_addrs = rtinfo.rti_addrs;
1310 error = SYSCTL_OUT(w->w_req, rtm, msglen);
984263bc
MD
1311 return (error);
1312 }
590b8cd4 1313 return (0);
984263bc
MD
1314}
1315
590b8cd4
JH
1316static int
1317sysctl_iflist(int af, struct walkarg *w)
984263bc 1318{
82ed7fc2 1319 struct ifnet *ifp;
590b8cd4
JH
1320 struct rt_addrinfo rtinfo;
1321 int msglen, error;
984263bc 1322
590b8cd4 1323 bzero(&rtinfo, sizeof(struct rt_addrinfo));
984263bc 1324 TAILQ_FOREACH(ifp, &ifnet, if_link) {
b2632176
SZ
1325 struct ifaddr_container *ifac;
1326 struct ifaddr *ifa;
1327
984263bc
MD
1328 if (w->w_arg && w->w_arg != ifp->if_index)
1329 continue;
b2632176
SZ
1330 ifac = TAILQ_FIRST(&ifp->if_addrheads[mycpuid]);
1331 ifa = ifac->ifa;
590b8cd4
JH
1332 rtinfo.rti_ifpaddr = ifa->ifa_addr;
1333 msglen = rt_msgsize(RTM_IFINFO, &rtinfo);
1334 if (w->w_tmemsize < msglen && resizewalkarg(w, msglen) != 0)
1335 return (ENOMEM);
1336 rt_msg_buffer(RTM_IFINFO, &rtinfo, w->w_tmem, msglen);
1337 rtinfo.rti_ifpaddr = NULL;
ef87f48d 1338 if (w->w_req != NULL && w->w_tmem != NULL) {
8e63efec 1339 struct if_msghdr *ifm = w->w_tmem;
984263bc 1340
984263bc 1341 ifm->ifm_index = ifp->if_index;
9c095379 1342 ifm->ifm_flags = ifp->if_flags;
984263bc 1343 ifm->ifm_data = ifp->if_data;
590b8cd4
JH
1344 ifm->ifm_addrs = rtinfo.rti_addrs;
1345 error = SYSCTL_OUT(w->w_req, ifm, msglen);
984263bc
MD
1346 if (error)
1347 return (error);
1348 }
b2632176
SZ
1349 while ((ifac = TAILQ_NEXT(ifac, ifa_link)) != NULL) {
1350 ifa = ifac->ifa;
1351
984263bc
MD
1352 if (af && af != ifa->ifa_addr->sa_family)
1353 continue;
590b8cd4 1354 if (curproc->p_ucred->cr_prison &&
87de5057 1355 prison_if(curproc->p_ucred, ifa->ifa_addr))
984263bc 1356 continue;
590b8cd4
JH
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)
1363 return (ENOMEM);
1364 rt_msg_buffer(RTM_NEWADDR, &rtinfo, w->w_tmem, msglen);
1365 if (w->w_req != NULL) {
8e63efec 1366 struct ifa_msghdr *ifam = w->w_tmem;
984263bc 1367
984263bc
MD
1368 ifam->ifam_index = ifa->ifa_ifp->if_index;
1369 ifam->ifam_flags = ifa->ifa_flags;
1370 ifam->ifam_metric = ifa->ifa_metric;
590b8cd4
JH
1371 ifam->ifam_addrs = rtinfo.rti_addrs;
1372 error = SYSCTL_OUT(w->w_req, w->w_tmem, msglen);
984263bc
MD
1373 if (error)
1374 return (error);
1375 }
1376 }
590b8cd4
JH
1377 rtinfo.rti_netmask = NULL;
1378 rtinfo.rti_ifaaddr = NULL;
1379 rtinfo.rti_bcastaddr = NULL;
984263bc
MD
1380 }
1381 return (0);
1382}
1383
1384static int
1385sysctl_rtsock(SYSCTL_HANDLER_ARGS)
1386{
1387 int *name = (int *)arg1;
1388 u_int namelen = arg2;
82ed7fc2 1389 struct radix_node_head *rnh;
a49aa710 1390 int i, error = EINVAL;
ecdefdda 1391 int origcpu;
984263bc
MD
1392 u_char af;
1393 struct walkarg w;
1394
1395 name ++;
1396 namelen--;
1397 if (req->newptr)
1398 return (EPERM);
ecdefdda 1399 if (namelen != 3 && namelen != 4)
984263bc
MD
1400 return (EINVAL);
1401 af = name[0];
5fe66e68 1402 bzero(&w, sizeof w);
984263bc
MD
1403 w.w_op = name[1];
1404 w.w_arg = name[2];
1405 w.w_req = req;
1406
ecdefdda
MD
1407 /*
1408 * Optional third argument specifies cpu, used primarily for
1409 * debugging the route table.
1410 */
1411 if (namelen == 4) {
1412 if (name[3] < 0 || name[3] >= ncpus)
1413 return (EINVAL);
1414 origcpu = mycpuid;
1415 lwkt_migratecpu(name[3]);
1416 } else {
1417 origcpu = -1;
1418 }
a49aa710 1419 crit_enter();
984263bc 1420 switch (w.w_op) {
984263bc
MD
1421 case NET_RT_DUMP:
1422 case NET_RT_FLAGS:
1423 for (i = 1; i <= AF_MAX; i++)
ecdefdda
MD
1424 if ((rnh = rt_tables[mycpuid][i]) &&
1425 (af == 0 || af == i) &&
984263bc 1426 (error = rnh->rnh_walktree(rnh,
f23061d4 1427 sysctl_dumpentry, &w)))
984263bc
MD
1428 break;
1429 break;
1430
1431 case NET_RT_IFLIST:
1432 error = sysctl_iflist(af, &w);
1433 }
a49aa710 1434 crit_exit();
590b8cd4 1435 if (w.w_tmem != NULL)
efda3bd0 1436 kfree(w.w_tmem, M_RTABLE);
ecdefdda
MD
1437 if (origcpu >= 0)
1438 lwkt_migratecpu(origcpu);
984263bc
MD
1439 return (error);
1440}
1441
1442SYSCTL_NODE(_net, PF_ROUTE, routetable, CTLFLAG_RD, sysctl_rtsock, "");
1443
1444/*
1445 * Definitions of protocols supported in the ROUTE domain.
1446 */
1447
ba4257b8 1448static struct domain routedomain; /* or at least forward */
984263bc
MD
1449
1450static struct protosw routesw[] = {
002c1265
MD
1451 {
1452 .pr_type = SOCK_RAW,
1453 .pr_domain = &routedomain,
1454 .pr_protocol = 0,
1455 .pr_flags = PR_ATOMIC|PR_ADDR,
1456 .pr_input = NULL,
1457 .pr_output = route_output,
1458 .pr_ctlinput = raw_ctlinput,
1459 .pr_ctloutput = NULL,
1460 .pr_ctlport = cpu0_ctlport,
1461
1462 .pr_init = raw_init,
1463 .pr_usrreqs = &route_usrreqs
1464 }
984263bc
MD
1465};
1466
ba4257b8 1467static struct domain routedomain = {
9c70fe43
JH
1468 PF_ROUTE, "route", NULL, NULL, NULL,
1469 routesw, &routesw[(sizeof routesw)/(sizeof routesw[0])],
1470};
984263bc
MD
1471
1472DOMAIN_SET(route);
47db7f9b 1473