USB - umass - fix reentrancy issues when rescanning the bus.
[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>
a49aa710 84#include <sys/thread2.h>
deffea2e 85
984263bc
MD
86#include <net/if.h>
87#include <net/route.h>
88#include <net/raw_cb.h>
47db7f9b 89#include <net/netmsg2.h>
984263bc 90
78812139
EN
91#ifdef SCTP
92extern void sctp_add_ip_address(struct ifaddr *ifa);
93extern void sctp_delete_ip_address(struct ifaddr *ifa);
94#endif /* SCTP */
95
984263bc
MD
96MALLOC_DEFINE(M_RTABLE, "routetbl", "routing tables");
97
f23061d4
JH
98static struct route_cb {
99 int ip_count;
100 int ip6_count;
101 int ipx_count;
102 int ns_count;
103 int any_count;
104} route_cb;
105
590b8cd4 106static const struct sockaddr route_src = { 2, PF_ROUTE, };
984263bc
MD
107
108struct walkarg {
109 int w_tmemsize;
110 int w_op, w_arg;
8e63efec 111 void *w_tmem;
984263bc
MD
112 struct sysctl_req *w_req;
113};
114
115static struct mbuf *
590b8cd4
JH
116 rt_msg_mbuf (int, struct rt_addrinfo *);
117static void rt_msg_buffer (int, struct rt_addrinfo *, void *buf, int len);
118static int rt_msgsize (int type, struct rt_addrinfo *rtinfo);
ef87f48d 119static int rt_xaddrs (char *, char *, struct rt_addrinfo *);
158abb01
RG
120static int sysctl_dumpentry (struct radix_node *rn, void *vw);
121static int sysctl_iflist (int af, struct walkarg *w);
deffea2e 122static int route_output(struct mbuf *, struct socket *, ...);
47e78ae8 123static void rt_setmetrics (u_long, struct rt_metrics *,
f23061d4 124 struct rt_metrics *);
984263bc 125
984263bc
MD
126/*
127 * It really doesn't make any sense at all for this code to share much
128 * with raw_usrreq.c, since its functionality is so restricted. XXX
129 */
130static int
131rts_abort(struct socket *so)
132{
a49aa710 133 int error;
f23061d4 134
a49aa710 135 crit_enter();
984263bc 136 error = raw_usrreqs.pru_abort(so);
a49aa710 137 crit_exit();
984263bc
MD
138 return error;
139}
140
141/* pru_accept is EOPNOTSUPP */
142
143static int
e4700d00 144rts_attach(struct socket *so, int proto, struct pru_attach_info *ai)
984263bc
MD
145{
146 struct rawcb *rp;
a49aa710 147 int error;
984263bc 148
ef87f48d 149 if (sotorawcb(so) != NULL)
984263bc 150 return EISCONN; /* XXX panic? */
f23061d4 151
efda3bd0 152 rp = kmalloc(sizeof *rp, M_PCB, M_WAITOK | M_ZERO);
984263bc
MD
153
154 /*
a49aa710 155 * The critical section is necessary to block protocols from sending
984263bc
MD
156 * error notifications (like RTM_REDIRECT or RTM_LOSING) while
157 * this PCB is extant but incompletely initialized.
158 * Probably we should try to do more of this work beforehand and
a49aa710 159 * eliminate the critical section.
984263bc 160 */
a49aa710 161 crit_enter();
ef87f48d 162 so->so_pcb = rp;
e4700d00 163 error = raw_attach(so, proto, ai->sb_rlimit);
984263bc
MD
164 rp = sotorawcb(so);
165 if (error) {
a49aa710 166 crit_exit();
efda3bd0 167 kfree(rp, M_PCB);
984263bc
MD
168 return error;
169 }
170 switch(rp->rcb_proto.sp_protocol) {
171 case AF_INET:
172 route_cb.ip_count++;
173 break;
174 case AF_INET6:
175 route_cb.ip6_count++;
176 break;
177 case AF_IPX:
178 route_cb.ipx_count++;
179 break;
180 case AF_NS:
181 route_cb.ns_count++;
182 break;
183 }
184 rp->rcb_faddr = &route_src;
185 route_cb.any_count++;
186 soisconnected(so);
187 so->so_options |= SO_USELOOPBACK;
a49aa710 188 crit_exit();
984263bc
MD
189 return 0;
190}
191
192static int
dadab5e9 193rts_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
984263bc 194{
a49aa710 195 int error;
f23061d4 196
a49aa710 197 crit_enter();
dadab5e9 198 error = raw_usrreqs.pru_bind(so, nam, td); /* xxx just EINVAL */
a49aa710 199 crit_exit();
984263bc
MD
200 return error;
201}
202
203static int
dadab5e9 204rts_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
984263bc 205{
a49aa710 206 int error;
f23061d4 207
a49aa710 208 crit_enter();
dadab5e9 209 error = raw_usrreqs.pru_connect(so, nam, td); /* XXX just EINVAL */
a49aa710 210 crit_exit();
984263bc
MD
211 return error;
212}
213
214/* pru_connect2 is EOPNOTSUPP */
215/* pru_control is EOPNOTSUPP */
216
217static int
218rts_detach(struct socket *so)
219{
220 struct rawcb *rp = sotorawcb(so);
a49aa710 221 int error;
984263bc 222
a49aa710 223 crit_enter();
ef87f48d 224 if (rp != NULL) {
984263bc
MD
225 switch(rp->rcb_proto.sp_protocol) {
226 case AF_INET:
227 route_cb.ip_count--;
228 break;
229 case AF_INET6:
230 route_cb.ip6_count--;
231 break;
232 case AF_IPX:
233 route_cb.ipx_count--;
234 break;
235 case AF_NS:
236 route_cb.ns_count--;
237 break;
238 }
239 route_cb.any_count--;
240 }
241 error = raw_usrreqs.pru_detach(so);
a49aa710 242 crit_exit();
984263bc
MD
243 return error;
244}
245
246static int
247rts_disconnect(struct socket *so)
248{
a49aa710 249 int error;
f23061d4 250
a49aa710 251 crit_enter();
984263bc 252 error = raw_usrreqs.pru_disconnect(so);
a49aa710 253 crit_exit();
984263bc
MD
254 return error;
255}
256
257/* pru_listen is EOPNOTSUPP */
258
259static int
260rts_peeraddr(struct socket *so, struct sockaddr **nam)
261{
a49aa710 262 int error;
f23061d4 263
a49aa710 264 crit_enter();
984263bc 265 error = raw_usrreqs.pru_peeraddr(so, nam);
a49aa710 266 crit_exit();
984263bc
MD
267 return error;
268}
269
270/* pru_rcvd is EOPNOTSUPP */
271/* pru_rcvoob is EOPNOTSUPP */
272
273static int
274rts_send(struct socket *so, int flags, struct mbuf *m, struct sockaddr *nam,
dadab5e9 275 struct mbuf *control, struct thread *td)
984263bc 276{
a49aa710 277 int error;
f23061d4 278
a49aa710 279 crit_enter();
dadab5e9 280 error = raw_usrreqs.pru_send(so, flags, m, nam, control, td);
a49aa710 281 crit_exit();
984263bc
MD
282 return error;
283}
284
285/* pru_sense is null */
286
287static int
288rts_shutdown(struct socket *so)
289{
a49aa710 290 int error;
f23061d4 291
a49aa710 292 crit_enter();
984263bc 293 error = raw_usrreqs.pru_shutdown(so);
a49aa710 294 crit_exit();
984263bc
MD
295 return error;
296}
297
298static int
299rts_sockaddr(struct socket *so, struct sockaddr **nam)
300{
a49aa710 301 int error;
f23061d4 302
a49aa710 303 crit_enter();
984263bc 304 error = raw_usrreqs.pru_sockaddr(so, nam);
a49aa710 305 crit_exit();
984263bc
MD
306 return error;
307}
308
309static struct pr_usrreqs route_usrreqs = {
fa5e758c
MD
310 .pru_abort = rts_abort,
311 .pru_accept = pru_accept_notsupp,
312 .pru_attach = rts_attach,
313 .pru_bind = rts_bind,
314 .pru_connect = rts_connect,
315 .pru_connect2 = pru_connect2_notsupp,
316 .pru_control = pru_control_notsupp,
317 .pru_detach = rts_detach,
318 .pru_disconnect = rts_disconnect,
319 .pru_listen = pru_listen_notsupp,
320 .pru_peeraddr = rts_peeraddr,
321 .pru_rcvd = pru_rcvd_notsupp,
322 .pru_rcvoob = pru_rcvoob_notsupp,
323 .pru_send = rts_send,
324 .pru_sense = pru_sense_null,
325 .pru_shutdown = rts_shutdown,
326 .pru_sockaddr = rts_sockaddr,
327 .pru_sosend = sosend,
328 .pru_soreceive = soreceive,
329 .pru_sopoll = sopoll
984263bc
MD
330};
331
590b8cd4
JH
332static __inline sa_family_t
333familyof(struct sockaddr *sa)
334{
335 return (sa != NULL ? sa->sa_family : 0);
336}
337
47db7f9b
MD
338/*
339 * Routing socket input function. The packet must be serialized onto cpu 0.
340 * We use the cpu0_soport() netisr processing loop to handle it.
341 *
342 * This looks messy but it means that anyone, including interrupt code,
343 * can send a message to the routing socket.
344 */
590b8cd4 345static void
47db7f9b 346rts_input_handler(struct netmsg *msg)
590b8cd4
JH
347{
348 static const struct sockaddr route_dst = { 2, PF_ROUTE, };
47db7f9b
MD
349 struct sockproto route_proto;
350 struct netmsg_packet *pmsg;
351 struct mbuf *m;
352 sa_family_t family;
d16fa201 353 struct rawcb *skip;
47db7f9b
MD
354
355 pmsg = (void *)msg;
47db7f9b
MD
356 family = pmsg->nm_netmsg.nm_lmsg.u.ms_result;
357 route_proto.sp_family = PF_ROUTE;
358 route_proto.sp_protocol = family;
590b8cd4 359
d16fa201
SZ
360 m = pmsg->nm_packet;
361 M_ASSERTPKTHDR(m);
362
363 skip = m->m_pkthdr.header;
364 m->m_pkthdr.header = NULL;
365
366 raw_input(m, &route_proto, &route_src, &route_dst, skip);
590b8cd4
JH
367}
368
47db7f9b 369static void
d16fa201 370rts_input_skip(struct mbuf *m, sa_family_t family, struct rawcb *skip)
47db7f9b
MD
371{
372 struct netmsg_packet *pmsg;
373 lwkt_port_t port;
374
d16fa201
SZ
375 M_ASSERTPKTHDR(m);
376
934c6849 377 port = cpu0_soport(NULL, NULL, NULL, 0);
47db7f9b
MD
378 pmsg = &m->m_hdr.mh_netmsg;
379 netmsg_init(&pmsg->nm_netmsg, &netisr_apanic_rport,
380 0, rts_input_handler);
381 pmsg->nm_packet = m;
382 pmsg->nm_netmsg.nm_lmsg.u.ms_result = family;
d16fa201 383 m->m_pkthdr.header = skip; /* XXX steal field in pkthdr */
47db7f9b
MD
384 lwkt_sendmsg(port, &pmsg->nm_netmsg.nm_lmsg);
385}
386
d16fa201
SZ
387static __inline void
388rts_input(struct mbuf *m, sa_family_t family)
389{
390 rts_input_skip(m, family, NULL);
391}
392
590b8cd4
JH
393static void *
394reallocbuf(void *ptr, size_t len, size_t olen)
395{
396 void *newptr;
397
efda3bd0 398 newptr = kmalloc(len, M_RTABLE, M_INTWAIT | M_NULLOK);
590b8cd4
JH
399 if (newptr == NULL)
400 return NULL;
401 bcopy(ptr, newptr, olen);
efda3bd0 402 kfree(ptr, M_RTABLE);
590b8cd4
JH
403 return (newptr);
404}
405
8e63efec
JH
406/*
407 * Internal helper routine for route_output().
408 */
590b8cd4
JH
409static int
410fillrtmsg(struct rt_msghdr **prtm, struct rtentry *rt,
411 struct rt_addrinfo *rtinfo)
412{
413 int msglen;
414 struct rt_msghdr *rtm = *prtm;
415
416 /* Fill in rt_addrinfo for call to rt_msg_buffer(). */
417 rtinfo->rti_dst = rt_key(rt);
418 rtinfo->rti_gateway = rt->rt_gateway;
419 rtinfo->rti_netmask = rt_mask(rt); /* might be NULL */
420 rtinfo->rti_genmask = rt->rt_genmask; /* might be NULL */
421 if (rtm->rtm_addrs & (RTA_IFP | RTA_IFA)) {
422 if (rt->rt_ifp != NULL) {
423 rtinfo->rti_ifpaddr =
b2632176
SZ
424 TAILQ_FIRST(&rt->rt_ifp->if_addrheads[mycpuid])
425 ->ifa->ifa_addr;
590b8cd4
JH
426 rtinfo->rti_ifaaddr = rt->rt_ifa->ifa_addr;
427 if (rt->rt_ifp->if_flags & IFF_POINTOPOINT)
428 rtinfo->rti_bcastaddr = rt->rt_ifa->ifa_dstaddr;
429 rtm->rtm_index = rt->rt_ifp->if_index;
430 } else {
431 rtinfo->rti_ifpaddr = NULL;
432 rtinfo->rti_ifaaddr = NULL;
1730f0e1 433 }
9d1f1679
SZ
434 } else if (rt->rt_ifp != NULL) {
435 rtm->rtm_index = rt->rt_ifp->if_index;
590b8cd4
JH
436 }
437
438 msglen = rt_msgsize(rtm->rtm_type, rtinfo);
439 if (rtm->rtm_msglen < msglen) {
440 rtm = reallocbuf(rtm, msglen, rtm->rtm_msglen);
441 if (rtm == NULL)
442 return (ENOBUFS);
443 *prtm = rtm;
444 }
445 rt_msg_buffer(rtm->rtm_type, rtinfo, rtm, msglen);
446
447 rtm->rtm_flags = rt->rt_flags;
448 rtm->rtm_rmx = rt->rt_rmx;
449 rtm->rtm_addrs = rtinfo->rti_addrs;
450
451 return (0);
452}
453
ecdefdda
MD
454static void route_output_add_callback(int, int, struct rt_addrinfo *,
455 struct rtentry *, void *);
456static void route_output_delete_callback(int, int, struct rt_addrinfo *,
457 struct rtentry *, void *);
00104f14 458static int route_output_get_callback(int, struct rt_addrinfo *,
b5e01dd3 459 struct rtentry *, void *, int);
2c62ae3b 460static int route_output_change_callback(int, struct rt_addrinfo *,
b5e01dd3 461 struct rtentry *, void *, int);
ae2a5ccd 462static int route_output_lock_callback(int, struct rt_addrinfo *,
b5e01dd3 463 struct rtentry *, void *, int);
ecdefdda 464
984263bc
MD
465/*ARGSUSED*/
466static int
deffea2e 467route_output(struct mbuf *m, struct socket *so, ...)
984263bc 468{
2e9572df 469 struct rt_msghdr *rtm = NULL;
f23061d4 470 struct rawcb *rp = NULL;
deffea2e 471 struct pr_output_info *oi;
590b8cd4 472 struct rt_addrinfo rtinfo;
ef87f48d 473 int len, error = 0;
deffea2e
JS
474 __va_list ap;
475
1cfcc9b6
SZ
476 M_ASSERTPKTHDR(m);
477
deffea2e
JS
478 __va_start(ap, so);
479 oi = __va_arg(ap, struct pr_output_info *);
480 __va_end(ap);
984263bc 481
ef87f48d 482#define gotoerr(e) { error = e; goto flush;}
590b8cd4
JH
483
484 if (m == NULL ||
485 (m->m_len < sizeof(long) &&
486 (m = m_pullup(m, sizeof(long))) == NULL))
984263bc 487 return (ENOBUFS);
984263bc 488 len = m->m_pkthdr.len;
590b8cd4 489 if (len < sizeof(struct rt_msghdr) ||
984263bc 490 len != mtod(m, struct rt_msghdr *)->rtm_msglen) {
590b8cd4 491 rtinfo.rti_dst = NULL;
ef87f48d 492 gotoerr(EINVAL);
984263bc 493 }
efda3bd0 494 rtm = kmalloc(len, M_RTABLE, M_INTWAIT | M_NULLOK);
ef87f48d 495 if (rtm == NULL) {
590b8cd4 496 rtinfo.rti_dst = NULL;
ef87f48d 497 gotoerr(ENOBUFS);
984263bc
MD
498 }
499 m_copydata(m, 0, len, (caddr_t)rtm);
500 if (rtm->rtm_version != RTM_VERSION) {
590b8cd4 501 rtinfo.rti_dst = NULL;
ef87f48d 502 gotoerr(EPROTONOSUPPORT);
984263bc 503 }
47e78ae8 504 rtm->rtm_pid = oi->p_pid;
590b8cd4
JH
505 bzero(&rtinfo, sizeof(struct rt_addrinfo));
506 rtinfo.rti_addrs = rtm->rtm_addrs;
507 if (rt_xaddrs((char *)(rtm + 1), (char *)rtm + len, &rtinfo) != 0) {
508 rtinfo.rti_dst = NULL;
ef87f48d 509 gotoerr(EINVAL);
984263bc 510 }
590b8cd4
JH
511 rtinfo.rti_flags = rtm->rtm_flags;
512 if (rtinfo.rti_dst == NULL || rtinfo.rti_dst->sa_family >= AF_MAX ||
513 (rtinfo.rti_gateway && rtinfo.rti_gateway->sa_family >= AF_MAX))
ef87f48d
JH
514 gotoerr(EINVAL);
515
590b8cd4 516 if (rtinfo.rti_genmask != NULL) {
ac0f13c5
SZ
517 error = rtmask_add_global(rtinfo.rti_genmask);
518 if (error)
519 goto flush;
984263bc
MD
520 }
521
522 /*
523 * Verify that the caller has the appropriate privilege; RTM_GET
524 * is the only operation the non-superuser is allowed.
525 */
02650706
SZ
526 if (rtm->rtm_type != RTM_GET &&
527 priv_check_cred(so->so_cred, PRIV_ROOT, 0) != 0)
ef87f48d 528 gotoerr(EPERM);
984263bc
MD
529
530 switch (rtm->rtm_type) {
984263bc 531 case RTM_ADD:
ecdefdda
MD
532 if (rtinfo.rti_gateway == NULL) {
533 error = EINVAL;
534 } else {
535 error = rtrequest1_global(RTM_ADD, &rtinfo,
536 route_output_add_callback, rtm);
984263bc
MD
537 }
538 break;
984263bc 539 case RTM_DELETE:
ecdefdda
MD
540 /*
541 * note: &rtm passed as argument so 'rtm' can be replaced.
542 */
543 error = rtrequest1_global(RTM_DELETE, &rtinfo,
544 route_output_delete_callback, &rtm);
984263bc 545 break;
984263bc 546 case RTM_GET:
00104f14
SZ
547 /*
548 * note: &rtm passed as argument so 'rtm' can be replaced.
549 */
550 error = rtsearch_global(RTM_GET, &rtinfo,
551 route_output_get_callback, &rtm,
552 RTS_NOEXACTMATCH);
ecdefdda
MD
553 break;
554 case RTM_CHANGE:
2c62ae3b
SZ
555 error = rtsearch_global(RTM_CHANGE, &rtinfo,
556 route_output_change_callback, rtm,
557 RTS_EXACTMATCH);
ecdefdda
MD
558 break;
559 case RTM_LOCK:
ae2a5ccd
SZ
560 error = rtsearch_global(RTM_LOCK, &rtinfo,
561 route_output_lock_callback, rtm,
562 RTS_EXACTMATCH);
590b8cd4 563 break;
984263bc 564 default:
ecdefdda
MD
565 error = EOPNOTSUPP;
566 break;
984263bc
MD
567 }
568
569flush:
590b8cd4
JH
570 if (rtm != NULL) {
571 if (error != 0)
984263bc
MD
572 rtm->rtm_errno = error;
573 else
574 rtm->rtm_flags |= RTF_DONE;
575 }
ecdefdda 576
984263bc
MD
577 /*
578 * Check to see if we don't want our own messages.
579 */
ef87f48d 580 if (!(so->so_options & SO_USELOOPBACK)) {
984263bc 581 if (route_cb.any_count <= 1) {
590b8cd4 582 if (rtm != NULL)
efda3bd0 583 kfree(rtm, M_RTABLE);
984263bc
MD
584 m_freem(m);
585 return (error);
586 }
587 /* There is another listener, so construct message */
588 rp = sotorawcb(so);
589 }
590b8cd4 590 if (rtm != NULL) {
984263bc
MD
591 m_copyback(m, 0, rtm->rtm_msglen, (caddr_t)rtm);
592 if (m->m_pkthdr.len < rtm->rtm_msglen) {
593 m_freem(m);
594 m = NULL;
595 } else if (m->m_pkthdr.len > rtm->rtm_msglen)
596 m_adj(m, rtm->rtm_msglen - m->m_pkthdr.len);
efda3bd0 597 kfree(rtm, M_RTABLE);
984263bc 598 }
ef87f48d 599 if (m != NULL)
d16fa201 600 rts_input_skip(m, familyof(rtinfo.rti_dst), rp);
984263bc
MD
601 return (error);
602}
603
ecdefdda
MD
604static void
605route_output_add_callback(int cmd, int error, struct rt_addrinfo *rtinfo,
606 struct rtentry *rt, void *arg)
607{
608 struct rt_msghdr *rtm = arg;
609
610 if (error == 0 && rt != NULL) {
611 rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx,
612 &rt->rt_rmx);
613 rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits);
614 rt->rt_rmx.rmx_locks |=
615 (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks);
ac0f13c5
SZ
616 if (rtinfo->rti_genmask != NULL) {
617 rt->rt_genmask = rtmask_purelookup(rtinfo->rti_genmask);
618 if (rt->rt_genmask == NULL) {
619 /*
620 * This should not happen, since we
621 * have already installed genmask
622 * on each CPU before we reach here.
623 */
624 panic("genmask is gone!?");
625 }
626 } else {
627 rt->rt_genmask = NULL;
628 }
31f89d97 629 rtm->rtm_index = rt->rt_ifp->if_index;
ecdefdda
MD
630 }
631}
632
633static void
634route_output_delete_callback(int cmd, int error, struct rt_addrinfo *rtinfo,
635 struct rtentry *rt, void *arg)
636{
637 struct rt_msghdr **rtm = arg;
638
639 if (error == 0 && rt) {
640 ++rt->rt_refcnt;
641 if (fillrtmsg(rtm, rt, rtinfo) != 0) {
642 error = ENOBUFS;
643 /* XXX no way to return the error */
644 }
645 --rt->rt_refcnt;
646 }
647}
648
00104f14
SZ
649static int
650route_output_get_callback(int cmd, struct rt_addrinfo *rtinfo,
b5e01dd3 651 struct rtentry *rt, void *arg, int found_cnt)
00104f14
SZ
652{
653 struct rt_msghdr **rtm = arg;
654 int error, found = 0;
655
656 if (((rtinfo->rti_flags ^ rt->rt_flags) & RTF_HOST) == 0)
657 found = 1;
658
659 error = fillrtmsg(rtm, rt, rtinfo);
660 if (!error && found) {
661 /* Got the exact match, we could return now! */
662 error = EJUSTRETURN;
663 }
664 return error;
665}
666
2c62ae3b
SZ
667static int
668route_output_change_callback(int cmd, struct rt_addrinfo *rtinfo,
b5e01dd3 669 struct rtentry *rt, void *arg, int found_cnt)
ecdefdda
MD
670{
671 struct rt_msghdr *rtm = arg;
672 struct ifaddr *ifa;
2c62ae3b 673 int error = 0;
ecdefdda
MD
674
675 /*
676 * new gateway could require new ifaddr, ifp;
677 * flags may also be different; ifp may be specified
678 * by ll sockaddr when protocol address is ambiguous
679 */
680 if (((rt->rt_flags & RTF_GATEWAY) && rtinfo->rti_gateway != NULL) ||
fb41af81
SZ
681 rtinfo->rti_ifpaddr != NULL ||
682 (rtinfo->rti_ifaaddr != NULL &&
683 !sa_equal(rtinfo->rti_ifaaddr, rt->rt_ifa->ifa_addr))) {
ecdefdda
MD
684 error = rt_getifa(rtinfo);
685 if (error != 0)
686 goto done;
687 }
688 if (rtinfo->rti_gateway != NULL) {
b5e01dd3
SZ
689 /*
690 * We only need to generate rtmsg upon the
691 * first route to be changed.
692 */
afe44b68 693 error = rt_setgate(rt, rt_key(rt), rtinfo->rti_gateway,
b5e01dd3 694 found_cnt == 1 ? RTL_REPORTMSG : RTL_DONTREPORT);
ecdefdda
MD
695 if (error != 0)
696 goto done;
697 }
698 if ((ifa = rtinfo->rti_ifa) != NULL) {
699 struct ifaddr *oifa = rt->rt_ifa;
700
701 if (oifa != ifa) {
702 if (oifa && oifa->ifa_rtrequest)
703 oifa->ifa_rtrequest(RTM_DELETE, rt, rtinfo);
704 IFAFREE(rt->rt_ifa);
705 IFAREF(ifa);
706 rt->rt_ifa = ifa;
707 rt->rt_ifp = rtinfo->rti_ifp;
708 }
709 }
710 rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx, &rt->rt_rmx);
711 if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest)
2c62ae3b 712 rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, rtinfo);
ac0f13c5
SZ
713 if (rtinfo->rti_genmask != NULL) {
714 rt->rt_genmask = rtmask_purelookup(rtinfo->rti_genmask);
715 if (rt->rt_genmask == NULL) {
716 /*
717 * This should not happen, since we
718 * have already installed genmask
719 * on each CPU before we reach here.
720 */
721 panic("genmask is gone!?\n");
722 }
723 }
31f89d97 724 rtm->rtm_index = rt->rt_ifp->if_index;
ecdefdda 725done:
2c62ae3b 726 return error;
ecdefdda
MD
727}
728
ae2a5ccd
SZ
729static int
730route_output_lock_callback(int cmd, struct rt_addrinfo *rtinfo,
b5e01dd3
SZ
731 struct rtentry *rt, void *arg,
732 int found_cnt __unused)
ecdefdda
MD
733{
734 struct rt_msghdr *rtm = arg;
735
736 rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits);
737 rt->rt_rmx.rmx_locks |=
738 (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks);
ae2a5ccd 739 return 0;
ecdefdda
MD
740}
741
984263bc 742static void
ef87f48d 743rt_setmetrics(u_long which, struct rt_metrics *in, struct rt_metrics *out)
984263bc 744{
2e9572df
JH
745#define setmetric(flag, elt) if (which & (flag)) out->elt = in->elt;
746 setmetric(RTV_RPIPE, rmx_recvpipe);
747 setmetric(RTV_SPIPE, rmx_sendpipe);
748 setmetric(RTV_SSTHRESH, rmx_ssthresh);
749 setmetric(RTV_RTT, rmx_rtt);
750 setmetric(RTV_RTTVAR, rmx_rttvar);
751 setmetric(RTV_HOPCOUNT, rmx_hopcount);
752 setmetric(RTV_MTU, rmx_mtu);
753 setmetric(RTV_EXPIRE, rmx_expire);
754#undef setmetric
984263bc
MD
755}
756
757#define ROUNDUP(a) \
758 ((a) > 0 ? (1 + (((a) - 1) | (sizeof(long) - 1))) : sizeof(long))
984263bc 759
984263bc
MD
760/*
761 * Extract the addresses of the passed sockaddrs.
762 * Do a little sanity checking so as to avoid bad memory references.
763 * This data is derived straight from userland.
764 */
765static int
ef87f48d 766rt_xaddrs(char *cp, char *cplim, struct rt_addrinfo *rtinfo)
984263bc 767{
82ed7fc2
RG
768 struct sockaddr *sa;
769 int i;
984263bc
MD
770
771 for (i = 0; (i < RTAX_MAX) && (cp < cplim); i++) {
772 if ((rtinfo->rti_addrs & (1 << i)) == 0)
773 continue;
774 sa = (struct sockaddr *)cp;
775 /*
776 * It won't fit.
777 */
0c3c561c 778 if ((cp + sa->sa_len) > cplim) {
984263bc
MD
779 return (EINVAL);
780 }
781
782 /*
ef87f48d 783 * There are no more... Quit now.
984263bc 784 * If there are more bits, they are in error.
ef87f48d 785 * I've seen this. route(1) can evidently generate these.
984263bc 786 * This causes kernel to core dump.
ef87f48d 787 * For compatibility, if we see this, point to a safe address.
984263bc
MD
788 */
789 if (sa->sa_len == 0) {
590b8cd4
JH
790 static struct sockaddr sa_zero = {
791 sizeof sa_zero, AF_INET,
792 };
793
984263bc 794 rtinfo->rti_info[i] = &sa_zero;
9b42cabe 795 kprintf("rtsock: received more addr bits than sockaddrs.\n");
984263bc
MD
796 return (0); /* should be EINVAL but for compat */
797 }
798
ef87f48d 799 /* Accept the sockaddr. */
984263bc 800 rtinfo->rti_info[i] = sa;
590b8cd4 801 cp += ROUNDUP(sa->sa_len);
984263bc
MD
802 }
803 return (0);
804}
805
590b8cd4
JH
806static int
807rt_msghdrsize(int type)
984263bc 808{
984263bc 809 switch (type) {
984263bc
MD
810 case RTM_DELADDR:
811 case RTM_NEWADDR:
590b8cd4 812 return sizeof(struct ifa_msghdr);
984263bc
MD
813 case RTM_DELMADDR:
814 case RTM_NEWMADDR:
590b8cd4 815 return sizeof(struct ifma_msghdr);
984263bc 816 case RTM_IFINFO:
590b8cd4 817 return sizeof(struct if_msghdr);
984263bc 818 case RTM_IFANNOUNCE:
841ab66c 819 case RTM_IEEE80211:
590b8cd4 820 return sizeof(struct if_announcemsghdr);
984263bc 821 default:
590b8cd4 822 return sizeof(struct rt_msghdr);
984263bc 823 }
590b8cd4
JH
824}
825
826static int
827rt_msgsize(int type, struct rt_addrinfo *rtinfo)
828{
829 int len, i;
830
831 len = rt_msghdrsize(type);
832 for (i = 0; i < RTAX_MAX; i++) {
833 if (rtinfo->rti_info[i] != NULL)
834 len += ROUNDUP(rtinfo->rti_info[i]->sa_len);
984263bc 835 }
590b8cd4
JH
836 len = ALIGN(len);
837 return len;
838}
839
840/*
841 * Build a routing message in a buffer.
842 * Copy the addresses in the rtinfo->rti_info[] sockaddr array
843 * to the end of the buffer after the message header.
844 *
845 * Set the rtinfo->rti_addrs bitmask of addresses present in rtinfo->rti_info[].
846 * This side-effect can be avoided if we reorder the addrs bitmask field in all
847 * the route messages to line up so we can set it here instead of back in the
848 * calling routine.
849 */
850static void
851rt_msg_buffer(int type, struct rt_addrinfo *rtinfo, void *buf, int msglen)
852{
853 struct rt_msghdr *rtm;
854 char *cp;
855 int dlen, i;
856
857 rtm = (struct rt_msghdr *) buf;
858 rtm->rtm_version = RTM_VERSION;
859 rtm->rtm_type = type;
860 rtm->rtm_msglen = msglen;
861
862 cp = (char *)buf + rt_msghdrsize(type);
863 rtinfo->rti_addrs = 0;
984263bc 864 for (i = 0; i < RTAX_MAX; i++) {
590b8cd4
JH
865 struct sockaddr *sa;
866
984263bc
MD
867 if ((sa = rtinfo->rti_info[i]) == NULL)
868 continue;
869 rtinfo->rti_addrs |= (1 << i);
870 dlen = ROUNDUP(sa->sa_len);
590b8cd4
JH
871 bcopy(sa, cp, dlen);
872 cp += dlen;
984263bc 873 }
984263bc
MD
874}
875
590b8cd4
JH
876/*
877 * Build a routing message in a mbuf chain.
878 * Copy the addresses in the rtinfo->rti_info[] sockaddr array
879 * to the end of the mbuf after the message header.
880 *
881 * Set the rtinfo->rti_addrs bitmask of addresses present in rtinfo->rti_info[].
882 * This side-effect can be avoided if we reorder the addrs bitmask field in all
883 * the route messages to line up so we can set it here instead of back in the
884 * calling routine.
885 */
886static struct mbuf *
887rt_msg_mbuf(int type, struct rt_addrinfo *rtinfo)
984263bc 888{
590b8cd4
JH
889 struct mbuf *m;
890 struct rt_msghdr *rtm;
891 int hlen, len;
82ed7fc2 892 int i;
984263bc 893
590b8cd4
JH
894 hlen = rt_msghdrsize(type);
895 KASSERT(hlen <= MCLBYTES, ("rt_msg_mbuf: hlen %d doesn't fit", hlen));
984263bc 896
50503f0f 897 m = m_getl(hlen, MB_DONTWAIT, MT_DATA, M_PKTHDR, NULL);
590b8cd4
JH
898 if (m == NULL)
899 return (NULL);
e9fa4b60 900 mbuftrackid(m, 32);
590b8cd4
JH
901 m->m_pkthdr.len = m->m_len = hlen;
902 m->m_pkthdr.rcvif = NULL;
903 rtinfo->rti_addrs = 0;
904 len = hlen;
984263bc 905 for (i = 0; i < RTAX_MAX; i++) {
82ed7fc2 906 struct sockaddr *sa;
590b8cd4 907 int dlen;
984263bc 908
ef87f48d 909 if ((sa = rtinfo->rti_info[i]) == NULL)
984263bc
MD
910 continue;
911 rtinfo->rti_addrs |= (1 << i);
912 dlen = ROUNDUP(sa->sa_len);
590b8cd4 913 m_copyback(m, len, dlen, (caddr_t)sa); /* can grow mbuf chain */
984263bc
MD
914 len += dlen;
915 }
590b8cd4
JH
916 if (m->m_pkthdr.len != len) { /* one of the m_copyback() calls failed */
917 m_freem(m);
918 return (NULL);
984263bc 919 }
590b8cd4
JH
920 rtm = mtod(m, struct rt_msghdr *);
921 bzero(rtm, hlen);
922 rtm->rtm_msglen = len;
923 rtm->rtm_version = RTM_VERSION;
924 rtm->rtm_type = type;
925 return (m);
984263bc
MD
926}
927
928/*
929 * This routine is called to generate a message from the routing
f23061d4 930 * socket indicating that a redirect has occurred, a routing lookup
984263bc
MD
931 * has failed, or that a protocol has detected timeouts to a particular
932 * destination.
933 */
934void
f23061d4 935rt_missmsg(int type, struct rt_addrinfo *rtinfo, int flags, int error)
984263bc 936{
0c3c561c 937 struct sockaddr *dst = rtinfo->rti_info[RTAX_DST];
82ed7fc2
RG
938 struct rt_msghdr *rtm;
939 struct mbuf *m;
984263bc
MD
940
941 if (route_cb.any_count == 0)
942 return;
590b8cd4 943 m = rt_msg_mbuf(type, rtinfo);
ef87f48d 944 if (m == NULL)
984263bc
MD
945 return;
946 rtm = mtod(m, struct rt_msghdr *);
947 rtm->rtm_flags = RTF_DONE | flags;
948 rtm->rtm_errno = error;
949 rtm->rtm_addrs = rtinfo->rti_addrs;
590b8cd4 950 rts_input(m, familyof(dst));
984263bc
MD
951}
952
f3ed2586
JH
953void
954rt_dstmsg(int type, struct sockaddr *dst, int error)
955{
956 struct rt_msghdr *rtm;
957 struct rt_addrinfo addrs;
958 struct mbuf *m;
959
960 if (route_cb.any_count == 0)
961 return;
962 bzero(&addrs, sizeof(struct rt_addrinfo));
963 addrs.rti_info[RTAX_DST] = dst;
590b8cd4 964 m = rt_msg_mbuf(type, &addrs);
f3ed2586
JH
965 if (m == NULL)
966 return;
967 rtm = mtod(m, struct rt_msghdr *);
968 rtm->rtm_flags = RTF_DONE;
969 rtm->rtm_errno = error;
970 rtm->rtm_addrs = addrs.rti_addrs;
590b8cd4 971 rts_input(m, familyof(dst));
f3ed2586
JH
972}
973
984263bc
MD
974/*
975 * This routine is called to generate a message from the routing
976 * socket indicating that the status of a network interface has changed.
977 */
978void
0c3c561c 979rt_ifmsg(struct ifnet *ifp)
984263bc 980{
82ed7fc2 981 struct if_msghdr *ifm;
984263bc 982 struct mbuf *m;
590b8cd4 983 struct rt_addrinfo rtinfo;
984263bc
MD
984
985 if (route_cb.any_count == 0)
986 return;
590b8cd4
JH
987 bzero(&rtinfo, sizeof(struct rt_addrinfo));
988 m = rt_msg_mbuf(RTM_IFINFO, &rtinfo);
ef87f48d 989 if (m == NULL)
984263bc
MD
990 return;
991 ifm = mtod(m, struct if_msghdr *);
992 ifm->ifm_index = ifp->if_index;
9c095379 993 ifm->ifm_flags = ifp->if_flags;
984263bc 994 ifm->ifm_data = ifp->if_data;
590b8cd4
JH
995 ifm->ifm_addrs = 0;
996 rts_input(m, 0);
984263bc
MD
997}
998
372316d9
JH
999static void
1000rt_ifamsg(int cmd, struct ifaddr *ifa)
1001{
1002 struct ifa_msghdr *ifam;
590b8cd4 1003 struct rt_addrinfo rtinfo;
372316d9 1004 struct mbuf *m;
372316d9
JH
1005 struct ifnet *ifp = ifa->ifa_ifp;
1006
590b8cd4
JH
1007 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1008 rtinfo.rti_ifaaddr = ifa->ifa_addr;
b2632176
SZ
1009 rtinfo.rti_ifpaddr =
1010 TAILQ_FIRST(&ifp->if_addrheads[mycpuid])->ifa->ifa_addr;
590b8cd4
JH
1011 rtinfo.rti_netmask = ifa->ifa_netmask;
1012 rtinfo.rti_bcastaddr = ifa->ifa_dstaddr;
f23061d4 1013
590b8cd4 1014 m = rt_msg_mbuf(cmd, &rtinfo);
f23061d4 1015 if (m == NULL)
372316d9 1016 return;
f23061d4 1017
372316d9
JH
1018 ifam = mtod(m, struct ifa_msghdr *);
1019 ifam->ifam_index = ifp->if_index;
1020 ifam->ifam_metric = ifa->ifa_metric;
1021 ifam->ifam_flags = ifa->ifa_flags;
590b8cd4 1022 ifam->ifam_addrs = rtinfo.rti_addrs;
f23061d4 1023
590b8cd4 1024 rts_input(m, familyof(ifa->ifa_addr));
372316d9
JH
1025}
1026
f3ed2586
JH
1027void
1028rt_rtmsg(int cmd, struct rtentry *rt, struct ifnet *ifp, int error)
372316d9
JH
1029{
1030 struct rt_msghdr *rtm;
590b8cd4 1031 struct rt_addrinfo rtinfo;
372316d9 1032 struct mbuf *m;
f3ed2586 1033 struct sockaddr *dst;
372316d9
JH
1034
1035 if (rt == NULL)
1036 return;
f23061d4 1037
590b8cd4
JH
1038 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1039 rtinfo.rti_dst = dst = rt_key(rt);
1040 rtinfo.rti_gateway = rt->rt_gateway;
1041 rtinfo.rti_netmask = rt_mask(rt);
b2632176
SZ
1042 if (ifp != NULL) {
1043 rtinfo.rti_ifpaddr =
1044 TAILQ_FIRST(&ifp->if_addrheads[mycpuid])->ifa->ifa_addr;
1045 }
590b8cd4 1046 rtinfo.rti_ifaaddr = rt->rt_ifa->ifa_addr;
f23061d4 1047
590b8cd4 1048 m = rt_msg_mbuf(cmd, &rtinfo);
f23061d4 1049 if (m == NULL)
372316d9 1050 return;
f23061d4 1051
372316d9 1052 rtm = mtod(m, struct rt_msghdr *);
f3ed2586
JH
1053 if (ifp != NULL)
1054 rtm->rtm_index = ifp->if_index;
372316d9
JH
1055 rtm->rtm_flags |= rt->rt_flags;
1056 rtm->rtm_errno = error;
590b8cd4 1057 rtm->rtm_addrs = rtinfo.rti_addrs;
f23061d4 1058
590b8cd4 1059 rts_input(m, familyof(dst));
372316d9
JH
1060}
1061
984263bc
MD
1062/*
1063 * This is called to generate messages from the routing socket
1064 * indicating a network interface has had addresses associated with it.
1065 * if we ever reverse the logic and replace messages TO the routing
1066 * socket indicate a request to configure interfaces, then it will
1067 * be unnecessary as the routing socket will automatically generate
1068 * copies of it.
1069 */
1070void
0c3c561c 1071rt_newaddrmsg(int cmd, struct ifaddr *ifa, int error, struct rtentry *rt)
984263bc 1072{
78812139
EN
1073#ifdef SCTP
1074 /*
1075 * notify the SCTP stack
1076 * this will only get called when an address is added/deleted
1077 * XXX pass the ifaddr struct instead if ifa->ifa_addr...
1078 */
1079 if (cmd == RTM_ADD)
1080 sctp_add_ip_address(ifa);
1081 else if (cmd == RTM_DELETE)
1082 sctp_delete_ip_address(ifa);
1083#endif /* SCTP */
1084
984263bc
MD
1085 if (route_cb.any_count == 0)
1086 return;
984263bc 1087
372316d9
JH
1088 if (cmd == RTM_ADD) {
1089 rt_ifamsg(RTM_NEWADDR, ifa);
f3ed2586 1090 rt_rtmsg(RTM_ADD, rt, ifa->ifa_ifp, error);
372316d9
JH
1091 } else {
1092 KASSERT((cmd == RTM_DELETE), ("unknown cmd %d", cmd));
f3ed2586 1093 rt_rtmsg(RTM_DELETE, rt, ifa->ifa_ifp, error);
372316d9 1094 rt_ifamsg(RTM_DELADDR, ifa);
984263bc
MD
1095 }
1096}
1097
1098/*
1099 * This is the analogue to the rt_newaddrmsg which performs the same
1100 * function but for multicast group memberhips. This is easier since
1101 * there is no route state to worry about.
1102 */
1103void
0c3c561c 1104rt_newmaddrmsg(int cmd, struct ifmultiaddr *ifma)
984263bc 1105{
590b8cd4 1106 struct rt_addrinfo rtinfo;
ef87f48d 1107 struct mbuf *m = NULL;
984263bc
MD
1108 struct ifnet *ifp = ifma->ifma_ifp;
1109 struct ifma_msghdr *ifmam;
1110
1111 if (route_cb.any_count == 0)
1112 return;
1113
590b8cd4
JH
1114 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1115 rtinfo.rti_ifaaddr = ifma->ifma_addr;
b2632176
SZ
1116 if (ifp != NULL && !TAILQ_EMPTY(&ifp->if_addrheads[mycpuid])) {
1117 rtinfo.rti_ifpaddr =
1118 TAILQ_FIRST(&ifp->if_addrheads[mycpuid])->ifa->ifa_addr;
1119 }
984263bc
MD
1120 /*
1121 * If a link-layer address is present, present it as a ``gateway''
1122 * (similarly to how ARP entries, e.g., are presented).
1123 */
590b8cd4 1124 rtinfo.rti_gateway = ifma->ifma_lladdr;
f23061d4 1125
590b8cd4 1126 m = rt_msg_mbuf(cmd, &rtinfo);
f23061d4 1127 if (m == NULL)
984263bc 1128 return;
f23061d4 1129
984263bc
MD
1130 ifmam = mtod(m, struct ifma_msghdr *);
1131 ifmam->ifmam_index = ifp->if_index;
590b8cd4 1132 ifmam->ifmam_addrs = rtinfo.rti_addrs;
f23061d4 1133
590b8cd4 1134 rts_input(m, familyof(ifma->ifma_addr));
984263bc
MD
1135}
1136
841ab66c
SZ
1137static struct mbuf *
1138rt_makeifannouncemsg(struct ifnet *ifp, int type, int what,
1139 struct rt_addrinfo *info)
984263bc 1140{
590b8cd4 1141 struct if_announcemsghdr *ifan;
841ab66c 1142 struct mbuf *m;
984263bc
MD
1143
1144 if (route_cb.any_count == 0)
841ab66c 1145 return NULL;
f23061d4 1146
841ab66c
SZ
1147 bzero(info, sizeof(*info));
1148 m = rt_msg_mbuf(type, info);
984263bc 1149 if (m == NULL)
841ab66c 1150 return NULL;
f23061d4 1151
984263bc
MD
1152 ifan = mtod(m, struct if_announcemsghdr *);
1153 ifan->ifan_index = ifp->if_index;
5fe66e68 1154 strlcpy(ifan->ifan_name, ifp->if_xname, sizeof ifan->ifan_name);
984263bc 1155 ifan->ifan_what = what;
841ab66c
SZ
1156 return m;
1157}
1158
1159/*
1160 * This is called to generate routing socket messages indicating
1161 * IEEE80211 wireless events.
1162 * XXX we piggyback on the RTM_IFANNOUNCE msg format in a clumsy way.
1163 */
1164void
1165rt_ieee80211msg(struct ifnet *ifp, int what, void *data, size_t data_len)
1166{
1167 struct rt_addrinfo info;
1168 struct mbuf *m;
f23061d4 1169
841ab66c
SZ
1170 m = rt_makeifannouncemsg(ifp, RTM_IEEE80211, what, &info);
1171 if (m == NULL)
1172 return;
1173
1174 /*
1175 * Append the ieee80211 data. Try to stick it in the
1176 * mbuf containing the ifannounce msg; otherwise allocate
1177 * a new mbuf and append.
1178 *
1179 * NB: we assume m is a single mbuf.
1180 */
1181 if (data_len > M_TRAILINGSPACE(m)) {
b11e1686 1182 struct mbuf *n = m_get(MB_DONTWAIT, MT_DATA);
841ab66c
SZ
1183 if (n == NULL) {
1184 m_freem(m);
1185 return;
1186 }
1187 bcopy(data, mtod(n, void *), data_len);
1188 n->m_len = data_len;
1189 m->m_next = n;
1190 } else if (data_len > 0) {
1191 bcopy(data, mtod(m, u_int8_t *) + m->m_len, data_len);
1192 m->m_len += data_len;
1193 }
e9fa4b60 1194 mbuftrackid(m, 33);
841ab66c
SZ
1195 if (m->m_flags & M_PKTHDR)
1196 m->m_pkthdr.len += data_len;
1197 mtod(m, struct if_announcemsghdr *)->ifan_msglen += data_len;
590b8cd4 1198 rts_input(m, 0);
841ab66c
SZ
1199}
1200
1201/*
1202 * This is called to generate routing socket messages indicating
1203 * network interface arrival and departure.
1204 */
1205void
1206rt_ifannouncemsg(struct ifnet *ifp, int what)
1207{
1208 struct rt_addrinfo addrinfo;
1209 struct mbuf *m;
1210
1211 m = rt_makeifannouncemsg(ifp, RTM_IFANNOUNCE, what, &addrinfo);
1212 if (m != NULL)
1213 rts_input(m, 0);
590b8cd4 1214}
f23061d4 1215
590b8cd4
JH
1216static int
1217resizewalkarg(struct walkarg *w, int len)
1218{
1219 void *newptr;
1220
efda3bd0 1221 newptr = kmalloc(len, M_RTABLE, M_INTWAIT | M_NULLOK);
590b8cd4
JH
1222 if (newptr == NULL)
1223 return (ENOMEM);
1224 if (w->w_tmem != NULL)
efda3bd0 1225 kfree(w->w_tmem, M_RTABLE);
590b8cd4
JH
1226 w->w_tmem = newptr;
1227 w->w_tmemsize = len;
1228 return (0);
1229}
984263bc
MD
1230
1231/*
1232 * This is used in dumping the kernel table via sysctl().
1233 */
1234int
590b8cd4 1235sysctl_dumpentry(struct radix_node *rn, void *vw)
984263bc 1236{
82ed7fc2
RG
1237 struct walkarg *w = vw;
1238 struct rtentry *rt = (struct rtentry *)rn;
590b8cd4
JH
1239 struct rt_addrinfo rtinfo;
1240 int error, msglen;
984263bc
MD
1241
1242 if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg))
1243 return 0;
f23061d4 1244
590b8cd4
JH
1245 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1246 rtinfo.rti_dst = rt_key(rt);
1247 rtinfo.rti_gateway = rt->rt_gateway;
1248 rtinfo.rti_netmask = rt_mask(rt);
1249 rtinfo.rti_genmask = rt->rt_genmask;
ef87f48d 1250 if (rt->rt_ifp != NULL) {
590b8cd4 1251 rtinfo.rti_ifpaddr =
b2632176 1252 TAILQ_FIRST(&rt->rt_ifp->if_addrheads[mycpuid])->ifa->ifa_addr;
590b8cd4 1253 rtinfo.rti_ifaaddr = rt->rt_ifa->ifa_addr;
984263bc 1254 if (rt->rt_ifp->if_flags & IFF_POINTOPOINT)
590b8cd4 1255 rtinfo.rti_bcastaddr = rt->rt_ifa->ifa_dstaddr;
984263bc 1256 }
590b8cd4
JH
1257 msglen = rt_msgsize(RTM_GET, &rtinfo);
1258 if (w->w_tmemsize < msglen && resizewalkarg(w, msglen) != 0)
1259 return (ENOMEM);
1260 rt_msg_buffer(RTM_GET, &rtinfo, w->w_tmem, msglen);
1261 if (w->w_req != NULL) {
8e63efec 1262 struct rt_msghdr *rtm = w->w_tmem;
984263bc
MD
1263
1264 rtm->rtm_flags = rt->rt_flags;
1265 rtm->rtm_use = rt->rt_use;
1266 rtm->rtm_rmx = rt->rt_rmx;
1267 rtm->rtm_index = rt->rt_ifp->if_index;
1268 rtm->rtm_errno = rtm->rtm_pid = rtm->rtm_seq = 0;
590b8cd4
JH
1269 rtm->rtm_addrs = rtinfo.rti_addrs;
1270 error = SYSCTL_OUT(w->w_req, rtm, msglen);
984263bc
MD
1271 return (error);
1272 }
590b8cd4 1273 return (0);
984263bc
MD
1274}
1275
590b8cd4
JH
1276static int
1277sysctl_iflist(int af, struct walkarg *w)
984263bc 1278{
82ed7fc2 1279 struct ifnet *ifp;
590b8cd4
JH
1280 struct rt_addrinfo rtinfo;
1281 int msglen, error;
984263bc 1282
590b8cd4 1283 bzero(&rtinfo, sizeof(struct rt_addrinfo));
984263bc 1284 TAILQ_FOREACH(ifp, &ifnet, if_link) {
b2632176
SZ
1285 struct ifaddr_container *ifac;
1286 struct ifaddr *ifa;
1287
984263bc
MD
1288 if (w->w_arg && w->w_arg != ifp->if_index)
1289 continue;
b2632176
SZ
1290 ifac = TAILQ_FIRST(&ifp->if_addrheads[mycpuid]);
1291 ifa = ifac->ifa;
590b8cd4
JH
1292 rtinfo.rti_ifpaddr = ifa->ifa_addr;
1293 msglen = rt_msgsize(RTM_IFINFO, &rtinfo);
1294 if (w->w_tmemsize < msglen && resizewalkarg(w, msglen) != 0)
1295 return (ENOMEM);
1296 rt_msg_buffer(RTM_IFINFO, &rtinfo, w->w_tmem, msglen);
1297 rtinfo.rti_ifpaddr = NULL;
ef87f48d 1298 if (w->w_req != NULL && w->w_tmem != NULL) {
8e63efec 1299 struct if_msghdr *ifm = w->w_tmem;
984263bc 1300
984263bc 1301 ifm->ifm_index = ifp->if_index;
9c095379 1302 ifm->ifm_flags = ifp->if_flags;
984263bc 1303 ifm->ifm_data = ifp->if_data;
590b8cd4
JH
1304 ifm->ifm_addrs = rtinfo.rti_addrs;
1305 error = SYSCTL_OUT(w->w_req, ifm, msglen);
984263bc
MD
1306 if (error)
1307 return (error);
1308 }
b2632176
SZ
1309 while ((ifac = TAILQ_NEXT(ifac, ifa_link)) != NULL) {
1310 ifa = ifac->ifa;
1311
984263bc
MD
1312 if (af && af != ifa->ifa_addr->sa_family)
1313 continue;
590b8cd4 1314 if (curproc->p_ucred->cr_prison &&
87de5057 1315 prison_if(curproc->p_ucred, ifa->ifa_addr))
984263bc 1316 continue;
590b8cd4
JH
1317 rtinfo.rti_ifaaddr = ifa->ifa_addr;
1318 rtinfo.rti_netmask = ifa->ifa_netmask;
1319 rtinfo.rti_bcastaddr = ifa->ifa_dstaddr;
1320 msglen = rt_msgsize(RTM_NEWADDR, &rtinfo);
1321 if (w->w_tmemsize < msglen &&
1322 resizewalkarg(w, msglen) != 0)
1323 return (ENOMEM);
1324 rt_msg_buffer(RTM_NEWADDR, &rtinfo, w->w_tmem, msglen);
1325 if (w->w_req != NULL) {
8e63efec 1326 struct ifa_msghdr *ifam = w->w_tmem;
984263bc 1327
984263bc
MD
1328 ifam->ifam_index = ifa->ifa_ifp->if_index;
1329 ifam->ifam_flags = ifa->ifa_flags;
1330 ifam->ifam_metric = ifa->ifa_metric;
590b8cd4
JH
1331 ifam->ifam_addrs = rtinfo.rti_addrs;
1332 error = SYSCTL_OUT(w->w_req, w->w_tmem, msglen);
984263bc
MD
1333 if (error)
1334 return (error);
1335 }
1336 }
590b8cd4
JH
1337 rtinfo.rti_netmask = NULL;
1338 rtinfo.rti_ifaaddr = NULL;
1339 rtinfo.rti_bcastaddr = NULL;
984263bc
MD
1340 }
1341 return (0);
1342}
1343
1344static int
1345sysctl_rtsock(SYSCTL_HANDLER_ARGS)
1346{
1347 int *name = (int *)arg1;
1348 u_int namelen = arg2;
82ed7fc2 1349 struct radix_node_head *rnh;
a49aa710 1350 int i, error = EINVAL;
ecdefdda 1351 int origcpu;
984263bc
MD
1352 u_char af;
1353 struct walkarg w;
1354
1355 name ++;
1356 namelen--;
1357 if (req->newptr)
1358 return (EPERM);
ecdefdda 1359 if (namelen != 3 && namelen != 4)
984263bc
MD
1360 return (EINVAL);
1361 af = name[0];
5fe66e68 1362 bzero(&w, sizeof w);
984263bc
MD
1363 w.w_op = name[1];
1364 w.w_arg = name[2];
1365 w.w_req = req;
1366
ecdefdda
MD
1367 /*
1368 * Optional third argument specifies cpu, used primarily for
1369 * debugging the route table.
1370 */
1371 if (namelen == 4) {
1372 if (name[3] < 0 || name[3] >= ncpus)
1373 return (EINVAL);
1374 origcpu = mycpuid;
1375 lwkt_migratecpu(name[3]);
1376 } else {
1377 origcpu = -1;
1378 }
a49aa710 1379 crit_enter();
984263bc 1380 switch (w.w_op) {
984263bc
MD
1381 case NET_RT_DUMP:
1382 case NET_RT_FLAGS:
1383 for (i = 1; i <= AF_MAX; i++)
ecdefdda
MD
1384 if ((rnh = rt_tables[mycpuid][i]) &&
1385 (af == 0 || af == i) &&
984263bc 1386 (error = rnh->rnh_walktree(rnh,
f23061d4 1387 sysctl_dumpentry, &w)))
984263bc
MD
1388 break;
1389 break;
1390
1391 case NET_RT_IFLIST:
1392 error = sysctl_iflist(af, &w);
1393 }
a49aa710 1394 crit_exit();
590b8cd4 1395 if (w.w_tmem != NULL)
efda3bd0 1396 kfree(w.w_tmem, M_RTABLE);
ecdefdda
MD
1397 if (origcpu >= 0)
1398 lwkt_migratecpu(origcpu);
984263bc
MD
1399 return (error);
1400}
1401
1402SYSCTL_NODE(_net, PF_ROUTE, routetable, CTLFLAG_RD, sysctl_rtsock, "");
1403
1404/*
1405 * Definitions of protocols supported in the ROUTE domain.
1406 */
1407
ba4257b8 1408static struct domain routedomain; /* or at least forward */
984263bc
MD
1409
1410static struct protosw routesw[] = {
1411{ SOCK_RAW, &routedomain, 0, PR_ATOMIC|PR_ADDR,
1412 0, route_output, raw_ctlinput, 0,
e3873585 1413 cpu0_soport, cpu0_ctlport,
984263bc
MD
1414 raw_init, 0, 0, 0,
1415 &route_usrreqs
1416}
1417};
1418
ba4257b8 1419static struct domain routedomain = {
9c70fe43
JH
1420 PF_ROUTE, "route", NULL, NULL, NULL,
1421 routesw, &routesw[(sizeof routesw)/(sizeof routesw[0])],
1422};
984263bc
MD
1423
1424DOMAIN_SET(route);
47db7f9b 1425