rtsock: Unbreak RTM_LOCK support by using rtsearch_global()
[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;
433 }
434 }
435
436 msglen = rt_msgsize(rtm->rtm_type, rtinfo);
437 if (rtm->rtm_msglen < msglen) {
438 rtm = reallocbuf(rtm, msglen, rtm->rtm_msglen);
439 if (rtm == NULL)
440 return (ENOBUFS);
441 *prtm = rtm;
442 }
443 rt_msg_buffer(rtm->rtm_type, rtinfo, rtm, msglen);
444
445 rtm->rtm_flags = rt->rt_flags;
446 rtm->rtm_rmx = rt->rt_rmx;
447 rtm->rtm_addrs = rtinfo->rti_addrs;
448
449 return (0);
450}
451
ecdefdda
MD
452static void route_output_add_callback(int, int, struct rt_addrinfo *,
453 struct rtentry *, void *);
454static void route_output_delete_callback(int, int, struct rt_addrinfo *,
455 struct rtentry *, void *);
2c62ae3b 456static int route_output_change_callback(int, struct rt_addrinfo *,
ecdefdda 457 struct rtentry *, void *);
ae2a5ccd
SZ
458static int route_output_lock_callback(int, struct rt_addrinfo *,
459 struct rtentry *, void *);
ecdefdda 460
984263bc
MD
461/*ARGSUSED*/
462static int
deffea2e 463route_output(struct mbuf *m, struct socket *so, ...)
984263bc 464{
2e9572df 465 struct rt_msghdr *rtm = NULL;
ecdefdda 466 struct rtentry *rt;
984263bc 467 struct radix_node_head *rnh;
f23061d4 468 struct rawcb *rp = NULL;
deffea2e 469 struct pr_output_info *oi;
590b8cd4 470 struct rt_addrinfo rtinfo;
ef87f48d 471 int len, error = 0;
deffea2e
JS
472 __va_list ap;
473
474 __va_start(ap, so);
475 oi = __va_arg(ap, struct pr_output_info *);
476 __va_end(ap);
984263bc 477
ef87f48d 478#define gotoerr(e) { error = e; goto flush;}
590b8cd4
JH
479
480 if (m == NULL ||
481 (m->m_len < sizeof(long) &&
482 (m = m_pullup(m, sizeof(long))) == NULL))
984263bc 483 return (ENOBUFS);
ef87f48d 484 if (!(m->m_flags & M_PKTHDR))
984263bc
MD
485 panic("route_output");
486 len = m->m_pkthdr.len;
590b8cd4 487 if (len < sizeof(struct rt_msghdr) ||
984263bc 488 len != mtod(m, struct rt_msghdr *)->rtm_msglen) {
590b8cd4 489 rtinfo.rti_dst = NULL;
ef87f48d 490 gotoerr(EINVAL);
984263bc 491 }
efda3bd0 492 rtm = kmalloc(len, M_RTABLE, M_INTWAIT | M_NULLOK);
ef87f48d 493 if (rtm == NULL) {
590b8cd4 494 rtinfo.rti_dst = NULL;
ef87f48d 495 gotoerr(ENOBUFS);
984263bc
MD
496 }
497 m_copydata(m, 0, len, (caddr_t)rtm);
498 if (rtm->rtm_version != RTM_VERSION) {
590b8cd4 499 rtinfo.rti_dst = NULL;
ef87f48d 500 gotoerr(EPROTONOSUPPORT);
984263bc 501 }
47e78ae8 502 rtm->rtm_pid = oi->p_pid;
590b8cd4
JH
503 bzero(&rtinfo, sizeof(struct rt_addrinfo));
504 rtinfo.rti_addrs = rtm->rtm_addrs;
505 if (rt_xaddrs((char *)(rtm + 1), (char *)rtm + len, &rtinfo) != 0) {
506 rtinfo.rti_dst = NULL;
ef87f48d 507 gotoerr(EINVAL);
984263bc 508 }
590b8cd4
JH
509 rtinfo.rti_flags = rtm->rtm_flags;
510 if (rtinfo.rti_dst == NULL || rtinfo.rti_dst->sa_family >= AF_MAX ||
511 (rtinfo.rti_gateway && rtinfo.rti_gateway->sa_family >= AF_MAX))
ef87f48d
JH
512 gotoerr(EINVAL);
513
590b8cd4
JH
514 if (rtinfo.rti_genmask != NULL) {
515 struct radix_node *n;
ef87f48d 516
590b8cd4
JH
517#define clen(s) (*(u_char *)(s))
518 n = rn_addmask((char *)rtinfo.rti_genmask, TRUE, 1);
519 if (n != NULL &&
520 rtinfo.rti_genmask->sa_len >= clen(n->rn_key) &&
521 bcmp((char *)rtinfo.rti_genmask + 1,
522 (char *)n->rn_key + 1, clen(n->rn_key) - 1) == 0)
523 rtinfo.rti_genmask = (struct sockaddr *)n->rn_key;
984263bc 524 else
ef87f48d 525 gotoerr(ENOBUFS);
984263bc
MD
526 }
527
528 /*
529 * Verify that the caller has the appropriate privilege; RTM_GET
530 * is the only operation the non-superuser is allowed.
531 */
895c1f85 532 if (rtm->rtm_type != RTM_GET && priv_check_cred(so->so_cred, PRIV_ROOT, 0) != 0)
ef87f48d 533 gotoerr(EPERM);
984263bc
MD
534
535 switch (rtm->rtm_type) {
984263bc 536 case RTM_ADD:
ecdefdda
MD
537 if (rtinfo.rti_gateway == NULL) {
538 error = EINVAL;
539 } else {
540 error = rtrequest1_global(RTM_ADD, &rtinfo,
541 route_output_add_callback, rtm);
984263bc
MD
542 }
543 break;
984263bc 544 case RTM_DELETE:
ecdefdda
MD
545 /*
546 * note: &rtm passed as argument so 'rtm' can be replaced.
547 */
548 error = rtrequest1_global(RTM_DELETE, &rtinfo,
549 route_output_delete_callback, &rtm);
984263bc 550 break;
984263bc 551 case RTM_GET:
ecdefdda
MD
552 rnh = rt_tables[mycpuid][rtinfo.rti_dst->sa_family];
553 if (rnh == NULL) {
554 error = EAFNOSUPPORT;
555 break;
556 }
590b8cd4
JH
557 rt = (struct rtentry *)
558 rnh->rnh_lookup((char *)rtinfo.rti_dst,
559 (char *)rtinfo.rti_netmask, rnh);
ecdefdda
MD
560 if (rt == NULL) {
561 error = ESRCH;
984263bc
MD
562 break;
563 }
ecdefdda
MD
564 rt->rt_refcnt++;
565 if (fillrtmsg(&rtm, rt, &rtinfo) != 0)
566 gotoerr(ENOBUFS);
567 --rt->rt_refcnt;
568 break;
569 case RTM_CHANGE:
2c62ae3b
SZ
570 error = rtsearch_global(RTM_CHANGE, &rtinfo,
571 route_output_change_callback, rtm,
572 RTS_EXACTMATCH);
ecdefdda
MD
573 break;
574 case RTM_LOCK:
ae2a5ccd
SZ
575 error = rtsearch_global(RTM_LOCK, &rtinfo,
576 route_output_lock_callback, rtm,
577 RTS_EXACTMATCH);
590b8cd4 578 break;
984263bc 579 default:
ecdefdda
MD
580 error = EOPNOTSUPP;
581 break;
984263bc
MD
582 }
583
584flush:
590b8cd4
JH
585 if (rtm != NULL) {
586 if (error != 0)
984263bc
MD
587 rtm->rtm_errno = error;
588 else
589 rtm->rtm_flags |= RTF_DONE;
590 }
ecdefdda 591
984263bc
MD
592 /*
593 * Check to see if we don't want our own messages.
594 */
ef87f48d 595 if (!(so->so_options & SO_USELOOPBACK)) {
984263bc 596 if (route_cb.any_count <= 1) {
590b8cd4 597 if (rtm != NULL)
efda3bd0 598 kfree(rtm, M_RTABLE);
984263bc
MD
599 m_freem(m);
600 return (error);
601 }
602 /* There is another listener, so construct message */
603 rp = sotorawcb(so);
604 }
590b8cd4 605 if (rtm != NULL) {
984263bc
MD
606 m_copyback(m, 0, rtm->rtm_msglen, (caddr_t)rtm);
607 if (m->m_pkthdr.len < rtm->rtm_msglen) {
608 m_freem(m);
609 m = NULL;
610 } else if (m->m_pkthdr.len > rtm->rtm_msglen)
611 m_adj(m, rtm->rtm_msglen - m->m_pkthdr.len);
efda3bd0 612 kfree(rtm, M_RTABLE);
984263bc 613 }
ef87f48d 614 if (m != NULL)
d16fa201 615 rts_input_skip(m, familyof(rtinfo.rti_dst), rp);
984263bc
MD
616 return (error);
617}
618
ecdefdda
MD
619static void
620route_output_add_callback(int cmd, int error, struct rt_addrinfo *rtinfo,
621 struct rtentry *rt, void *arg)
622{
623 struct rt_msghdr *rtm = arg;
624
625 if (error == 0 && rt != NULL) {
626 rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx,
627 &rt->rt_rmx);
628 rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits);
629 rt->rt_rmx.rmx_locks |=
630 (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks);
631 rt->rt_genmask = rtinfo->rti_genmask;
632 }
633}
634
635static void
636route_output_delete_callback(int cmd, int error, struct rt_addrinfo *rtinfo,
637 struct rtentry *rt, void *arg)
638{
639 struct rt_msghdr **rtm = arg;
640
641 if (error == 0 && rt) {
642 ++rt->rt_refcnt;
643 if (fillrtmsg(rtm, rt, rtinfo) != 0) {
644 error = ENOBUFS;
645 /* XXX no way to return the error */
646 }
647 --rt->rt_refcnt;
648 }
649}
650
2c62ae3b
SZ
651static int
652route_output_change_callback(int cmd, struct rt_addrinfo *rtinfo,
653 struct rtentry *rt, void *arg)
ecdefdda
MD
654{
655 struct rt_msghdr *rtm = arg;
656 struct ifaddr *ifa;
2c62ae3b 657 int error = 0;
ecdefdda
MD
658
659 /*
660 * new gateway could require new ifaddr, ifp;
661 * flags may also be different; ifp may be specified
662 * by ll sockaddr when protocol address is ambiguous
663 */
664 if (((rt->rt_flags & RTF_GATEWAY) && rtinfo->rti_gateway != NULL) ||
fb41af81
SZ
665 rtinfo->rti_ifpaddr != NULL ||
666 (rtinfo->rti_ifaaddr != NULL &&
667 !sa_equal(rtinfo->rti_ifaaddr, rt->rt_ifa->ifa_addr))) {
ecdefdda
MD
668 error = rt_getifa(rtinfo);
669 if (error != 0)
670 goto done;
671 }
672 if (rtinfo->rti_gateway != NULL) {
673 error = rt_setgate(rt, rt_key(rt), rtinfo->rti_gateway);
674 if (error != 0)
675 goto done;
676 }
677 if ((ifa = rtinfo->rti_ifa) != NULL) {
678 struct ifaddr *oifa = rt->rt_ifa;
679
680 if (oifa != ifa) {
681 if (oifa && oifa->ifa_rtrequest)
682 oifa->ifa_rtrequest(RTM_DELETE, rt, rtinfo);
683 IFAFREE(rt->rt_ifa);
684 IFAREF(ifa);
685 rt->rt_ifa = ifa;
686 rt->rt_ifp = rtinfo->rti_ifp;
687 }
688 }
689 rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx, &rt->rt_rmx);
690 if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest)
2c62ae3b 691 rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, rtinfo);
ecdefdda
MD
692 if (rtinfo->rti_genmask != NULL)
693 rt->rt_genmask = rtinfo->rti_genmask;
694done:
2c62ae3b 695 return error;
ecdefdda
MD
696}
697
ae2a5ccd
SZ
698static int
699route_output_lock_callback(int cmd, struct rt_addrinfo *rtinfo,
ecdefdda
MD
700 struct rtentry *rt, void *arg)
701{
702 struct rt_msghdr *rtm = arg;
703
704 rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits);
705 rt->rt_rmx.rmx_locks |=
706 (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks);
ae2a5ccd 707 return 0;
ecdefdda
MD
708}
709
984263bc 710static void
ef87f48d 711rt_setmetrics(u_long which, struct rt_metrics *in, struct rt_metrics *out)
984263bc 712{
2e9572df
JH
713#define setmetric(flag, elt) if (which & (flag)) out->elt = in->elt;
714 setmetric(RTV_RPIPE, rmx_recvpipe);
715 setmetric(RTV_SPIPE, rmx_sendpipe);
716 setmetric(RTV_SSTHRESH, rmx_ssthresh);
717 setmetric(RTV_RTT, rmx_rtt);
718 setmetric(RTV_RTTVAR, rmx_rttvar);
719 setmetric(RTV_HOPCOUNT, rmx_hopcount);
720 setmetric(RTV_MTU, rmx_mtu);
721 setmetric(RTV_EXPIRE, rmx_expire);
722#undef setmetric
984263bc
MD
723}
724
725#define ROUNDUP(a) \
726 ((a) > 0 ? (1 + (((a) - 1) | (sizeof(long) - 1))) : sizeof(long))
984263bc 727
984263bc
MD
728/*
729 * Extract the addresses of the passed sockaddrs.
730 * Do a little sanity checking so as to avoid bad memory references.
731 * This data is derived straight from userland.
732 */
733static int
ef87f48d 734rt_xaddrs(char *cp, char *cplim, struct rt_addrinfo *rtinfo)
984263bc 735{
82ed7fc2
RG
736 struct sockaddr *sa;
737 int i;
984263bc
MD
738
739 for (i = 0; (i < RTAX_MAX) && (cp < cplim); i++) {
740 if ((rtinfo->rti_addrs & (1 << i)) == 0)
741 continue;
742 sa = (struct sockaddr *)cp;
743 /*
744 * It won't fit.
745 */
0c3c561c 746 if ((cp + sa->sa_len) > cplim) {
984263bc
MD
747 return (EINVAL);
748 }
749
750 /*
ef87f48d 751 * There are no more... Quit now.
984263bc 752 * If there are more bits, they are in error.
ef87f48d 753 * I've seen this. route(1) can evidently generate these.
984263bc 754 * This causes kernel to core dump.
ef87f48d 755 * For compatibility, if we see this, point to a safe address.
984263bc
MD
756 */
757 if (sa->sa_len == 0) {
590b8cd4
JH
758 static struct sockaddr sa_zero = {
759 sizeof sa_zero, AF_INET,
760 };
761
984263bc 762 rtinfo->rti_info[i] = &sa_zero;
9b42cabe 763 kprintf("rtsock: received more addr bits than sockaddrs.\n");
984263bc
MD
764 return (0); /* should be EINVAL but for compat */
765 }
766
ef87f48d 767 /* Accept the sockaddr. */
984263bc 768 rtinfo->rti_info[i] = sa;
590b8cd4 769 cp += ROUNDUP(sa->sa_len);
984263bc
MD
770 }
771 return (0);
772}
773
590b8cd4
JH
774static int
775rt_msghdrsize(int type)
984263bc 776{
984263bc 777 switch (type) {
984263bc
MD
778 case RTM_DELADDR:
779 case RTM_NEWADDR:
590b8cd4 780 return sizeof(struct ifa_msghdr);
984263bc
MD
781 case RTM_DELMADDR:
782 case RTM_NEWMADDR:
590b8cd4 783 return sizeof(struct ifma_msghdr);
984263bc 784 case RTM_IFINFO:
590b8cd4 785 return sizeof(struct if_msghdr);
984263bc 786 case RTM_IFANNOUNCE:
841ab66c 787 case RTM_IEEE80211:
590b8cd4 788 return sizeof(struct if_announcemsghdr);
984263bc 789 default:
590b8cd4 790 return sizeof(struct rt_msghdr);
984263bc 791 }
590b8cd4
JH
792}
793
794static int
795rt_msgsize(int type, struct rt_addrinfo *rtinfo)
796{
797 int len, i;
798
799 len = rt_msghdrsize(type);
800 for (i = 0; i < RTAX_MAX; i++) {
801 if (rtinfo->rti_info[i] != NULL)
802 len += ROUNDUP(rtinfo->rti_info[i]->sa_len);
984263bc 803 }
590b8cd4
JH
804 len = ALIGN(len);
805 return len;
806}
807
808/*
809 * Build a routing message in a buffer.
810 * Copy the addresses in the rtinfo->rti_info[] sockaddr array
811 * to the end of the buffer after the message header.
812 *
813 * Set the rtinfo->rti_addrs bitmask of addresses present in rtinfo->rti_info[].
814 * This side-effect can be avoided if we reorder the addrs bitmask field in all
815 * the route messages to line up so we can set it here instead of back in the
816 * calling routine.
817 */
818static void
819rt_msg_buffer(int type, struct rt_addrinfo *rtinfo, void *buf, int msglen)
820{
821 struct rt_msghdr *rtm;
822 char *cp;
823 int dlen, i;
824
825 rtm = (struct rt_msghdr *) buf;
826 rtm->rtm_version = RTM_VERSION;
827 rtm->rtm_type = type;
828 rtm->rtm_msglen = msglen;
829
830 cp = (char *)buf + rt_msghdrsize(type);
831 rtinfo->rti_addrs = 0;
984263bc 832 for (i = 0; i < RTAX_MAX; i++) {
590b8cd4
JH
833 struct sockaddr *sa;
834
984263bc
MD
835 if ((sa = rtinfo->rti_info[i]) == NULL)
836 continue;
837 rtinfo->rti_addrs |= (1 << i);
838 dlen = ROUNDUP(sa->sa_len);
590b8cd4
JH
839 bcopy(sa, cp, dlen);
840 cp += dlen;
984263bc 841 }
984263bc
MD
842}
843
590b8cd4
JH
844/*
845 * Build a routing message in a mbuf chain.
846 * Copy the addresses in the rtinfo->rti_info[] sockaddr array
847 * to the end of the mbuf after the message header.
848 *
849 * Set the rtinfo->rti_addrs bitmask of addresses present in rtinfo->rti_info[].
850 * This side-effect can be avoided if we reorder the addrs bitmask field in all
851 * the route messages to line up so we can set it here instead of back in the
852 * calling routine.
853 */
854static struct mbuf *
855rt_msg_mbuf(int type, struct rt_addrinfo *rtinfo)
984263bc 856{
590b8cd4
JH
857 struct mbuf *m;
858 struct rt_msghdr *rtm;
859 int hlen, len;
82ed7fc2 860 int i;
984263bc 861
590b8cd4
JH
862 hlen = rt_msghdrsize(type);
863 KASSERT(hlen <= MCLBYTES, ("rt_msg_mbuf: hlen %d doesn't fit", hlen));
984263bc 864
50503f0f 865 m = m_getl(hlen, MB_DONTWAIT, MT_DATA, M_PKTHDR, NULL);
590b8cd4
JH
866 if (m == NULL)
867 return (NULL);
e9fa4b60 868 mbuftrackid(m, 32);
590b8cd4
JH
869 m->m_pkthdr.len = m->m_len = hlen;
870 m->m_pkthdr.rcvif = NULL;
871 rtinfo->rti_addrs = 0;
872 len = hlen;
984263bc 873 for (i = 0; i < RTAX_MAX; i++) {
82ed7fc2 874 struct sockaddr *sa;
590b8cd4 875 int dlen;
984263bc 876
ef87f48d 877 if ((sa = rtinfo->rti_info[i]) == NULL)
984263bc
MD
878 continue;
879 rtinfo->rti_addrs |= (1 << i);
880 dlen = ROUNDUP(sa->sa_len);
590b8cd4 881 m_copyback(m, len, dlen, (caddr_t)sa); /* can grow mbuf chain */
984263bc
MD
882 len += dlen;
883 }
590b8cd4
JH
884 if (m->m_pkthdr.len != len) { /* one of the m_copyback() calls failed */
885 m_freem(m);
886 return (NULL);
984263bc 887 }
590b8cd4
JH
888 rtm = mtod(m, struct rt_msghdr *);
889 bzero(rtm, hlen);
890 rtm->rtm_msglen = len;
891 rtm->rtm_version = RTM_VERSION;
892 rtm->rtm_type = type;
893 return (m);
984263bc
MD
894}
895
896/*
897 * This routine is called to generate a message from the routing
f23061d4 898 * socket indicating that a redirect has occurred, a routing lookup
984263bc
MD
899 * has failed, or that a protocol has detected timeouts to a particular
900 * destination.
901 */
902void
f23061d4 903rt_missmsg(int type, struct rt_addrinfo *rtinfo, int flags, int error)
984263bc 904{
0c3c561c 905 struct sockaddr *dst = rtinfo->rti_info[RTAX_DST];
82ed7fc2
RG
906 struct rt_msghdr *rtm;
907 struct mbuf *m;
984263bc
MD
908
909 if (route_cb.any_count == 0)
910 return;
590b8cd4 911 m = rt_msg_mbuf(type, rtinfo);
ef87f48d 912 if (m == NULL)
984263bc
MD
913 return;
914 rtm = mtod(m, struct rt_msghdr *);
915 rtm->rtm_flags = RTF_DONE | flags;
916 rtm->rtm_errno = error;
917 rtm->rtm_addrs = rtinfo->rti_addrs;
590b8cd4 918 rts_input(m, familyof(dst));
984263bc
MD
919}
920
f3ed2586
JH
921void
922rt_dstmsg(int type, struct sockaddr *dst, int error)
923{
924 struct rt_msghdr *rtm;
925 struct rt_addrinfo addrs;
926 struct mbuf *m;
927
928 if (route_cb.any_count == 0)
929 return;
930 bzero(&addrs, sizeof(struct rt_addrinfo));
931 addrs.rti_info[RTAX_DST] = dst;
590b8cd4 932 m = rt_msg_mbuf(type, &addrs);
f3ed2586
JH
933 if (m == NULL)
934 return;
935 rtm = mtod(m, struct rt_msghdr *);
936 rtm->rtm_flags = RTF_DONE;
937 rtm->rtm_errno = error;
938 rtm->rtm_addrs = addrs.rti_addrs;
590b8cd4 939 rts_input(m, familyof(dst));
f3ed2586
JH
940}
941
984263bc
MD
942/*
943 * This routine is called to generate a message from the routing
944 * socket indicating that the status of a network interface has changed.
945 */
946void
0c3c561c 947rt_ifmsg(struct ifnet *ifp)
984263bc 948{
82ed7fc2 949 struct if_msghdr *ifm;
984263bc 950 struct mbuf *m;
590b8cd4 951 struct rt_addrinfo rtinfo;
984263bc
MD
952
953 if (route_cb.any_count == 0)
954 return;
590b8cd4
JH
955 bzero(&rtinfo, sizeof(struct rt_addrinfo));
956 m = rt_msg_mbuf(RTM_IFINFO, &rtinfo);
ef87f48d 957 if (m == NULL)
984263bc
MD
958 return;
959 ifm = mtod(m, struct if_msghdr *);
960 ifm->ifm_index = ifp->if_index;
9c095379 961 ifm->ifm_flags = ifp->if_flags;
984263bc 962 ifm->ifm_data = ifp->if_data;
590b8cd4
JH
963 ifm->ifm_addrs = 0;
964 rts_input(m, 0);
984263bc
MD
965}
966
372316d9
JH
967static void
968rt_ifamsg(int cmd, struct ifaddr *ifa)
969{
970 struct ifa_msghdr *ifam;
590b8cd4 971 struct rt_addrinfo rtinfo;
372316d9 972 struct mbuf *m;
372316d9
JH
973 struct ifnet *ifp = ifa->ifa_ifp;
974
590b8cd4
JH
975 bzero(&rtinfo, sizeof(struct rt_addrinfo));
976 rtinfo.rti_ifaaddr = ifa->ifa_addr;
b2632176
SZ
977 rtinfo.rti_ifpaddr =
978 TAILQ_FIRST(&ifp->if_addrheads[mycpuid])->ifa->ifa_addr;
590b8cd4
JH
979 rtinfo.rti_netmask = ifa->ifa_netmask;
980 rtinfo.rti_bcastaddr = ifa->ifa_dstaddr;
f23061d4 981
590b8cd4 982 m = rt_msg_mbuf(cmd, &rtinfo);
f23061d4 983 if (m == NULL)
372316d9 984 return;
f23061d4 985
372316d9
JH
986 ifam = mtod(m, struct ifa_msghdr *);
987 ifam->ifam_index = ifp->if_index;
988 ifam->ifam_metric = ifa->ifa_metric;
989 ifam->ifam_flags = ifa->ifa_flags;
590b8cd4 990 ifam->ifam_addrs = rtinfo.rti_addrs;
f23061d4 991
590b8cd4 992 rts_input(m, familyof(ifa->ifa_addr));
372316d9
JH
993}
994
f3ed2586
JH
995void
996rt_rtmsg(int cmd, struct rtentry *rt, struct ifnet *ifp, int error)
372316d9
JH
997{
998 struct rt_msghdr *rtm;
590b8cd4 999 struct rt_addrinfo rtinfo;
372316d9 1000 struct mbuf *m;
f3ed2586 1001 struct sockaddr *dst;
372316d9
JH
1002
1003 if (rt == NULL)
1004 return;
f23061d4 1005
590b8cd4
JH
1006 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1007 rtinfo.rti_dst = dst = rt_key(rt);
1008 rtinfo.rti_gateway = rt->rt_gateway;
1009 rtinfo.rti_netmask = rt_mask(rt);
b2632176
SZ
1010 if (ifp != NULL) {
1011 rtinfo.rti_ifpaddr =
1012 TAILQ_FIRST(&ifp->if_addrheads[mycpuid])->ifa->ifa_addr;
1013 }
590b8cd4 1014 rtinfo.rti_ifaaddr = rt->rt_ifa->ifa_addr;
f23061d4 1015
590b8cd4 1016 m = rt_msg_mbuf(cmd, &rtinfo);
f23061d4 1017 if (m == NULL)
372316d9 1018 return;
f23061d4 1019
372316d9 1020 rtm = mtod(m, struct rt_msghdr *);
f3ed2586
JH
1021 if (ifp != NULL)
1022 rtm->rtm_index = ifp->if_index;
372316d9
JH
1023 rtm->rtm_flags |= rt->rt_flags;
1024 rtm->rtm_errno = error;
590b8cd4 1025 rtm->rtm_addrs = rtinfo.rti_addrs;
f23061d4 1026
590b8cd4 1027 rts_input(m, familyof(dst));
372316d9
JH
1028}
1029
984263bc
MD
1030/*
1031 * This is called to generate messages from the routing socket
1032 * indicating a network interface has had addresses associated with it.
1033 * if we ever reverse the logic and replace messages TO the routing
1034 * socket indicate a request to configure interfaces, then it will
1035 * be unnecessary as the routing socket will automatically generate
1036 * copies of it.
1037 */
1038void
0c3c561c 1039rt_newaddrmsg(int cmd, struct ifaddr *ifa, int error, struct rtentry *rt)
984263bc 1040{
78812139
EN
1041#ifdef SCTP
1042 /*
1043 * notify the SCTP stack
1044 * this will only get called when an address is added/deleted
1045 * XXX pass the ifaddr struct instead if ifa->ifa_addr...
1046 */
1047 if (cmd == RTM_ADD)
1048 sctp_add_ip_address(ifa);
1049 else if (cmd == RTM_DELETE)
1050 sctp_delete_ip_address(ifa);
1051#endif /* SCTP */
1052
984263bc
MD
1053 if (route_cb.any_count == 0)
1054 return;
984263bc 1055
372316d9
JH
1056 if (cmd == RTM_ADD) {
1057 rt_ifamsg(RTM_NEWADDR, ifa);
f3ed2586 1058 rt_rtmsg(RTM_ADD, rt, ifa->ifa_ifp, error);
372316d9
JH
1059 } else {
1060 KASSERT((cmd == RTM_DELETE), ("unknown cmd %d", cmd));
f3ed2586 1061 rt_rtmsg(RTM_DELETE, rt, ifa->ifa_ifp, error);
372316d9 1062 rt_ifamsg(RTM_DELADDR, ifa);
984263bc
MD
1063 }
1064}
1065
1066/*
1067 * This is the analogue to the rt_newaddrmsg which performs the same
1068 * function but for multicast group memberhips. This is easier since
1069 * there is no route state to worry about.
1070 */
1071void
0c3c561c 1072rt_newmaddrmsg(int cmd, struct ifmultiaddr *ifma)
984263bc 1073{
590b8cd4 1074 struct rt_addrinfo rtinfo;
ef87f48d 1075 struct mbuf *m = NULL;
984263bc
MD
1076 struct ifnet *ifp = ifma->ifma_ifp;
1077 struct ifma_msghdr *ifmam;
1078
1079 if (route_cb.any_count == 0)
1080 return;
1081
590b8cd4
JH
1082 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1083 rtinfo.rti_ifaaddr = ifma->ifma_addr;
b2632176
SZ
1084 if (ifp != NULL && !TAILQ_EMPTY(&ifp->if_addrheads[mycpuid])) {
1085 rtinfo.rti_ifpaddr =
1086 TAILQ_FIRST(&ifp->if_addrheads[mycpuid])->ifa->ifa_addr;
1087 }
984263bc
MD
1088 /*
1089 * If a link-layer address is present, present it as a ``gateway''
1090 * (similarly to how ARP entries, e.g., are presented).
1091 */
590b8cd4 1092 rtinfo.rti_gateway = ifma->ifma_lladdr;
f23061d4 1093
590b8cd4 1094 m = rt_msg_mbuf(cmd, &rtinfo);
f23061d4 1095 if (m == NULL)
984263bc 1096 return;
f23061d4 1097
984263bc
MD
1098 ifmam = mtod(m, struct ifma_msghdr *);
1099 ifmam->ifmam_index = ifp->if_index;
590b8cd4 1100 ifmam->ifmam_addrs = rtinfo.rti_addrs;
f23061d4 1101
590b8cd4 1102 rts_input(m, familyof(ifma->ifma_addr));
984263bc
MD
1103}
1104
841ab66c
SZ
1105static struct mbuf *
1106rt_makeifannouncemsg(struct ifnet *ifp, int type, int what,
1107 struct rt_addrinfo *info)
984263bc 1108{
590b8cd4 1109 struct if_announcemsghdr *ifan;
841ab66c 1110 struct mbuf *m;
984263bc
MD
1111
1112 if (route_cb.any_count == 0)
841ab66c 1113 return NULL;
f23061d4 1114
841ab66c
SZ
1115 bzero(info, sizeof(*info));
1116 m = rt_msg_mbuf(type, info);
984263bc 1117 if (m == NULL)
841ab66c 1118 return NULL;
f23061d4 1119
984263bc
MD
1120 ifan = mtod(m, struct if_announcemsghdr *);
1121 ifan->ifan_index = ifp->if_index;
5fe66e68 1122 strlcpy(ifan->ifan_name, ifp->if_xname, sizeof ifan->ifan_name);
984263bc 1123 ifan->ifan_what = what;
841ab66c
SZ
1124 return m;
1125}
1126
1127/*
1128 * This is called to generate routing socket messages indicating
1129 * IEEE80211 wireless events.
1130 * XXX we piggyback on the RTM_IFANNOUNCE msg format in a clumsy way.
1131 */
1132void
1133rt_ieee80211msg(struct ifnet *ifp, int what, void *data, size_t data_len)
1134{
1135 struct rt_addrinfo info;
1136 struct mbuf *m;
f23061d4 1137
841ab66c
SZ
1138 m = rt_makeifannouncemsg(ifp, RTM_IEEE80211, what, &info);
1139 if (m == NULL)
1140 return;
1141
1142 /*
1143 * Append the ieee80211 data. Try to stick it in the
1144 * mbuf containing the ifannounce msg; otherwise allocate
1145 * a new mbuf and append.
1146 *
1147 * NB: we assume m is a single mbuf.
1148 */
1149 if (data_len > M_TRAILINGSPACE(m)) {
b11e1686 1150 struct mbuf *n = m_get(MB_DONTWAIT, MT_DATA);
841ab66c
SZ
1151 if (n == NULL) {
1152 m_freem(m);
1153 return;
1154 }
1155 bcopy(data, mtod(n, void *), data_len);
1156 n->m_len = data_len;
1157 m->m_next = n;
1158 } else if (data_len > 0) {
1159 bcopy(data, mtod(m, u_int8_t *) + m->m_len, data_len);
1160 m->m_len += data_len;
1161 }
e9fa4b60 1162 mbuftrackid(m, 33);
841ab66c
SZ
1163 if (m->m_flags & M_PKTHDR)
1164 m->m_pkthdr.len += data_len;
1165 mtod(m, struct if_announcemsghdr *)->ifan_msglen += data_len;
590b8cd4 1166 rts_input(m, 0);
841ab66c
SZ
1167}
1168
1169/*
1170 * This is called to generate routing socket messages indicating
1171 * network interface arrival and departure.
1172 */
1173void
1174rt_ifannouncemsg(struct ifnet *ifp, int what)
1175{
1176 struct rt_addrinfo addrinfo;
1177 struct mbuf *m;
1178
1179 m = rt_makeifannouncemsg(ifp, RTM_IFANNOUNCE, what, &addrinfo);
1180 if (m != NULL)
1181 rts_input(m, 0);
590b8cd4 1182}
f23061d4 1183
590b8cd4
JH
1184static int
1185resizewalkarg(struct walkarg *w, int len)
1186{
1187 void *newptr;
1188
efda3bd0 1189 newptr = kmalloc(len, M_RTABLE, M_INTWAIT | M_NULLOK);
590b8cd4
JH
1190 if (newptr == NULL)
1191 return (ENOMEM);
1192 if (w->w_tmem != NULL)
efda3bd0 1193 kfree(w->w_tmem, M_RTABLE);
590b8cd4
JH
1194 w->w_tmem = newptr;
1195 w->w_tmemsize = len;
1196 return (0);
1197}
984263bc
MD
1198
1199/*
1200 * This is used in dumping the kernel table via sysctl().
1201 */
1202int
590b8cd4 1203sysctl_dumpentry(struct radix_node *rn, void *vw)
984263bc 1204{
82ed7fc2
RG
1205 struct walkarg *w = vw;
1206 struct rtentry *rt = (struct rtentry *)rn;
590b8cd4
JH
1207 struct rt_addrinfo rtinfo;
1208 int error, msglen;
984263bc
MD
1209
1210 if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg))
1211 return 0;
f23061d4 1212
590b8cd4
JH
1213 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1214 rtinfo.rti_dst = rt_key(rt);
1215 rtinfo.rti_gateway = rt->rt_gateway;
1216 rtinfo.rti_netmask = rt_mask(rt);
1217 rtinfo.rti_genmask = rt->rt_genmask;
ef87f48d 1218 if (rt->rt_ifp != NULL) {
590b8cd4 1219 rtinfo.rti_ifpaddr =
b2632176 1220 TAILQ_FIRST(&rt->rt_ifp->if_addrheads[mycpuid])->ifa->ifa_addr;
590b8cd4 1221 rtinfo.rti_ifaaddr = rt->rt_ifa->ifa_addr;
984263bc 1222 if (rt->rt_ifp->if_flags & IFF_POINTOPOINT)
590b8cd4 1223 rtinfo.rti_bcastaddr = rt->rt_ifa->ifa_dstaddr;
984263bc 1224 }
590b8cd4
JH
1225 msglen = rt_msgsize(RTM_GET, &rtinfo);
1226 if (w->w_tmemsize < msglen && resizewalkarg(w, msglen) != 0)
1227 return (ENOMEM);
1228 rt_msg_buffer(RTM_GET, &rtinfo, w->w_tmem, msglen);
1229 if (w->w_req != NULL) {
8e63efec 1230 struct rt_msghdr *rtm = w->w_tmem;
984263bc
MD
1231
1232 rtm->rtm_flags = rt->rt_flags;
1233 rtm->rtm_use = rt->rt_use;
1234 rtm->rtm_rmx = rt->rt_rmx;
1235 rtm->rtm_index = rt->rt_ifp->if_index;
1236 rtm->rtm_errno = rtm->rtm_pid = rtm->rtm_seq = 0;
590b8cd4
JH
1237 rtm->rtm_addrs = rtinfo.rti_addrs;
1238 error = SYSCTL_OUT(w->w_req, rtm, msglen);
984263bc
MD
1239 return (error);
1240 }
590b8cd4 1241 return (0);
984263bc
MD
1242}
1243
590b8cd4
JH
1244static int
1245sysctl_iflist(int af, struct walkarg *w)
984263bc 1246{
82ed7fc2 1247 struct ifnet *ifp;
590b8cd4
JH
1248 struct rt_addrinfo rtinfo;
1249 int msglen, error;
984263bc 1250
590b8cd4 1251 bzero(&rtinfo, sizeof(struct rt_addrinfo));
984263bc 1252 TAILQ_FOREACH(ifp, &ifnet, if_link) {
b2632176
SZ
1253 struct ifaddr_container *ifac;
1254 struct ifaddr *ifa;
1255
984263bc
MD
1256 if (w->w_arg && w->w_arg != ifp->if_index)
1257 continue;
b2632176
SZ
1258 ifac = TAILQ_FIRST(&ifp->if_addrheads[mycpuid]);
1259 ifa = ifac->ifa;
590b8cd4
JH
1260 rtinfo.rti_ifpaddr = ifa->ifa_addr;
1261 msglen = rt_msgsize(RTM_IFINFO, &rtinfo);
1262 if (w->w_tmemsize < msglen && resizewalkarg(w, msglen) != 0)
1263 return (ENOMEM);
1264 rt_msg_buffer(RTM_IFINFO, &rtinfo, w->w_tmem, msglen);
1265 rtinfo.rti_ifpaddr = NULL;
ef87f48d 1266 if (w->w_req != NULL && w->w_tmem != NULL) {
8e63efec 1267 struct if_msghdr *ifm = w->w_tmem;
984263bc 1268
984263bc 1269 ifm->ifm_index = ifp->if_index;
9c095379 1270 ifm->ifm_flags = ifp->if_flags;
984263bc 1271 ifm->ifm_data = ifp->if_data;
590b8cd4
JH
1272 ifm->ifm_addrs = rtinfo.rti_addrs;
1273 error = SYSCTL_OUT(w->w_req, ifm, msglen);
984263bc
MD
1274 if (error)
1275 return (error);
1276 }
b2632176
SZ
1277 while ((ifac = TAILQ_NEXT(ifac, ifa_link)) != NULL) {
1278 ifa = ifac->ifa;
1279
984263bc
MD
1280 if (af && af != ifa->ifa_addr->sa_family)
1281 continue;
590b8cd4 1282 if (curproc->p_ucred->cr_prison &&
87de5057 1283 prison_if(curproc->p_ucred, ifa->ifa_addr))
984263bc 1284 continue;
590b8cd4
JH
1285 rtinfo.rti_ifaaddr = ifa->ifa_addr;
1286 rtinfo.rti_netmask = ifa->ifa_netmask;
1287 rtinfo.rti_bcastaddr = ifa->ifa_dstaddr;
1288 msglen = rt_msgsize(RTM_NEWADDR, &rtinfo);
1289 if (w->w_tmemsize < msglen &&
1290 resizewalkarg(w, msglen) != 0)
1291 return (ENOMEM);
1292 rt_msg_buffer(RTM_NEWADDR, &rtinfo, w->w_tmem, msglen);
1293 if (w->w_req != NULL) {
8e63efec 1294 struct ifa_msghdr *ifam = w->w_tmem;
984263bc 1295
984263bc
MD
1296 ifam->ifam_index = ifa->ifa_ifp->if_index;
1297 ifam->ifam_flags = ifa->ifa_flags;
1298 ifam->ifam_metric = ifa->ifa_metric;
590b8cd4
JH
1299 ifam->ifam_addrs = rtinfo.rti_addrs;
1300 error = SYSCTL_OUT(w->w_req, w->w_tmem, msglen);
984263bc
MD
1301 if (error)
1302 return (error);
1303 }
1304 }
590b8cd4
JH
1305 rtinfo.rti_netmask = NULL;
1306 rtinfo.rti_ifaaddr = NULL;
1307 rtinfo.rti_bcastaddr = NULL;
984263bc
MD
1308 }
1309 return (0);
1310}
1311
1312static int
1313sysctl_rtsock(SYSCTL_HANDLER_ARGS)
1314{
1315 int *name = (int *)arg1;
1316 u_int namelen = arg2;
82ed7fc2 1317 struct radix_node_head *rnh;
a49aa710 1318 int i, error = EINVAL;
ecdefdda 1319 int origcpu;
984263bc
MD
1320 u_char af;
1321 struct walkarg w;
1322
1323 name ++;
1324 namelen--;
1325 if (req->newptr)
1326 return (EPERM);
ecdefdda 1327 if (namelen != 3 && namelen != 4)
984263bc
MD
1328 return (EINVAL);
1329 af = name[0];
5fe66e68 1330 bzero(&w, sizeof w);
984263bc
MD
1331 w.w_op = name[1];
1332 w.w_arg = name[2];
1333 w.w_req = req;
1334
ecdefdda
MD
1335 /*
1336 * Optional third argument specifies cpu, used primarily for
1337 * debugging the route table.
1338 */
1339 if (namelen == 4) {
1340 if (name[3] < 0 || name[3] >= ncpus)
1341 return (EINVAL);
1342 origcpu = mycpuid;
1343 lwkt_migratecpu(name[3]);
1344 } else {
1345 origcpu = -1;
1346 }
a49aa710 1347 crit_enter();
984263bc 1348 switch (w.w_op) {
984263bc
MD
1349 case NET_RT_DUMP:
1350 case NET_RT_FLAGS:
1351 for (i = 1; i <= AF_MAX; i++)
ecdefdda
MD
1352 if ((rnh = rt_tables[mycpuid][i]) &&
1353 (af == 0 || af == i) &&
984263bc 1354 (error = rnh->rnh_walktree(rnh,
f23061d4 1355 sysctl_dumpentry, &w)))
984263bc
MD
1356 break;
1357 break;
1358
1359 case NET_RT_IFLIST:
1360 error = sysctl_iflist(af, &w);
1361 }
a49aa710 1362 crit_exit();
590b8cd4 1363 if (w.w_tmem != NULL)
efda3bd0 1364 kfree(w.w_tmem, M_RTABLE);
ecdefdda
MD
1365 if (origcpu >= 0)
1366 lwkt_migratecpu(origcpu);
984263bc
MD
1367 return (error);
1368}
1369
1370SYSCTL_NODE(_net, PF_ROUTE, routetable, CTLFLAG_RD, sysctl_rtsock, "");
1371
1372/*
1373 * Definitions of protocols supported in the ROUTE domain.
1374 */
1375
ba4257b8 1376static struct domain routedomain; /* or at least forward */
984263bc
MD
1377
1378static struct protosw routesw[] = {
1379{ SOCK_RAW, &routedomain, 0, PR_ATOMIC|PR_ADDR,
1380 0, route_output, raw_ctlinput, 0,
e3873585 1381 cpu0_soport, cpu0_ctlport,
984263bc
MD
1382 raw_init, 0, 0, 0,
1383 &route_usrreqs
1384}
1385};
1386
ba4257b8 1387static struct domain routedomain = {
9c70fe43
JH
1388 PF_ROUTE, "route", NULL, NULL, NULL,
1389 routesw, &routesw[(sizeof routesw)/(sizeof routesw[0])],
1390};
984263bc
MD
1391
1392DOMAIN_SET(route);
47db7f9b 1393