network - Completely revamp the netisr / dispatch code
[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,
8b5c39bb 328 .pru_soreceive = soreceive
984263bc
MD
329};
330
590b8cd4
JH
331static __inline sa_family_t
332familyof(struct sockaddr *sa)
333{
334 return (sa != NULL ? sa->sa_family : 0);
335}
336
47db7f9b
MD
337/*
338 * Routing socket input function. The packet must be serialized onto cpu 0.
339 * We use the cpu0_soport() netisr processing loop to handle it.
340 *
341 * This looks messy but it means that anyone, including interrupt code,
342 * can send a message to the routing socket.
343 */
590b8cd4 344static void
47db7f9b 345rts_input_handler(struct netmsg *msg)
590b8cd4
JH
346{
347 static const struct sockaddr route_dst = { 2, PF_ROUTE, };
47db7f9b
MD
348 struct sockproto route_proto;
349 struct netmsg_packet *pmsg;
350 struct mbuf *m;
351 sa_family_t family;
d16fa201 352 struct rawcb *skip;
47db7f9b
MD
353
354 pmsg = (void *)msg;
47db7f9b
MD
355 family = pmsg->nm_netmsg.nm_lmsg.u.ms_result;
356 route_proto.sp_family = PF_ROUTE;
357 route_proto.sp_protocol = family;
590b8cd4 358
d16fa201
SZ
359 m = pmsg->nm_packet;
360 M_ASSERTPKTHDR(m);
361
362 skip = m->m_pkthdr.header;
363 m->m_pkthdr.header = NULL;
364
365 raw_input(m, &route_proto, &route_src, &route_dst, skip);
590b8cd4
JH
366}
367
47db7f9b 368static void
d16fa201 369rts_input_skip(struct mbuf *m, sa_family_t family, struct rawcb *skip)
47db7f9b
MD
370{
371 struct netmsg_packet *pmsg;
372 lwkt_port_t port;
373
d16fa201
SZ
374 M_ASSERTPKTHDR(m);
375
48e7b118 376 port = cpu0_soport(NULL, NULL, NULL); /* same as for routing socket */
47db7f9b 377 pmsg = &m->m_hdr.mh_netmsg;
48e7b118 378 netmsg_init(&pmsg->nm_netmsg, NULL, &netisr_apanic_rport,
47db7f9b
MD
379 0, rts_input_handler);
380 pmsg->nm_packet = m;
381 pmsg->nm_netmsg.nm_lmsg.u.ms_result = family;
d16fa201 382 m->m_pkthdr.header = skip; /* XXX steal field in pkthdr */
47db7f9b
MD
383 lwkt_sendmsg(port, &pmsg->nm_netmsg.nm_lmsg);
384}
385
d16fa201
SZ
386static __inline void
387rts_input(struct mbuf *m, sa_family_t family)
388{
389 rts_input_skip(m, family, NULL);
390}
391
590b8cd4 392static void *
6e6b3e50 393reallocbuf_nofree(void *ptr, size_t len, size_t olen)
590b8cd4
JH
394{
395 void *newptr;
396
efda3bd0 397 newptr = kmalloc(len, M_RTABLE, M_INTWAIT | M_NULLOK);
590b8cd4
JH
398 if (newptr == NULL)
399 return NULL;
400 bcopy(ptr, newptr, olen);
590b8cd4
JH
401 return (newptr);
402}
403
8e63efec
JH
404/*
405 * Internal helper routine for route_output().
406 */
590b8cd4 407static int
6e6b3e50
SZ
408_fillrtmsg(struct rt_msghdr **prtm, struct rtentry *rt,
409 struct rt_addrinfo *rtinfo)
590b8cd4
JH
410{
411 int msglen;
412 struct rt_msghdr *rtm = *prtm;
413
414 /* Fill in rt_addrinfo for call to rt_msg_buffer(). */
415 rtinfo->rti_dst = rt_key(rt);
416 rtinfo->rti_gateway = rt->rt_gateway;
417 rtinfo->rti_netmask = rt_mask(rt); /* might be NULL */
418 rtinfo->rti_genmask = rt->rt_genmask; /* might be NULL */
419 if (rtm->rtm_addrs & (RTA_IFP | RTA_IFA)) {
420 if (rt->rt_ifp != NULL) {
421 rtinfo->rti_ifpaddr =
b2632176
SZ
422 TAILQ_FIRST(&rt->rt_ifp->if_addrheads[mycpuid])
423 ->ifa->ifa_addr;
590b8cd4
JH
424 rtinfo->rti_ifaaddr = rt->rt_ifa->ifa_addr;
425 if (rt->rt_ifp->if_flags & IFF_POINTOPOINT)
426 rtinfo->rti_bcastaddr = rt->rt_ifa->ifa_dstaddr;
427 rtm->rtm_index = rt->rt_ifp->if_index;
428 } else {
429 rtinfo->rti_ifpaddr = NULL;
430 rtinfo->rti_ifaaddr = NULL;
1730f0e1 431 }
9d1f1679
SZ
432 } else if (rt->rt_ifp != NULL) {
433 rtm->rtm_index = rt->rt_ifp->if_index;
590b8cd4
JH
434 }
435
436 msglen = rt_msgsize(rtm->rtm_type, rtinfo);
437 if (rtm->rtm_msglen < msglen) {
6e6b3e50
SZ
438 /* NOTE: Caller will free the old rtm accordingly */
439 rtm = reallocbuf_nofree(rtm, msglen, rtm->rtm_msglen);
590b8cd4
JH
440 if (rtm == NULL)
441 return (ENOBUFS);
442 *prtm = rtm;
443 }
444 rt_msg_buffer(rtm->rtm_type, rtinfo, rtm, msglen);
445
446 rtm->rtm_flags = rt->rt_flags;
447 rtm->rtm_rmx = rt->rt_rmx;
448 rtm->rtm_addrs = rtinfo->rti_addrs;
449
450 return (0);
451}
452
6e6b3e50
SZ
453struct rtm_arg {
454 struct rt_msghdr *bak_rtm;
455 struct rt_msghdr *new_rtm;
456};
457
458static int
459fillrtmsg(struct rtm_arg *arg, struct rtentry *rt,
460 struct rt_addrinfo *rtinfo)
461{
462 struct rt_msghdr *rtm = arg->new_rtm;
463 int error;
464
465 error = _fillrtmsg(&rtm, rt, rtinfo);
466 if (!error) {
467 if (arg->new_rtm != rtm) {
468 /*
469 * _fillrtmsg() just allocated a new rtm;
470 * if the previously allocated rtm is not
471 * the backing rtm, it should be freed.
472 */
473 if (arg->new_rtm != arg->bak_rtm)
474 kfree(arg->new_rtm, M_RTABLE);
475 arg->new_rtm = rtm;
476 }
477 }
478 return error;
479}
480
ecdefdda
MD
481static void route_output_add_callback(int, int, struct rt_addrinfo *,
482 struct rtentry *, void *);
483static void route_output_delete_callback(int, int, struct rt_addrinfo *,
484 struct rtentry *, void *);
00104f14 485static int route_output_get_callback(int, struct rt_addrinfo *,
b5e01dd3 486 struct rtentry *, void *, int);
2c62ae3b 487static int route_output_change_callback(int, struct rt_addrinfo *,
b5e01dd3 488 struct rtentry *, void *, int);
ae2a5ccd 489static int route_output_lock_callback(int, struct rt_addrinfo *,
b5e01dd3 490 struct rtentry *, void *, int);
ecdefdda 491
984263bc
MD
492/*ARGSUSED*/
493static int
deffea2e 494route_output(struct mbuf *m, struct socket *so, ...)
984263bc 495{
6e6b3e50 496 struct rtm_arg arg;
2e9572df 497 struct rt_msghdr *rtm = NULL;
f23061d4 498 struct rawcb *rp = NULL;
deffea2e 499 struct pr_output_info *oi;
590b8cd4 500 struct rt_addrinfo rtinfo;
a8c218b2 501 sa_family_t family;
ef87f48d 502 int len, error = 0;
deffea2e
JS
503 __va_list ap;
504
1cfcc9b6
SZ
505 M_ASSERTPKTHDR(m);
506
deffea2e
JS
507 __va_start(ap, so);
508 oi = __va_arg(ap, struct pr_output_info *);
509 __va_end(ap);
984263bc 510
a8c218b2
SZ
511 family = familyof(NULL);
512
ef87f48d 513#define gotoerr(e) { error = e; goto flush;}
590b8cd4
JH
514
515 if (m == NULL ||
516 (m->m_len < sizeof(long) &&
517 (m = m_pullup(m, sizeof(long))) == NULL))
984263bc 518 return (ENOBUFS);
984263bc 519 len = m->m_pkthdr.len;
590b8cd4 520 if (len < sizeof(struct rt_msghdr) ||
a8c218b2 521 len != mtod(m, struct rt_msghdr *)->rtm_msglen)
ef87f48d 522 gotoerr(EINVAL);
a8c218b2 523
efda3bd0 524 rtm = kmalloc(len, M_RTABLE, M_INTWAIT | M_NULLOK);
a8c218b2 525 if (rtm == NULL)
ef87f48d 526 gotoerr(ENOBUFS);
a8c218b2 527
984263bc 528 m_copydata(m, 0, len, (caddr_t)rtm);
a8c218b2 529 if (rtm->rtm_version != RTM_VERSION)
ef87f48d 530 gotoerr(EPROTONOSUPPORT);
a8c218b2 531
47e78ae8 532 rtm->rtm_pid = oi->p_pid;
590b8cd4
JH
533 bzero(&rtinfo, sizeof(struct rt_addrinfo));
534 rtinfo.rti_addrs = rtm->rtm_addrs;
a8c218b2 535 if (rt_xaddrs((char *)(rtm + 1), (char *)rtm + len, &rtinfo) != 0)
ef87f48d 536 gotoerr(EINVAL);
a8c218b2 537
590b8cd4
JH
538 rtinfo.rti_flags = rtm->rtm_flags;
539 if (rtinfo.rti_dst == NULL || rtinfo.rti_dst->sa_family >= AF_MAX ||
540 (rtinfo.rti_gateway && rtinfo.rti_gateway->sa_family >= AF_MAX))
ef87f48d
JH
541 gotoerr(EINVAL);
542
a8c218b2
SZ
543 family = familyof(rtinfo.rti_dst);
544
590b8cd4 545 if (rtinfo.rti_genmask != NULL) {
ac0f13c5
SZ
546 error = rtmask_add_global(rtinfo.rti_genmask);
547 if (error)
548 goto flush;
984263bc
MD
549 }
550
551 /*
552 * Verify that the caller has the appropriate privilege; RTM_GET
553 * is the only operation the non-superuser is allowed.
554 */
02650706
SZ
555 if (rtm->rtm_type != RTM_GET &&
556 priv_check_cred(so->so_cred, PRIV_ROOT, 0) != 0)
ef87f48d 557 gotoerr(EPERM);
984263bc
MD
558
559 switch (rtm->rtm_type) {
984263bc 560 case RTM_ADD:
ecdefdda
MD
561 if (rtinfo.rti_gateway == NULL) {
562 error = EINVAL;
563 } else {
564 error = rtrequest1_global(RTM_ADD, &rtinfo,
565 route_output_add_callback, rtm);
984263bc
MD
566 }
567 break;
984263bc 568 case RTM_DELETE:
ecdefdda 569 /*
6e6b3e50
SZ
570 * Backing rtm (bak_rtm) could _not_ be freed during
571 * rtrequest1_global or rtsearch_global, even if the
572 * callback reallocates the rtm due to its size changes,
573 * since rtinfo points to the backing rtm's memory area.
574 * After rtrequest1_global or rtsearch_global returns,
575 * it is safe to free the backing rtm, since rtinfo will
576 * not be used anymore.
577 *
578 * new_rtm will be used to save the new rtm allocated
579 * by rtrequest1_global or rtsearch_global.
ecdefdda 580 */
6e6b3e50
SZ
581 arg.bak_rtm = rtm;
582 arg.new_rtm = rtm;
ecdefdda 583 error = rtrequest1_global(RTM_DELETE, &rtinfo,
6e6b3e50
SZ
584 route_output_delete_callback, &arg);
585 rtm = arg.new_rtm;
586 if (rtm != arg.bak_rtm)
587 kfree(arg.bak_rtm, M_RTABLE);
984263bc 588 break;
984263bc 589 case RTM_GET:
6e6b3e50
SZ
590 /* See the comment in RTM_DELETE */
591 arg.bak_rtm = rtm;
592 arg.new_rtm = rtm;
00104f14 593 error = rtsearch_global(RTM_GET, &rtinfo,
6e6b3e50 594 route_output_get_callback, &arg,
00104f14 595 RTS_NOEXACTMATCH);
6e6b3e50
SZ
596 rtm = arg.new_rtm;
597 if (rtm != arg.bak_rtm)
598 kfree(arg.bak_rtm, M_RTABLE);
ecdefdda
MD
599 break;
600 case RTM_CHANGE:
2c62ae3b
SZ
601 error = rtsearch_global(RTM_CHANGE, &rtinfo,
602 route_output_change_callback, rtm,
603 RTS_EXACTMATCH);
ecdefdda
MD
604 break;
605 case RTM_LOCK:
ae2a5ccd
SZ
606 error = rtsearch_global(RTM_LOCK, &rtinfo,
607 route_output_lock_callback, rtm,
608 RTS_EXACTMATCH);
590b8cd4 609 break;
984263bc 610 default:
ecdefdda
MD
611 error = EOPNOTSUPP;
612 break;
984263bc 613 }
984263bc 614flush:
590b8cd4
JH
615 if (rtm != NULL) {
616 if (error != 0)
984263bc
MD
617 rtm->rtm_errno = error;
618 else
619 rtm->rtm_flags |= RTF_DONE;
620 }
ecdefdda 621
984263bc
MD
622 /*
623 * Check to see if we don't want our own messages.
624 */
ef87f48d 625 if (!(so->so_options & SO_USELOOPBACK)) {
984263bc 626 if (route_cb.any_count <= 1) {
590b8cd4 627 if (rtm != NULL)
efda3bd0 628 kfree(rtm, M_RTABLE);
984263bc
MD
629 m_freem(m);
630 return (error);
631 }
632 /* There is another listener, so construct message */
633 rp = sotorawcb(so);
634 }
590b8cd4 635 if (rtm != NULL) {
984263bc
MD
636 m_copyback(m, 0, rtm->rtm_msglen, (caddr_t)rtm);
637 if (m->m_pkthdr.len < rtm->rtm_msglen) {
638 m_freem(m);
639 m = NULL;
640 } else if (m->m_pkthdr.len > rtm->rtm_msglen)
641 m_adj(m, rtm->rtm_msglen - m->m_pkthdr.len);
efda3bd0 642 kfree(rtm, M_RTABLE);
984263bc 643 }
ef87f48d 644 if (m != NULL)
a8c218b2 645 rts_input_skip(m, family, rp);
984263bc
MD
646 return (error);
647}
648
ecdefdda
MD
649static void
650route_output_add_callback(int cmd, int error, struct rt_addrinfo *rtinfo,
651 struct rtentry *rt, void *arg)
652{
653 struct rt_msghdr *rtm = arg;
654
655 if (error == 0 && rt != NULL) {
656 rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx,
657 &rt->rt_rmx);
658 rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits);
659 rt->rt_rmx.rmx_locks |=
660 (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks);
ac0f13c5
SZ
661 if (rtinfo->rti_genmask != NULL) {
662 rt->rt_genmask = rtmask_purelookup(rtinfo->rti_genmask);
663 if (rt->rt_genmask == NULL) {
664 /*
665 * This should not happen, since we
666 * have already installed genmask
667 * on each CPU before we reach here.
668 */
669 panic("genmask is gone!?");
670 }
671 } else {
672 rt->rt_genmask = NULL;
673 }
31f89d97 674 rtm->rtm_index = rt->rt_ifp->if_index;
ecdefdda
MD
675 }
676}
677
678static void
679route_output_delete_callback(int cmd, int error, struct rt_addrinfo *rtinfo,
680 struct rtentry *rt, void *arg)
681{
ecdefdda
MD
682 if (error == 0 && rt) {
683 ++rt->rt_refcnt;
6e6b3e50 684 if (fillrtmsg(arg, rt, rtinfo) != 0) {
ecdefdda
MD
685 error = ENOBUFS;
686 /* XXX no way to return the error */
687 }
688 --rt->rt_refcnt;
689 }
6e6b3e50
SZ
690 if (rt && rt->rt_refcnt == 0) {
691 ++rt->rt_refcnt;
692 rtfree(rt);
693 }
ecdefdda
MD
694}
695
00104f14
SZ
696static int
697route_output_get_callback(int cmd, struct rt_addrinfo *rtinfo,
b5e01dd3 698 struct rtentry *rt, void *arg, int found_cnt)
00104f14 699{
00104f14
SZ
700 int error, found = 0;
701
702 if (((rtinfo->rti_flags ^ rt->rt_flags) & RTF_HOST) == 0)
703 found = 1;
704
6e6b3e50 705 error = fillrtmsg(arg, rt, rtinfo);
00104f14
SZ
706 if (!error && found) {
707 /* Got the exact match, we could return now! */
708 error = EJUSTRETURN;
709 }
710 return error;
711}
712
2c62ae3b
SZ
713static int
714route_output_change_callback(int cmd, struct rt_addrinfo *rtinfo,
b5e01dd3 715 struct rtentry *rt, void *arg, int found_cnt)
ecdefdda
MD
716{
717 struct rt_msghdr *rtm = arg;
718 struct ifaddr *ifa;
2c62ae3b 719 int error = 0;
ecdefdda
MD
720
721 /*
722 * new gateway could require new ifaddr, ifp;
723 * flags may also be different; ifp may be specified
724 * by ll sockaddr when protocol address is ambiguous
725 */
726 if (((rt->rt_flags & RTF_GATEWAY) && rtinfo->rti_gateway != NULL) ||
fb41af81
SZ
727 rtinfo->rti_ifpaddr != NULL ||
728 (rtinfo->rti_ifaaddr != NULL &&
729 !sa_equal(rtinfo->rti_ifaaddr, rt->rt_ifa->ifa_addr))) {
ecdefdda
MD
730 error = rt_getifa(rtinfo);
731 if (error != 0)
732 goto done;
733 }
734 if (rtinfo->rti_gateway != NULL) {
b5e01dd3
SZ
735 /*
736 * We only need to generate rtmsg upon the
737 * first route to be changed.
738 */
afe44b68 739 error = rt_setgate(rt, rt_key(rt), rtinfo->rti_gateway,
b5e01dd3 740 found_cnt == 1 ? RTL_REPORTMSG : RTL_DONTREPORT);
ecdefdda
MD
741 if (error != 0)
742 goto done;
743 }
744 if ((ifa = rtinfo->rti_ifa) != NULL) {
745 struct ifaddr *oifa = rt->rt_ifa;
746
747 if (oifa != ifa) {
748 if (oifa && oifa->ifa_rtrequest)
749 oifa->ifa_rtrequest(RTM_DELETE, rt, rtinfo);
750 IFAFREE(rt->rt_ifa);
751 IFAREF(ifa);
752 rt->rt_ifa = ifa;
753 rt->rt_ifp = rtinfo->rti_ifp;
754 }
755 }
756 rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx, &rt->rt_rmx);
757 if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest)
2c62ae3b 758 rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, rtinfo);
ac0f13c5
SZ
759 if (rtinfo->rti_genmask != NULL) {
760 rt->rt_genmask = rtmask_purelookup(rtinfo->rti_genmask);
761 if (rt->rt_genmask == NULL) {
762 /*
763 * This should not happen, since we
764 * have already installed genmask
765 * on each CPU before we reach here.
766 */
767 panic("genmask is gone!?\n");
768 }
769 }
31f89d97 770 rtm->rtm_index = rt->rt_ifp->if_index;
ecdefdda 771done:
2c62ae3b 772 return error;
ecdefdda
MD
773}
774
ae2a5ccd
SZ
775static int
776route_output_lock_callback(int cmd, struct rt_addrinfo *rtinfo,
b5e01dd3
SZ
777 struct rtentry *rt, void *arg,
778 int found_cnt __unused)
ecdefdda
MD
779{
780 struct rt_msghdr *rtm = arg;
781
782 rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits);
783 rt->rt_rmx.rmx_locks |=
784 (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks);
ae2a5ccd 785 return 0;
ecdefdda
MD
786}
787
984263bc 788static void
ef87f48d 789rt_setmetrics(u_long which, struct rt_metrics *in, struct rt_metrics *out)
984263bc 790{
2e9572df
JH
791#define setmetric(flag, elt) if (which & (flag)) out->elt = in->elt;
792 setmetric(RTV_RPIPE, rmx_recvpipe);
793 setmetric(RTV_SPIPE, rmx_sendpipe);
794 setmetric(RTV_SSTHRESH, rmx_ssthresh);
795 setmetric(RTV_RTT, rmx_rtt);
796 setmetric(RTV_RTTVAR, rmx_rttvar);
797 setmetric(RTV_HOPCOUNT, rmx_hopcount);
798 setmetric(RTV_MTU, rmx_mtu);
799 setmetric(RTV_EXPIRE, rmx_expire);
800#undef setmetric
984263bc
MD
801}
802
803#define ROUNDUP(a) \
804 ((a) > 0 ? (1 + (((a) - 1) | (sizeof(long) - 1))) : sizeof(long))
984263bc 805
984263bc
MD
806/*
807 * Extract the addresses of the passed sockaddrs.
808 * Do a little sanity checking so as to avoid bad memory references.
809 * This data is derived straight from userland.
810 */
811static int
ef87f48d 812rt_xaddrs(char *cp, char *cplim, struct rt_addrinfo *rtinfo)
984263bc 813{
82ed7fc2
RG
814 struct sockaddr *sa;
815 int i;
984263bc
MD
816
817 for (i = 0; (i < RTAX_MAX) && (cp < cplim); i++) {
818 if ((rtinfo->rti_addrs & (1 << i)) == 0)
819 continue;
820 sa = (struct sockaddr *)cp;
821 /*
822 * It won't fit.
823 */
0c3c561c 824 if ((cp + sa->sa_len) > cplim) {
984263bc
MD
825 return (EINVAL);
826 }
827
828 /*
ef87f48d 829 * There are no more... Quit now.
984263bc 830 * If there are more bits, they are in error.
ef87f48d 831 * I've seen this. route(1) can evidently generate these.
984263bc 832 * This causes kernel to core dump.
ef87f48d 833 * For compatibility, if we see this, point to a safe address.
984263bc
MD
834 */
835 if (sa->sa_len == 0) {
590b8cd4
JH
836 static struct sockaddr sa_zero = {
837 sizeof sa_zero, AF_INET,
838 };
839
984263bc 840 rtinfo->rti_info[i] = &sa_zero;
9b42cabe 841 kprintf("rtsock: received more addr bits than sockaddrs.\n");
984263bc
MD
842 return (0); /* should be EINVAL but for compat */
843 }
844
ef87f48d 845 /* Accept the sockaddr. */
984263bc 846 rtinfo->rti_info[i] = sa;
590b8cd4 847 cp += ROUNDUP(sa->sa_len);
984263bc
MD
848 }
849 return (0);
850}
851
590b8cd4
JH
852static int
853rt_msghdrsize(int type)
984263bc 854{
984263bc 855 switch (type) {
984263bc
MD
856 case RTM_DELADDR:
857 case RTM_NEWADDR:
590b8cd4 858 return sizeof(struct ifa_msghdr);
984263bc
MD
859 case RTM_DELMADDR:
860 case RTM_NEWMADDR:
590b8cd4 861 return sizeof(struct ifma_msghdr);
984263bc 862 case RTM_IFINFO:
590b8cd4 863 return sizeof(struct if_msghdr);
984263bc 864 case RTM_IFANNOUNCE:
841ab66c 865 case RTM_IEEE80211:
590b8cd4 866 return sizeof(struct if_announcemsghdr);
984263bc 867 default:
590b8cd4 868 return sizeof(struct rt_msghdr);
984263bc 869 }
590b8cd4
JH
870}
871
872static int
873rt_msgsize(int type, struct rt_addrinfo *rtinfo)
874{
875 int len, i;
876
877 len = rt_msghdrsize(type);
878 for (i = 0; i < RTAX_MAX; i++) {
879 if (rtinfo->rti_info[i] != NULL)
880 len += ROUNDUP(rtinfo->rti_info[i]->sa_len);
984263bc 881 }
590b8cd4
JH
882 len = ALIGN(len);
883 return len;
884}
885
886/*
887 * Build a routing message in a buffer.
888 * Copy the addresses in the rtinfo->rti_info[] sockaddr array
889 * to the end of the buffer after the message header.
890 *
891 * Set the rtinfo->rti_addrs bitmask of addresses present in rtinfo->rti_info[].
892 * This side-effect can be avoided if we reorder the addrs bitmask field in all
893 * the route messages to line up so we can set it here instead of back in the
894 * calling routine.
895 */
896static void
897rt_msg_buffer(int type, struct rt_addrinfo *rtinfo, void *buf, int msglen)
898{
899 struct rt_msghdr *rtm;
900 char *cp;
901 int dlen, i;
902
903 rtm = (struct rt_msghdr *) buf;
904 rtm->rtm_version = RTM_VERSION;
905 rtm->rtm_type = type;
906 rtm->rtm_msglen = msglen;
907
908 cp = (char *)buf + rt_msghdrsize(type);
909 rtinfo->rti_addrs = 0;
984263bc 910 for (i = 0; i < RTAX_MAX; i++) {
590b8cd4
JH
911 struct sockaddr *sa;
912
984263bc
MD
913 if ((sa = rtinfo->rti_info[i]) == NULL)
914 continue;
915 rtinfo->rti_addrs |= (1 << i);
916 dlen = ROUNDUP(sa->sa_len);
590b8cd4
JH
917 bcopy(sa, cp, dlen);
918 cp += dlen;
984263bc 919 }
984263bc
MD
920}
921
590b8cd4
JH
922/*
923 * Build a routing message in a mbuf chain.
924 * Copy the addresses in the rtinfo->rti_info[] sockaddr array
925 * to the end of the mbuf after the message header.
926 *
927 * Set the rtinfo->rti_addrs bitmask of addresses present in rtinfo->rti_info[].
928 * This side-effect can be avoided if we reorder the addrs bitmask field in all
929 * the route messages to line up so we can set it here instead of back in the
930 * calling routine.
931 */
932static struct mbuf *
933rt_msg_mbuf(int type, struct rt_addrinfo *rtinfo)
984263bc 934{
590b8cd4
JH
935 struct mbuf *m;
936 struct rt_msghdr *rtm;
937 int hlen, len;
82ed7fc2 938 int i;
984263bc 939
590b8cd4
JH
940 hlen = rt_msghdrsize(type);
941 KASSERT(hlen <= MCLBYTES, ("rt_msg_mbuf: hlen %d doesn't fit", hlen));
984263bc 942
50503f0f 943 m = m_getl(hlen, MB_DONTWAIT, MT_DATA, M_PKTHDR, NULL);
590b8cd4
JH
944 if (m == NULL)
945 return (NULL);
e9fa4b60 946 mbuftrackid(m, 32);
590b8cd4
JH
947 m->m_pkthdr.len = m->m_len = hlen;
948 m->m_pkthdr.rcvif = NULL;
949 rtinfo->rti_addrs = 0;
950 len = hlen;
984263bc 951 for (i = 0; i < RTAX_MAX; i++) {
82ed7fc2 952 struct sockaddr *sa;
590b8cd4 953 int dlen;
984263bc 954
ef87f48d 955 if ((sa = rtinfo->rti_info[i]) == NULL)
984263bc
MD
956 continue;
957 rtinfo->rti_addrs |= (1 << i);
958 dlen = ROUNDUP(sa->sa_len);
590b8cd4 959 m_copyback(m, len, dlen, (caddr_t)sa); /* can grow mbuf chain */
984263bc
MD
960 len += dlen;
961 }
590b8cd4
JH
962 if (m->m_pkthdr.len != len) { /* one of the m_copyback() calls failed */
963 m_freem(m);
964 return (NULL);
984263bc 965 }
590b8cd4
JH
966 rtm = mtod(m, struct rt_msghdr *);
967 bzero(rtm, hlen);
968 rtm->rtm_msglen = len;
969 rtm->rtm_version = RTM_VERSION;
970 rtm->rtm_type = type;
971 return (m);
984263bc
MD
972}
973
974/*
975 * This routine is called to generate a message from the routing
f23061d4 976 * socket indicating that a redirect has occurred, a routing lookup
984263bc
MD
977 * has failed, or that a protocol has detected timeouts to a particular
978 * destination.
979 */
980void
f23061d4 981rt_missmsg(int type, struct rt_addrinfo *rtinfo, int flags, int error)
984263bc 982{
0c3c561c 983 struct sockaddr *dst = rtinfo->rti_info[RTAX_DST];
82ed7fc2
RG
984 struct rt_msghdr *rtm;
985 struct mbuf *m;
984263bc
MD
986
987 if (route_cb.any_count == 0)
988 return;
590b8cd4 989 m = rt_msg_mbuf(type, rtinfo);
ef87f48d 990 if (m == NULL)
984263bc
MD
991 return;
992 rtm = mtod(m, struct rt_msghdr *);
993 rtm->rtm_flags = RTF_DONE | flags;
994 rtm->rtm_errno = error;
995 rtm->rtm_addrs = rtinfo->rti_addrs;
590b8cd4 996 rts_input(m, familyof(dst));
984263bc
MD
997}
998
f3ed2586
JH
999void
1000rt_dstmsg(int type, struct sockaddr *dst, int error)
1001{
1002 struct rt_msghdr *rtm;
1003 struct rt_addrinfo addrs;
1004 struct mbuf *m;
1005
1006 if (route_cb.any_count == 0)
1007 return;
1008 bzero(&addrs, sizeof(struct rt_addrinfo));
1009 addrs.rti_info[RTAX_DST] = dst;
590b8cd4 1010 m = rt_msg_mbuf(type, &addrs);
f3ed2586
JH
1011 if (m == NULL)
1012 return;
1013 rtm = mtod(m, struct rt_msghdr *);
1014 rtm->rtm_flags = RTF_DONE;
1015 rtm->rtm_errno = error;
1016 rtm->rtm_addrs = addrs.rti_addrs;
590b8cd4 1017 rts_input(m, familyof(dst));
f3ed2586
JH
1018}
1019
984263bc
MD
1020/*
1021 * This routine is called to generate a message from the routing
1022 * socket indicating that the status of a network interface has changed.
1023 */
1024void
0c3c561c 1025rt_ifmsg(struct ifnet *ifp)
984263bc 1026{
82ed7fc2 1027 struct if_msghdr *ifm;
984263bc 1028 struct mbuf *m;
590b8cd4 1029 struct rt_addrinfo rtinfo;
984263bc
MD
1030
1031 if (route_cb.any_count == 0)
1032 return;
590b8cd4
JH
1033 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1034 m = rt_msg_mbuf(RTM_IFINFO, &rtinfo);
ef87f48d 1035 if (m == NULL)
984263bc
MD
1036 return;
1037 ifm = mtod(m, struct if_msghdr *);
1038 ifm->ifm_index = ifp->if_index;
9c095379 1039 ifm->ifm_flags = ifp->if_flags;
984263bc 1040 ifm->ifm_data = ifp->if_data;
590b8cd4
JH
1041 ifm->ifm_addrs = 0;
1042 rts_input(m, 0);
984263bc
MD
1043}
1044
372316d9
JH
1045static void
1046rt_ifamsg(int cmd, struct ifaddr *ifa)
1047{
1048 struct ifa_msghdr *ifam;
590b8cd4 1049 struct rt_addrinfo rtinfo;
372316d9 1050 struct mbuf *m;
372316d9
JH
1051 struct ifnet *ifp = ifa->ifa_ifp;
1052
590b8cd4
JH
1053 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1054 rtinfo.rti_ifaaddr = ifa->ifa_addr;
b2632176
SZ
1055 rtinfo.rti_ifpaddr =
1056 TAILQ_FIRST(&ifp->if_addrheads[mycpuid])->ifa->ifa_addr;
590b8cd4
JH
1057 rtinfo.rti_netmask = ifa->ifa_netmask;
1058 rtinfo.rti_bcastaddr = ifa->ifa_dstaddr;
f23061d4 1059
590b8cd4 1060 m = rt_msg_mbuf(cmd, &rtinfo);
f23061d4 1061 if (m == NULL)
372316d9 1062 return;
f23061d4 1063
372316d9
JH
1064 ifam = mtod(m, struct ifa_msghdr *);
1065 ifam->ifam_index = ifp->if_index;
1066 ifam->ifam_metric = ifa->ifa_metric;
1067 ifam->ifam_flags = ifa->ifa_flags;
590b8cd4 1068 ifam->ifam_addrs = rtinfo.rti_addrs;
f23061d4 1069
590b8cd4 1070 rts_input(m, familyof(ifa->ifa_addr));
372316d9
JH
1071}
1072
f3ed2586
JH
1073void
1074rt_rtmsg(int cmd, struct rtentry *rt, struct ifnet *ifp, int error)
372316d9
JH
1075{
1076 struct rt_msghdr *rtm;
590b8cd4 1077 struct rt_addrinfo rtinfo;
372316d9 1078 struct mbuf *m;
f3ed2586 1079 struct sockaddr *dst;
372316d9
JH
1080
1081 if (rt == NULL)
1082 return;
f23061d4 1083
590b8cd4
JH
1084 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1085 rtinfo.rti_dst = dst = rt_key(rt);
1086 rtinfo.rti_gateway = rt->rt_gateway;
1087 rtinfo.rti_netmask = rt_mask(rt);
b2632176
SZ
1088 if (ifp != NULL) {
1089 rtinfo.rti_ifpaddr =
1090 TAILQ_FIRST(&ifp->if_addrheads[mycpuid])->ifa->ifa_addr;
1091 }
590b8cd4 1092 rtinfo.rti_ifaaddr = rt->rt_ifa->ifa_addr;
f23061d4 1093
590b8cd4 1094 m = rt_msg_mbuf(cmd, &rtinfo);
f23061d4 1095 if (m == NULL)
372316d9 1096 return;
f23061d4 1097
372316d9 1098 rtm = mtod(m, struct rt_msghdr *);
f3ed2586
JH
1099 if (ifp != NULL)
1100 rtm->rtm_index = ifp->if_index;
372316d9
JH
1101 rtm->rtm_flags |= rt->rt_flags;
1102 rtm->rtm_errno = error;
590b8cd4 1103 rtm->rtm_addrs = rtinfo.rti_addrs;
f23061d4 1104
590b8cd4 1105 rts_input(m, familyof(dst));
372316d9
JH
1106}
1107
984263bc
MD
1108/*
1109 * This is called to generate messages from the routing socket
1110 * indicating a network interface has had addresses associated with it.
1111 * if we ever reverse the logic and replace messages TO the routing
1112 * socket indicate a request to configure interfaces, then it will
1113 * be unnecessary as the routing socket will automatically generate
1114 * copies of it.
1115 */
1116void
0c3c561c 1117rt_newaddrmsg(int cmd, struct ifaddr *ifa, int error, struct rtentry *rt)
984263bc 1118{
78812139
EN
1119#ifdef SCTP
1120 /*
1121 * notify the SCTP stack
1122 * this will only get called when an address is added/deleted
1123 * XXX pass the ifaddr struct instead if ifa->ifa_addr...
1124 */
1125 if (cmd == RTM_ADD)
1126 sctp_add_ip_address(ifa);
1127 else if (cmd == RTM_DELETE)
1128 sctp_delete_ip_address(ifa);
1129#endif /* SCTP */
1130
984263bc
MD
1131 if (route_cb.any_count == 0)
1132 return;
984263bc 1133
372316d9
JH
1134 if (cmd == RTM_ADD) {
1135 rt_ifamsg(RTM_NEWADDR, ifa);
f3ed2586 1136 rt_rtmsg(RTM_ADD, rt, ifa->ifa_ifp, error);
372316d9
JH
1137 } else {
1138 KASSERT((cmd == RTM_DELETE), ("unknown cmd %d", cmd));
f3ed2586 1139 rt_rtmsg(RTM_DELETE, rt, ifa->ifa_ifp, error);
372316d9 1140 rt_ifamsg(RTM_DELADDR, ifa);
984263bc
MD
1141 }
1142}
1143
1144/*
1145 * This is the analogue to the rt_newaddrmsg which performs the same
1146 * function but for multicast group memberhips. This is easier since
1147 * there is no route state to worry about.
1148 */
1149void
0c3c561c 1150rt_newmaddrmsg(int cmd, struct ifmultiaddr *ifma)
984263bc 1151{
590b8cd4 1152 struct rt_addrinfo rtinfo;
ef87f48d 1153 struct mbuf *m = NULL;
984263bc
MD
1154 struct ifnet *ifp = ifma->ifma_ifp;
1155 struct ifma_msghdr *ifmam;
1156
1157 if (route_cb.any_count == 0)
1158 return;
1159
590b8cd4
JH
1160 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1161 rtinfo.rti_ifaaddr = ifma->ifma_addr;
b2632176
SZ
1162 if (ifp != NULL && !TAILQ_EMPTY(&ifp->if_addrheads[mycpuid])) {
1163 rtinfo.rti_ifpaddr =
1164 TAILQ_FIRST(&ifp->if_addrheads[mycpuid])->ifa->ifa_addr;
1165 }
984263bc
MD
1166 /*
1167 * If a link-layer address is present, present it as a ``gateway''
1168 * (similarly to how ARP entries, e.g., are presented).
1169 */
590b8cd4 1170 rtinfo.rti_gateway = ifma->ifma_lladdr;
f23061d4 1171
590b8cd4 1172 m = rt_msg_mbuf(cmd, &rtinfo);
f23061d4 1173 if (m == NULL)
984263bc 1174 return;
f23061d4 1175
984263bc
MD
1176 ifmam = mtod(m, struct ifma_msghdr *);
1177 ifmam->ifmam_index = ifp->if_index;
590b8cd4 1178 ifmam->ifmam_addrs = rtinfo.rti_addrs;
f23061d4 1179
590b8cd4 1180 rts_input(m, familyof(ifma->ifma_addr));
984263bc
MD
1181}
1182
841ab66c
SZ
1183static struct mbuf *
1184rt_makeifannouncemsg(struct ifnet *ifp, int type, int what,
1185 struct rt_addrinfo *info)
984263bc 1186{
590b8cd4 1187 struct if_announcemsghdr *ifan;
841ab66c 1188 struct mbuf *m;
984263bc
MD
1189
1190 if (route_cb.any_count == 0)
841ab66c 1191 return NULL;
f23061d4 1192
841ab66c
SZ
1193 bzero(info, sizeof(*info));
1194 m = rt_msg_mbuf(type, info);
984263bc 1195 if (m == NULL)
841ab66c 1196 return NULL;
f23061d4 1197
984263bc
MD
1198 ifan = mtod(m, struct if_announcemsghdr *);
1199 ifan->ifan_index = ifp->if_index;
5fe66e68 1200 strlcpy(ifan->ifan_name, ifp->if_xname, sizeof ifan->ifan_name);
984263bc 1201 ifan->ifan_what = what;
841ab66c
SZ
1202 return m;
1203}
1204
1205/*
1206 * This is called to generate routing socket messages indicating
1207 * IEEE80211 wireless events.
1208 * XXX we piggyback on the RTM_IFANNOUNCE msg format in a clumsy way.
1209 */
1210void
1211rt_ieee80211msg(struct ifnet *ifp, int what, void *data, size_t data_len)
1212{
1213 struct rt_addrinfo info;
1214 struct mbuf *m;
f23061d4 1215
841ab66c
SZ
1216 m = rt_makeifannouncemsg(ifp, RTM_IEEE80211, what, &info);
1217 if (m == NULL)
1218 return;
1219
1220 /*
1221 * Append the ieee80211 data. Try to stick it in the
1222 * mbuf containing the ifannounce msg; otherwise allocate
1223 * a new mbuf and append.
1224 *
1225 * NB: we assume m is a single mbuf.
1226 */
1227 if (data_len > M_TRAILINGSPACE(m)) {
b11e1686 1228 struct mbuf *n = m_get(MB_DONTWAIT, MT_DATA);
841ab66c
SZ
1229 if (n == NULL) {
1230 m_freem(m);
1231 return;
1232 }
1233 bcopy(data, mtod(n, void *), data_len);
1234 n->m_len = data_len;
1235 m->m_next = n;
1236 } else if (data_len > 0) {
1237 bcopy(data, mtod(m, u_int8_t *) + m->m_len, data_len);
1238 m->m_len += data_len;
1239 }
e9fa4b60 1240 mbuftrackid(m, 33);
841ab66c
SZ
1241 if (m->m_flags & M_PKTHDR)
1242 m->m_pkthdr.len += data_len;
1243 mtod(m, struct if_announcemsghdr *)->ifan_msglen += data_len;
590b8cd4 1244 rts_input(m, 0);
841ab66c
SZ
1245}
1246
1247/*
1248 * This is called to generate routing socket messages indicating
1249 * network interface arrival and departure.
1250 */
1251void
1252rt_ifannouncemsg(struct ifnet *ifp, int what)
1253{
1254 struct rt_addrinfo addrinfo;
1255 struct mbuf *m;
1256
1257 m = rt_makeifannouncemsg(ifp, RTM_IFANNOUNCE, what, &addrinfo);
1258 if (m != NULL)
1259 rts_input(m, 0);
590b8cd4 1260}
f23061d4 1261
590b8cd4
JH
1262static int
1263resizewalkarg(struct walkarg *w, int len)
1264{
1265 void *newptr;
1266
efda3bd0 1267 newptr = kmalloc(len, M_RTABLE, M_INTWAIT | M_NULLOK);
590b8cd4
JH
1268 if (newptr == NULL)
1269 return (ENOMEM);
1270 if (w->w_tmem != NULL)
efda3bd0 1271 kfree(w->w_tmem, M_RTABLE);
590b8cd4
JH
1272 w->w_tmem = newptr;
1273 w->w_tmemsize = len;
1274 return (0);
1275}
984263bc
MD
1276
1277/*
1278 * This is used in dumping the kernel table via sysctl().
1279 */
1280int
590b8cd4 1281sysctl_dumpentry(struct radix_node *rn, void *vw)
984263bc 1282{
82ed7fc2
RG
1283 struct walkarg *w = vw;
1284 struct rtentry *rt = (struct rtentry *)rn;
590b8cd4
JH
1285 struct rt_addrinfo rtinfo;
1286 int error, msglen;
984263bc
MD
1287
1288 if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg))
1289 return 0;
f23061d4 1290
590b8cd4
JH
1291 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1292 rtinfo.rti_dst = rt_key(rt);
1293 rtinfo.rti_gateway = rt->rt_gateway;
1294 rtinfo.rti_netmask = rt_mask(rt);
1295 rtinfo.rti_genmask = rt->rt_genmask;
ef87f48d 1296 if (rt->rt_ifp != NULL) {
590b8cd4 1297 rtinfo.rti_ifpaddr =
b2632176 1298 TAILQ_FIRST(&rt->rt_ifp->if_addrheads[mycpuid])->ifa->ifa_addr;
590b8cd4 1299 rtinfo.rti_ifaaddr = rt->rt_ifa->ifa_addr;
984263bc 1300 if (rt->rt_ifp->if_flags & IFF_POINTOPOINT)
590b8cd4 1301 rtinfo.rti_bcastaddr = rt->rt_ifa->ifa_dstaddr;
984263bc 1302 }
590b8cd4
JH
1303 msglen = rt_msgsize(RTM_GET, &rtinfo);
1304 if (w->w_tmemsize < msglen && resizewalkarg(w, msglen) != 0)
1305 return (ENOMEM);
1306 rt_msg_buffer(RTM_GET, &rtinfo, w->w_tmem, msglen);
1307 if (w->w_req != NULL) {
8e63efec 1308 struct rt_msghdr *rtm = w->w_tmem;
984263bc
MD
1309
1310 rtm->rtm_flags = rt->rt_flags;
1311 rtm->rtm_use = rt->rt_use;
1312 rtm->rtm_rmx = rt->rt_rmx;
1313 rtm->rtm_index = rt->rt_ifp->if_index;
1314 rtm->rtm_errno = rtm->rtm_pid = rtm->rtm_seq = 0;
590b8cd4
JH
1315 rtm->rtm_addrs = rtinfo.rti_addrs;
1316 error = SYSCTL_OUT(w->w_req, rtm, msglen);
984263bc
MD
1317 return (error);
1318 }
590b8cd4 1319 return (0);
984263bc
MD
1320}
1321
590b8cd4
JH
1322static int
1323sysctl_iflist(int af, struct walkarg *w)
984263bc 1324{
82ed7fc2 1325 struct ifnet *ifp;
590b8cd4
JH
1326 struct rt_addrinfo rtinfo;
1327 int msglen, error;
984263bc 1328
590b8cd4 1329 bzero(&rtinfo, sizeof(struct rt_addrinfo));
984263bc 1330 TAILQ_FOREACH(ifp, &ifnet, if_link) {
b2632176
SZ
1331 struct ifaddr_container *ifac;
1332 struct ifaddr *ifa;
1333
984263bc
MD
1334 if (w->w_arg && w->w_arg != ifp->if_index)
1335 continue;
b2632176
SZ
1336 ifac = TAILQ_FIRST(&ifp->if_addrheads[mycpuid]);
1337 ifa = ifac->ifa;
590b8cd4
JH
1338 rtinfo.rti_ifpaddr = ifa->ifa_addr;
1339 msglen = rt_msgsize(RTM_IFINFO, &rtinfo);
1340 if (w->w_tmemsize < msglen && resizewalkarg(w, msglen) != 0)
1341 return (ENOMEM);
1342 rt_msg_buffer(RTM_IFINFO, &rtinfo, w->w_tmem, msglen);
1343 rtinfo.rti_ifpaddr = NULL;
ef87f48d 1344 if (w->w_req != NULL && w->w_tmem != NULL) {
8e63efec 1345 struct if_msghdr *ifm = w->w_tmem;
984263bc 1346
984263bc 1347 ifm->ifm_index = ifp->if_index;
9c095379 1348 ifm->ifm_flags = ifp->if_flags;
984263bc 1349 ifm->ifm_data = ifp->if_data;
590b8cd4
JH
1350 ifm->ifm_addrs = rtinfo.rti_addrs;
1351 error = SYSCTL_OUT(w->w_req, ifm, msglen);
984263bc
MD
1352 if (error)
1353 return (error);
1354 }
b2632176
SZ
1355 while ((ifac = TAILQ_NEXT(ifac, ifa_link)) != NULL) {
1356 ifa = ifac->ifa;
1357
984263bc
MD
1358 if (af && af != ifa->ifa_addr->sa_family)
1359 continue;
590b8cd4 1360 if (curproc->p_ucred->cr_prison &&
87de5057 1361 prison_if(curproc->p_ucred, ifa->ifa_addr))
984263bc 1362 continue;
590b8cd4
JH
1363 rtinfo.rti_ifaaddr = ifa->ifa_addr;
1364 rtinfo.rti_netmask = ifa->ifa_netmask;
1365 rtinfo.rti_bcastaddr = ifa->ifa_dstaddr;
1366 msglen = rt_msgsize(RTM_NEWADDR, &rtinfo);
1367 if (w->w_tmemsize < msglen &&
1368 resizewalkarg(w, msglen) != 0)
1369 return (ENOMEM);
1370 rt_msg_buffer(RTM_NEWADDR, &rtinfo, w->w_tmem, msglen);
1371 if (w->w_req != NULL) {
8e63efec 1372 struct ifa_msghdr *ifam = w->w_tmem;
984263bc 1373
984263bc
MD
1374 ifam->ifam_index = ifa->ifa_ifp->if_index;
1375 ifam->ifam_flags = ifa->ifa_flags;
1376 ifam->ifam_metric = ifa->ifa_metric;
590b8cd4
JH
1377 ifam->ifam_addrs = rtinfo.rti_addrs;
1378 error = SYSCTL_OUT(w->w_req, w->w_tmem, msglen);
984263bc
MD
1379 if (error)
1380 return (error);
1381 }
1382 }
590b8cd4
JH
1383 rtinfo.rti_netmask = NULL;
1384 rtinfo.rti_ifaaddr = NULL;
1385 rtinfo.rti_bcastaddr = NULL;
984263bc
MD
1386 }
1387 return (0);
1388}
1389
1390static int
1391sysctl_rtsock(SYSCTL_HANDLER_ARGS)
1392{
1393 int *name = (int *)arg1;
1394 u_int namelen = arg2;
82ed7fc2 1395 struct radix_node_head *rnh;
a49aa710 1396 int i, error = EINVAL;
ecdefdda 1397 int origcpu;
984263bc
MD
1398 u_char af;
1399 struct walkarg w;
1400
1401 name ++;
1402 namelen--;
1403 if (req->newptr)
1404 return (EPERM);
ecdefdda 1405 if (namelen != 3 && namelen != 4)
984263bc
MD
1406 return (EINVAL);
1407 af = name[0];
5fe66e68 1408 bzero(&w, sizeof w);
984263bc
MD
1409 w.w_op = name[1];
1410 w.w_arg = name[2];
1411 w.w_req = req;
1412
ecdefdda
MD
1413 /*
1414 * Optional third argument specifies cpu, used primarily for
1415 * debugging the route table.
1416 */
1417 if (namelen == 4) {
1418 if (name[3] < 0 || name[3] >= ncpus)
1419 return (EINVAL);
1420 origcpu = mycpuid;
1421 lwkt_migratecpu(name[3]);
1422 } else {
1423 origcpu = -1;
1424 }
a49aa710 1425 crit_enter();
984263bc 1426 switch (w.w_op) {
984263bc
MD
1427 case NET_RT_DUMP:
1428 case NET_RT_FLAGS:
1429 for (i = 1; i <= AF_MAX; i++)
ecdefdda
MD
1430 if ((rnh = rt_tables[mycpuid][i]) &&
1431 (af == 0 || af == i) &&
984263bc 1432 (error = rnh->rnh_walktree(rnh,
f23061d4 1433 sysctl_dumpentry, &w)))
984263bc
MD
1434 break;
1435 break;
1436
1437 case NET_RT_IFLIST:
1438 error = sysctl_iflist(af, &w);
1439 }
a49aa710 1440 crit_exit();
590b8cd4 1441 if (w.w_tmem != NULL)
efda3bd0 1442 kfree(w.w_tmem, M_RTABLE);
ecdefdda
MD
1443 if (origcpu >= 0)
1444 lwkt_migratecpu(origcpu);
984263bc
MD
1445 return (error);
1446}
1447
1448SYSCTL_NODE(_net, PF_ROUTE, routetable, CTLFLAG_RD, sysctl_rtsock, "");
1449
1450/*
1451 * Definitions of protocols supported in the ROUTE domain.
1452 */
1453
ba4257b8 1454static struct domain routedomain; /* or at least forward */
984263bc
MD
1455
1456static struct protosw routesw[] = {
1457{ SOCK_RAW, &routedomain, 0, PR_ATOMIC|PR_ADDR,
1458 0, route_output, raw_ctlinput, 0,
e3873585 1459 cpu0_soport, cpu0_ctlport,
984263bc
MD
1460 raw_init, 0, 0, 0,
1461 &route_usrreqs
1462}
1463};
1464
ba4257b8 1465static struct domain routedomain = {
9c70fe43
JH
1466 PF_ROUTE, "route", NULL, NULL, NULL,
1467 routesw, &routesw[(sizeof routesw)/(sizeof routesw[0])],
1468};
984263bc
MD
1469
1470DOMAIN_SET(route);
47db7f9b 1471