backtrace: Use print_backtrace()
[dragonfly.git] / sys / net / rtsock.c
... / ...
CommitLineData
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
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 $
67 * $DragonFly: src/sys/net/rtsock.c,v 1.45 2008/10/27 02:56:30 sephe Exp $
68 */
69
70#include "opt_sctp.h"
71
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>
77#include <sys/priv.h>
78#include <sys/malloc.h>
79#include <sys/mbuf.h>
80#include <sys/protosw.h>
81#include <sys/socket.h>
82#include <sys/socketvar.h>
83#include <sys/domain.h>
84#include <sys/thread2.h>
85
86#include <net/if.h>
87#include <net/route.h>
88#include <net/raw_cb.h>
89#include <net/netmsg2.h>
90
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
96MALLOC_DEFINE(M_RTABLE, "routetbl", "routing tables");
97
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
106static const struct sockaddr route_src = { 2, PF_ROUTE, };
107
108struct walkarg {
109 int w_tmemsize;
110 int w_op, w_arg;
111 void *w_tmem;
112 struct sysctl_req *w_req;
113};
114
115static struct mbuf *
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);
119static int rt_xaddrs (char *, char *, struct rt_addrinfo *);
120static int sysctl_dumpentry (struct radix_node *rn, void *vw);
121static int sysctl_iflist (int af, struct walkarg *w);
122static int route_output(struct mbuf *, struct socket *, ...);
123static void rt_setmetrics (u_long, struct rt_metrics *,
124 struct rt_metrics *);
125
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{
133 int error;
134
135 crit_enter();
136 error = raw_usrreqs.pru_abort(so);
137 crit_exit();
138 return error;
139}
140
141/* pru_accept is EOPNOTSUPP */
142
143static int
144rts_attach(struct socket *so, int proto, struct pru_attach_info *ai)
145{
146 struct rawcb *rp;
147 int error;
148
149 if (sotorawcb(so) != NULL)
150 return EISCONN; /* XXX panic? */
151
152 rp = kmalloc(sizeof *rp, M_PCB, M_WAITOK | M_ZERO);
153
154 /*
155 * The critical section is necessary to block protocols from sending
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
159 * eliminate the critical section.
160 */
161 crit_enter();
162 so->so_pcb = rp;
163 error = raw_attach(so, proto, ai->sb_rlimit);
164 rp = sotorawcb(so);
165 if (error) {
166 crit_exit();
167 kfree(rp, M_PCB);
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;
188 crit_exit();
189 return 0;
190}
191
192static int
193rts_bind(struct socket *so, struct sockaddr *nam, struct thread *td)
194{
195 int error;
196
197 crit_enter();
198 error = raw_usrreqs.pru_bind(so, nam, td); /* xxx just EINVAL */
199 crit_exit();
200 return error;
201}
202
203static int
204rts_connect(struct socket *so, struct sockaddr *nam, struct thread *td)
205{
206 int error;
207
208 crit_enter();
209 error = raw_usrreqs.pru_connect(so, nam, td); /* XXX just EINVAL */
210 crit_exit();
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);
221 int error;
222
223 crit_enter();
224 if (rp != NULL) {
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);
242 crit_exit();
243 return error;
244}
245
246static int
247rts_disconnect(struct socket *so)
248{
249 int error;
250
251 crit_enter();
252 error = raw_usrreqs.pru_disconnect(so);
253 crit_exit();
254 return error;
255}
256
257/* pru_listen is EOPNOTSUPP */
258
259static int
260rts_peeraddr(struct socket *so, struct sockaddr **nam)
261{
262 int error;
263
264 crit_enter();
265 error = raw_usrreqs.pru_peeraddr(so, nam);
266 crit_exit();
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,
275 struct mbuf *control, struct thread *td)
276{
277 int error;
278
279 crit_enter();
280 error = raw_usrreqs.pru_send(so, flags, m, nam, control, td);
281 crit_exit();
282 return error;
283}
284
285/* pru_sense is null */
286
287static int
288rts_shutdown(struct socket *so)
289{
290 int error;
291
292 crit_enter();
293 error = raw_usrreqs.pru_shutdown(so);
294 crit_exit();
295 return error;
296}
297
298static int
299rts_sockaddr(struct socket *so, struct sockaddr **nam)
300{
301 int error;
302
303 crit_enter();
304 error = raw_usrreqs.pru_sockaddr(so, nam);
305 crit_exit();
306 return error;
307}
308
309static struct pr_usrreqs route_usrreqs = {
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
330};
331
332static __inline sa_family_t
333familyof(struct sockaddr *sa)
334{
335 return (sa != NULL ? sa->sa_family : 0);
336}
337
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 */
345static void
346rts_input_handler(struct netmsg *msg)
347{
348 static const struct sockaddr route_dst = { 2, PF_ROUTE, };
349 struct sockproto route_proto;
350 struct netmsg_packet *pmsg;
351 struct mbuf *m;
352 sa_family_t family;
353 struct rawcb *skip;
354
355 pmsg = (void *)msg;
356 family = pmsg->nm_netmsg.nm_lmsg.u.ms_result;
357 route_proto.sp_family = PF_ROUTE;
358 route_proto.sp_protocol = family;
359
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);
367}
368
369static void
370rts_input_skip(struct mbuf *m, sa_family_t family, struct rawcb *skip)
371{
372 struct netmsg_packet *pmsg;
373 lwkt_port_t port;
374
375 M_ASSERTPKTHDR(m);
376
377 port = cpu0_soport(NULL, NULL, NULL, 0);
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;
383 m->m_pkthdr.header = skip; /* XXX steal field in pkthdr */
384 lwkt_sendmsg(port, &pmsg->nm_netmsg.nm_lmsg);
385}
386
387static __inline void
388rts_input(struct mbuf *m, sa_family_t family)
389{
390 rts_input_skip(m, family, NULL);
391}
392
393static void *
394reallocbuf(void *ptr, size_t len, size_t olen)
395{
396 void *newptr;
397
398 newptr = kmalloc(len, M_RTABLE, M_INTWAIT | M_NULLOK);
399 if (newptr == NULL)
400 return NULL;
401 bcopy(ptr, newptr, olen);
402 kfree(ptr, M_RTABLE);
403 return (newptr);
404}
405
406/*
407 * Internal helper routine for route_output().
408 */
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 =
424 TAILQ_FIRST(&rt->rt_ifp->if_addrheads[mycpuid])
425 ->ifa->ifa_addr;
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 } else if (rt->rt_ifp != NULL) {
435 rtm->rtm_index = rt->rt_ifp->if_index;
436 }
437
438 msglen = rt_msgsize(rtm->rtm_type, rtinfo);
439 if (rtm->rtm_msglen < msglen) {
440 rtm = reallocbuf(rtm, msglen, rtm->rtm_msglen);
441 if (rtm == NULL)
442 return (ENOBUFS);
443 *prtm = rtm;
444 }
445 rt_msg_buffer(rtm->rtm_type, rtinfo, rtm, msglen);
446
447 rtm->rtm_flags = rt->rt_flags;
448 rtm->rtm_rmx = rt->rt_rmx;
449 rtm->rtm_addrs = rtinfo->rti_addrs;
450
451 return (0);
452}
453
454static void route_output_add_callback(int, int, struct rt_addrinfo *,
455 struct rtentry *, void *);
456static void route_output_delete_callback(int, int, struct rt_addrinfo *,
457 struct rtentry *, void *);
458static int route_output_get_callback(int, struct rt_addrinfo *,
459 struct rtentry *, void *, int);
460static int route_output_change_callback(int, struct rt_addrinfo *,
461 struct rtentry *, void *, int);
462static int route_output_lock_callback(int, struct rt_addrinfo *,
463 struct rtentry *, void *, int);
464
465/*ARGSUSED*/
466static int
467route_output(struct mbuf *m, struct socket *so, ...)
468{
469 struct rt_msghdr *rtm = NULL;
470 struct rawcb *rp = NULL;
471 struct pr_output_info *oi;
472 struct rt_addrinfo rtinfo;
473 int len, error = 0;
474 __va_list ap;
475
476 M_ASSERTPKTHDR(m);
477
478 __va_start(ap, so);
479 oi = __va_arg(ap, struct pr_output_info *);
480 __va_end(ap);
481
482#define gotoerr(e) { error = e; goto flush;}
483
484 if (m == NULL ||
485 (m->m_len < sizeof(long) &&
486 (m = m_pullup(m, sizeof(long))) == NULL))
487 return (ENOBUFS);
488 len = m->m_pkthdr.len;
489 if (len < sizeof(struct rt_msghdr) ||
490 len != mtod(m, struct rt_msghdr *)->rtm_msglen) {
491 rtinfo.rti_dst = NULL;
492 gotoerr(EINVAL);
493 }
494 rtm = kmalloc(len, M_RTABLE, M_INTWAIT | M_NULLOK);
495 if (rtm == NULL) {
496 rtinfo.rti_dst = NULL;
497 gotoerr(ENOBUFS);
498 }
499 m_copydata(m, 0, len, (caddr_t)rtm);
500 if (rtm->rtm_version != RTM_VERSION) {
501 rtinfo.rti_dst = NULL;
502 gotoerr(EPROTONOSUPPORT);
503 }
504 rtm->rtm_pid = oi->p_pid;
505 bzero(&rtinfo, sizeof(struct rt_addrinfo));
506 rtinfo.rti_addrs = rtm->rtm_addrs;
507 if (rt_xaddrs((char *)(rtm + 1), (char *)rtm + len, &rtinfo) != 0) {
508 rtinfo.rti_dst = NULL;
509 gotoerr(EINVAL);
510 }
511 rtinfo.rti_flags = rtm->rtm_flags;
512 if (rtinfo.rti_dst == NULL || rtinfo.rti_dst->sa_family >= AF_MAX ||
513 (rtinfo.rti_gateway && rtinfo.rti_gateway->sa_family >= AF_MAX))
514 gotoerr(EINVAL);
515
516 if (rtinfo.rti_genmask != NULL) {
517 struct radix_node *n;
518
519#define clen(s) (*(u_char *)(s))
520 n = rn_addmask((char *)rtinfo.rti_genmask, TRUE, 1);
521 if (n != NULL &&
522 rtinfo.rti_genmask->sa_len >= clen(n->rn_key) &&
523 bcmp((char *)rtinfo.rti_genmask + 1,
524 (char *)n->rn_key + 1, clen(n->rn_key) - 1) == 0)
525 rtinfo.rti_genmask = (struct sockaddr *)n->rn_key;
526 else
527 gotoerr(ENOBUFS);
528 }
529
530 /*
531 * Verify that the caller has the appropriate privilege; RTM_GET
532 * is the only operation the non-superuser is allowed.
533 */
534 if (rtm->rtm_type != RTM_GET &&
535 priv_check_cred(so->so_cred, PRIV_ROOT, 0) != 0)
536 gotoerr(EPERM);
537
538 switch (rtm->rtm_type) {
539 case RTM_ADD:
540 if (rtinfo.rti_gateway == NULL) {
541 error = EINVAL;
542 } else {
543 error = rtrequest1_global(RTM_ADD, &rtinfo,
544 route_output_add_callback, rtm);
545 }
546 break;
547 case RTM_DELETE:
548 /*
549 * note: &rtm passed as argument so 'rtm' can be replaced.
550 */
551 error = rtrequest1_global(RTM_DELETE, &rtinfo,
552 route_output_delete_callback, &rtm);
553 break;
554 case RTM_GET:
555 /*
556 * note: &rtm passed as argument so 'rtm' can be replaced.
557 */
558 error = rtsearch_global(RTM_GET, &rtinfo,
559 route_output_get_callback, &rtm,
560 RTS_NOEXACTMATCH);
561 break;
562 case RTM_CHANGE:
563 error = rtsearch_global(RTM_CHANGE, &rtinfo,
564 route_output_change_callback, rtm,
565 RTS_EXACTMATCH);
566 break;
567 case RTM_LOCK:
568 error = rtsearch_global(RTM_LOCK, &rtinfo,
569 route_output_lock_callback, rtm,
570 RTS_EXACTMATCH);
571 break;
572 default:
573 error = EOPNOTSUPP;
574 break;
575 }
576
577flush:
578 if (rtm != NULL) {
579 if (error != 0)
580 rtm->rtm_errno = error;
581 else
582 rtm->rtm_flags |= RTF_DONE;
583 }
584
585 /*
586 * Check to see if we don't want our own messages.
587 */
588 if (!(so->so_options & SO_USELOOPBACK)) {
589 if (route_cb.any_count <= 1) {
590 if (rtm != NULL)
591 kfree(rtm, M_RTABLE);
592 m_freem(m);
593 return (error);
594 }
595 /* There is another listener, so construct message */
596 rp = sotorawcb(so);
597 }
598 if (rtm != NULL) {
599 m_copyback(m, 0, rtm->rtm_msglen, (caddr_t)rtm);
600 if (m->m_pkthdr.len < rtm->rtm_msglen) {
601 m_freem(m);
602 m = NULL;
603 } else if (m->m_pkthdr.len > rtm->rtm_msglen)
604 m_adj(m, rtm->rtm_msglen - m->m_pkthdr.len);
605 kfree(rtm, M_RTABLE);
606 }
607 if (m != NULL)
608 rts_input_skip(m, familyof(rtinfo.rti_dst), rp);
609 return (error);
610}
611
612static void
613route_output_add_callback(int cmd, int error, struct rt_addrinfo *rtinfo,
614 struct rtentry *rt, void *arg)
615{
616 struct rt_msghdr *rtm = arg;
617
618 if (error == 0 && rt != NULL) {
619 rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx,
620 &rt->rt_rmx);
621 rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits);
622 rt->rt_rmx.rmx_locks |=
623 (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks);
624 rt->rt_genmask = rtinfo->rti_genmask;
625 rtm->rtm_index = rt->rt_ifp->if_index;
626 }
627}
628
629static void
630route_output_delete_callback(int cmd, int error, struct rt_addrinfo *rtinfo,
631 struct rtentry *rt, void *arg)
632{
633 struct rt_msghdr **rtm = arg;
634
635 if (error == 0 && rt) {
636 ++rt->rt_refcnt;
637 if (fillrtmsg(rtm, rt, rtinfo) != 0) {
638 error = ENOBUFS;
639 /* XXX no way to return the error */
640 }
641 --rt->rt_refcnt;
642 }
643}
644
645static int
646route_output_get_callback(int cmd, struct rt_addrinfo *rtinfo,
647 struct rtentry *rt, void *arg, int found_cnt)
648{
649 struct rt_msghdr **rtm = arg;
650 int error, found = 0;
651
652 if (((rtinfo->rti_flags ^ rt->rt_flags) & RTF_HOST) == 0)
653 found = 1;
654
655 error = fillrtmsg(rtm, rt, rtinfo);
656 if (!error && found) {
657 /* Got the exact match, we could return now! */
658 error = EJUSTRETURN;
659 }
660 return error;
661}
662
663static int
664route_output_change_callback(int cmd, struct rt_addrinfo *rtinfo,
665 struct rtentry *rt, void *arg, int found_cnt)
666{
667 struct rt_msghdr *rtm = arg;
668 struct ifaddr *ifa;
669 int error = 0;
670
671 /*
672 * new gateway could require new ifaddr, ifp;
673 * flags may also be different; ifp may be specified
674 * by ll sockaddr when protocol address is ambiguous
675 */
676 if (((rt->rt_flags & RTF_GATEWAY) && rtinfo->rti_gateway != NULL) ||
677 rtinfo->rti_ifpaddr != NULL ||
678 (rtinfo->rti_ifaaddr != NULL &&
679 !sa_equal(rtinfo->rti_ifaaddr, rt->rt_ifa->ifa_addr))) {
680 error = rt_getifa(rtinfo);
681 if (error != 0)
682 goto done;
683 }
684 if (rtinfo->rti_gateway != NULL) {
685 /*
686 * We only need to generate rtmsg upon the
687 * first route to be changed.
688 */
689 error = rt_setgate(rt, rt_key(rt), rtinfo->rti_gateway,
690 found_cnt == 1 ? RTL_REPORTMSG : RTL_DONTREPORT);
691 if (error != 0)
692 goto done;
693 }
694 if ((ifa = rtinfo->rti_ifa) != NULL) {
695 struct ifaddr *oifa = rt->rt_ifa;
696
697 if (oifa != ifa) {
698 if (oifa && oifa->ifa_rtrequest)
699 oifa->ifa_rtrequest(RTM_DELETE, rt, rtinfo);
700 IFAFREE(rt->rt_ifa);
701 IFAREF(ifa);
702 rt->rt_ifa = ifa;
703 rt->rt_ifp = rtinfo->rti_ifp;
704 }
705 }
706 rt_setmetrics(rtm->rtm_inits, &rtm->rtm_rmx, &rt->rt_rmx);
707 if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest)
708 rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, rtinfo);
709 if (rtinfo->rti_genmask != NULL)
710 rt->rt_genmask = rtinfo->rti_genmask;
711 rtm->rtm_index = rt->rt_ifp->if_index;
712done:
713 return error;
714}
715
716static int
717route_output_lock_callback(int cmd, struct rt_addrinfo *rtinfo,
718 struct rtentry *rt, void *arg,
719 int found_cnt __unused)
720{
721 struct rt_msghdr *rtm = arg;
722
723 rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits);
724 rt->rt_rmx.rmx_locks |=
725 (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks);
726 return 0;
727}
728
729static void
730rt_setmetrics(u_long which, struct rt_metrics *in, struct rt_metrics *out)
731{
732#define setmetric(flag, elt) if (which & (flag)) out->elt = in->elt;
733 setmetric(RTV_RPIPE, rmx_recvpipe);
734 setmetric(RTV_SPIPE, rmx_sendpipe);
735 setmetric(RTV_SSTHRESH, rmx_ssthresh);
736 setmetric(RTV_RTT, rmx_rtt);
737 setmetric(RTV_RTTVAR, rmx_rttvar);
738 setmetric(RTV_HOPCOUNT, rmx_hopcount);
739 setmetric(RTV_MTU, rmx_mtu);
740 setmetric(RTV_EXPIRE, rmx_expire);
741#undef setmetric
742}
743
744#define ROUNDUP(a) \
745 ((a) > 0 ? (1 + (((a) - 1) | (sizeof(long) - 1))) : sizeof(long))
746
747/*
748 * Extract the addresses of the passed sockaddrs.
749 * Do a little sanity checking so as to avoid bad memory references.
750 * This data is derived straight from userland.
751 */
752static int
753rt_xaddrs(char *cp, char *cplim, struct rt_addrinfo *rtinfo)
754{
755 struct sockaddr *sa;
756 int i;
757
758 for (i = 0; (i < RTAX_MAX) && (cp < cplim); i++) {
759 if ((rtinfo->rti_addrs & (1 << i)) == 0)
760 continue;
761 sa = (struct sockaddr *)cp;
762 /*
763 * It won't fit.
764 */
765 if ((cp + sa->sa_len) > cplim) {
766 return (EINVAL);
767 }
768
769 /*
770 * There are no more... Quit now.
771 * If there are more bits, they are in error.
772 * I've seen this. route(1) can evidently generate these.
773 * This causes kernel to core dump.
774 * For compatibility, if we see this, point to a safe address.
775 */
776 if (sa->sa_len == 0) {
777 static struct sockaddr sa_zero = {
778 sizeof sa_zero, AF_INET,
779 };
780
781 rtinfo->rti_info[i] = &sa_zero;
782 kprintf("rtsock: received more addr bits than sockaddrs.\n");
783 return (0); /* should be EINVAL but for compat */
784 }
785
786 /* Accept the sockaddr. */
787 rtinfo->rti_info[i] = sa;
788 cp += ROUNDUP(sa->sa_len);
789 }
790 return (0);
791}
792
793static int
794rt_msghdrsize(int type)
795{
796 switch (type) {
797 case RTM_DELADDR:
798 case RTM_NEWADDR:
799 return sizeof(struct ifa_msghdr);
800 case RTM_DELMADDR:
801 case RTM_NEWMADDR:
802 return sizeof(struct ifma_msghdr);
803 case RTM_IFINFO:
804 return sizeof(struct if_msghdr);
805 case RTM_IFANNOUNCE:
806 case RTM_IEEE80211:
807 return sizeof(struct if_announcemsghdr);
808 default:
809 return sizeof(struct rt_msghdr);
810 }
811}
812
813static int
814rt_msgsize(int type, struct rt_addrinfo *rtinfo)
815{
816 int len, i;
817
818 len = rt_msghdrsize(type);
819 for (i = 0; i < RTAX_MAX; i++) {
820 if (rtinfo->rti_info[i] != NULL)
821 len += ROUNDUP(rtinfo->rti_info[i]->sa_len);
822 }
823 len = ALIGN(len);
824 return len;
825}
826
827/*
828 * Build a routing message in a buffer.
829 * Copy the addresses in the rtinfo->rti_info[] sockaddr array
830 * to the end of the buffer after the message header.
831 *
832 * Set the rtinfo->rti_addrs bitmask of addresses present in rtinfo->rti_info[].
833 * This side-effect can be avoided if we reorder the addrs bitmask field in all
834 * the route messages to line up so we can set it here instead of back in the
835 * calling routine.
836 */
837static void
838rt_msg_buffer(int type, struct rt_addrinfo *rtinfo, void *buf, int msglen)
839{
840 struct rt_msghdr *rtm;
841 char *cp;
842 int dlen, i;
843
844 rtm = (struct rt_msghdr *) buf;
845 rtm->rtm_version = RTM_VERSION;
846 rtm->rtm_type = type;
847 rtm->rtm_msglen = msglen;
848
849 cp = (char *)buf + rt_msghdrsize(type);
850 rtinfo->rti_addrs = 0;
851 for (i = 0; i < RTAX_MAX; i++) {
852 struct sockaddr *sa;
853
854 if ((sa = rtinfo->rti_info[i]) == NULL)
855 continue;
856 rtinfo->rti_addrs |= (1 << i);
857 dlen = ROUNDUP(sa->sa_len);
858 bcopy(sa, cp, dlen);
859 cp += dlen;
860 }
861}
862
863/*
864 * Build a routing message in a mbuf chain.
865 * Copy the addresses in the rtinfo->rti_info[] sockaddr array
866 * to the end of the mbuf after the message header.
867 *
868 * Set the rtinfo->rti_addrs bitmask of addresses present in rtinfo->rti_info[].
869 * This side-effect can be avoided if we reorder the addrs bitmask field in all
870 * the route messages to line up so we can set it here instead of back in the
871 * calling routine.
872 */
873static struct mbuf *
874rt_msg_mbuf(int type, struct rt_addrinfo *rtinfo)
875{
876 struct mbuf *m;
877 struct rt_msghdr *rtm;
878 int hlen, len;
879 int i;
880
881 hlen = rt_msghdrsize(type);
882 KASSERT(hlen <= MCLBYTES, ("rt_msg_mbuf: hlen %d doesn't fit", hlen));
883
884 m = m_getl(hlen, MB_DONTWAIT, MT_DATA, M_PKTHDR, NULL);
885 if (m == NULL)
886 return (NULL);
887 mbuftrackid(m, 32);
888 m->m_pkthdr.len = m->m_len = hlen;
889 m->m_pkthdr.rcvif = NULL;
890 rtinfo->rti_addrs = 0;
891 len = hlen;
892 for (i = 0; i < RTAX_MAX; i++) {
893 struct sockaddr *sa;
894 int dlen;
895
896 if ((sa = rtinfo->rti_info[i]) == NULL)
897 continue;
898 rtinfo->rti_addrs |= (1 << i);
899 dlen = ROUNDUP(sa->sa_len);
900 m_copyback(m, len, dlen, (caddr_t)sa); /* can grow mbuf chain */
901 len += dlen;
902 }
903 if (m->m_pkthdr.len != len) { /* one of the m_copyback() calls failed */
904 m_freem(m);
905 return (NULL);
906 }
907 rtm = mtod(m, struct rt_msghdr *);
908 bzero(rtm, hlen);
909 rtm->rtm_msglen = len;
910 rtm->rtm_version = RTM_VERSION;
911 rtm->rtm_type = type;
912 return (m);
913}
914
915/*
916 * This routine is called to generate a message from the routing
917 * socket indicating that a redirect has occurred, a routing lookup
918 * has failed, or that a protocol has detected timeouts to a particular
919 * destination.
920 */
921void
922rt_missmsg(int type, struct rt_addrinfo *rtinfo, int flags, int error)
923{
924 struct sockaddr *dst = rtinfo->rti_info[RTAX_DST];
925 struct rt_msghdr *rtm;
926 struct mbuf *m;
927
928 if (route_cb.any_count == 0)
929 return;
930 m = rt_msg_mbuf(type, rtinfo);
931 if (m == NULL)
932 return;
933 rtm = mtod(m, struct rt_msghdr *);
934 rtm->rtm_flags = RTF_DONE | flags;
935 rtm->rtm_errno = error;
936 rtm->rtm_addrs = rtinfo->rti_addrs;
937 rts_input(m, familyof(dst));
938}
939
940void
941rt_dstmsg(int type, struct sockaddr *dst, int error)
942{
943 struct rt_msghdr *rtm;
944 struct rt_addrinfo addrs;
945 struct mbuf *m;
946
947 if (route_cb.any_count == 0)
948 return;
949 bzero(&addrs, sizeof(struct rt_addrinfo));
950 addrs.rti_info[RTAX_DST] = dst;
951 m = rt_msg_mbuf(type, &addrs);
952 if (m == NULL)
953 return;
954 rtm = mtod(m, struct rt_msghdr *);
955 rtm->rtm_flags = RTF_DONE;
956 rtm->rtm_errno = error;
957 rtm->rtm_addrs = addrs.rti_addrs;
958 rts_input(m, familyof(dst));
959}
960
961/*
962 * This routine is called to generate a message from the routing
963 * socket indicating that the status of a network interface has changed.
964 */
965void
966rt_ifmsg(struct ifnet *ifp)
967{
968 struct if_msghdr *ifm;
969 struct mbuf *m;
970 struct rt_addrinfo rtinfo;
971
972 if (route_cb.any_count == 0)
973 return;
974 bzero(&rtinfo, sizeof(struct rt_addrinfo));
975 m = rt_msg_mbuf(RTM_IFINFO, &rtinfo);
976 if (m == NULL)
977 return;
978 ifm = mtod(m, struct if_msghdr *);
979 ifm->ifm_index = ifp->if_index;
980 ifm->ifm_flags = ifp->if_flags;
981 ifm->ifm_data = ifp->if_data;
982 ifm->ifm_addrs = 0;
983 rts_input(m, 0);
984}
985
986static void
987rt_ifamsg(int cmd, struct ifaddr *ifa)
988{
989 struct ifa_msghdr *ifam;
990 struct rt_addrinfo rtinfo;
991 struct mbuf *m;
992 struct ifnet *ifp = ifa->ifa_ifp;
993
994 bzero(&rtinfo, sizeof(struct rt_addrinfo));
995 rtinfo.rti_ifaaddr = ifa->ifa_addr;
996 rtinfo.rti_ifpaddr =
997 TAILQ_FIRST(&ifp->if_addrheads[mycpuid])->ifa->ifa_addr;
998 rtinfo.rti_netmask = ifa->ifa_netmask;
999 rtinfo.rti_bcastaddr = ifa->ifa_dstaddr;
1000
1001 m = rt_msg_mbuf(cmd, &rtinfo);
1002 if (m == NULL)
1003 return;
1004
1005 ifam = mtod(m, struct ifa_msghdr *);
1006 ifam->ifam_index = ifp->if_index;
1007 ifam->ifam_metric = ifa->ifa_metric;
1008 ifam->ifam_flags = ifa->ifa_flags;
1009 ifam->ifam_addrs = rtinfo.rti_addrs;
1010
1011 rts_input(m, familyof(ifa->ifa_addr));
1012}
1013
1014void
1015rt_rtmsg(int cmd, struct rtentry *rt, struct ifnet *ifp, int error)
1016{
1017 struct rt_msghdr *rtm;
1018 struct rt_addrinfo rtinfo;
1019 struct mbuf *m;
1020 struct sockaddr *dst;
1021
1022 if (rt == NULL)
1023 return;
1024
1025 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1026 rtinfo.rti_dst = dst = rt_key(rt);
1027 rtinfo.rti_gateway = rt->rt_gateway;
1028 rtinfo.rti_netmask = rt_mask(rt);
1029 if (ifp != NULL) {
1030 rtinfo.rti_ifpaddr =
1031 TAILQ_FIRST(&ifp->if_addrheads[mycpuid])->ifa->ifa_addr;
1032 }
1033 rtinfo.rti_ifaaddr = rt->rt_ifa->ifa_addr;
1034
1035 m = rt_msg_mbuf(cmd, &rtinfo);
1036 if (m == NULL)
1037 return;
1038
1039 rtm = mtod(m, struct rt_msghdr *);
1040 if (ifp != NULL)
1041 rtm->rtm_index = ifp->if_index;
1042 rtm->rtm_flags |= rt->rt_flags;
1043 rtm->rtm_errno = error;
1044 rtm->rtm_addrs = rtinfo.rti_addrs;
1045
1046 rts_input(m, familyof(dst));
1047}
1048
1049/*
1050 * This is called to generate messages from the routing socket
1051 * indicating a network interface has had addresses associated with it.
1052 * if we ever reverse the logic and replace messages TO the routing
1053 * socket indicate a request to configure interfaces, then it will
1054 * be unnecessary as the routing socket will automatically generate
1055 * copies of it.
1056 */
1057void
1058rt_newaddrmsg(int cmd, struct ifaddr *ifa, int error, struct rtentry *rt)
1059{
1060#ifdef SCTP
1061 /*
1062 * notify the SCTP stack
1063 * this will only get called when an address is added/deleted
1064 * XXX pass the ifaddr struct instead if ifa->ifa_addr...
1065 */
1066 if (cmd == RTM_ADD)
1067 sctp_add_ip_address(ifa);
1068 else if (cmd == RTM_DELETE)
1069 sctp_delete_ip_address(ifa);
1070#endif /* SCTP */
1071
1072 if (route_cb.any_count == 0)
1073 return;
1074
1075 if (cmd == RTM_ADD) {
1076 rt_ifamsg(RTM_NEWADDR, ifa);
1077 rt_rtmsg(RTM_ADD, rt, ifa->ifa_ifp, error);
1078 } else {
1079 KASSERT((cmd == RTM_DELETE), ("unknown cmd %d", cmd));
1080 rt_rtmsg(RTM_DELETE, rt, ifa->ifa_ifp, error);
1081 rt_ifamsg(RTM_DELADDR, ifa);
1082 }
1083}
1084
1085/*
1086 * This is the analogue to the rt_newaddrmsg which performs the same
1087 * function but for multicast group memberhips. This is easier since
1088 * there is no route state to worry about.
1089 */
1090void
1091rt_newmaddrmsg(int cmd, struct ifmultiaddr *ifma)
1092{
1093 struct rt_addrinfo rtinfo;
1094 struct mbuf *m = NULL;
1095 struct ifnet *ifp = ifma->ifma_ifp;
1096 struct ifma_msghdr *ifmam;
1097
1098 if (route_cb.any_count == 0)
1099 return;
1100
1101 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1102 rtinfo.rti_ifaaddr = ifma->ifma_addr;
1103 if (ifp != NULL && !TAILQ_EMPTY(&ifp->if_addrheads[mycpuid])) {
1104 rtinfo.rti_ifpaddr =
1105 TAILQ_FIRST(&ifp->if_addrheads[mycpuid])->ifa->ifa_addr;
1106 }
1107 /*
1108 * If a link-layer address is present, present it as a ``gateway''
1109 * (similarly to how ARP entries, e.g., are presented).
1110 */
1111 rtinfo.rti_gateway = ifma->ifma_lladdr;
1112
1113 m = rt_msg_mbuf(cmd, &rtinfo);
1114 if (m == NULL)
1115 return;
1116
1117 ifmam = mtod(m, struct ifma_msghdr *);
1118 ifmam->ifmam_index = ifp->if_index;
1119 ifmam->ifmam_addrs = rtinfo.rti_addrs;
1120
1121 rts_input(m, familyof(ifma->ifma_addr));
1122}
1123
1124static struct mbuf *
1125rt_makeifannouncemsg(struct ifnet *ifp, int type, int what,
1126 struct rt_addrinfo *info)
1127{
1128 struct if_announcemsghdr *ifan;
1129 struct mbuf *m;
1130
1131 if (route_cb.any_count == 0)
1132 return NULL;
1133
1134 bzero(info, sizeof(*info));
1135 m = rt_msg_mbuf(type, info);
1136 if (m == NULL)
1137 return NULL;
1138
1139 ifan = mtod(m, struct if_announcemsghdr *);
1140 ifan->ifan_index = ifp->if_index;
1141 strlcpy(ifan->ifan_name, ifp->if_xname, sizeof ifan->ifan_name);
1142 ifan->ifan_what = what;
1143 return m;
1144}
1145
1146/*
1147 * This is called to generate routing socket messages indicating
1148 * IEEE80211 wireless events.
1149 * XXX we piggyback on the RTM_IFANNOUNCE msg format in a clumsy way.
1150 */
1151void
1152rt_ieee80211msg(struct ifnet *ifp, int what, void *data, size_t data_len)
1153{
1154 struct rt_addrinfo info;
1155 struct mbuf *m;
1156
1157 m = rt_makeifannouncemsg(ifp, RTM_IEEE80211, what, &info);
1158 if (m == NULL)
1159 return;
1160
1161 /*
1162 * Append the ieee80211 data. Try to stick it in the
1163 * mbuf containing the ifannounce msg; otherwise allocate
1164 * a new mbuf and append.
1165 *
1166 * NB: we assume m is a single mbuf.
1167 */
1168 if (data_len > M_TRAILINGSPACE(m)) {
1169 struct mbuf *n = m_get(MB_DONTWAIT, MT_DATA);
1170 if (n == NULL) {
1171 m_freem(m);
1172 return;
1173 }
1174 bcopy(data, mtod(n, void *), data_len);
1175 n->m_len = data_len;
1176 m->m_next = n;
1177 } else if (data_len > 0) {
1178 bcopy(data, mtod(m, u_int8_t *) + m->m_len, data_len);
1179 m->m_len += data_len;
1180 }
1181 mbuftrackid(m, 33);
1182 if (m->m_flags & M_PKTHDR)
1183 m->m_pkthdr.len += data_len;
1184 mtod(m, struct if_announcemsghdr *)->ifan_msglen += data_len;
1185 rts_input(m, 0);
1186}
1187
1188/*
1189 * This is called to generate routing socket messages indicating
1190 * network interface arrival and departure.
1191 */
1192void
1193rt_ifannouncemsg(struct ifnet *ifp, int what)
1194{
1195 struct rt_addrinfo addrinfo;
1196 struct mbuf *m;
1197
1198 m = rt_makeifannouncemsg(ifp, RTM_IFANNOUNCE, what, &addrinfo);
1199 if (m != NULL)
1200 rts_input(m, 0);
1201}
1202
1203static int
1204resizewalkarg(struct walkarg *w, int len)
1205{
1206 void *newptr;
1207
1208 newptr = kmalloc(len, M_RTABLE, M_INTWAIT | M_NULLOK);
1209 if (newptr == NULL)
1210 return (ENOMEM);
1211 if (w->w_tmem != NULL)
1212 kfree(w->w_tmem, M_RTABLE);
1213 w->w_tmem = newptr;
1214 w->w_tmemsize = len;
1215 return (0);
1216}
1217
1218/*
1219 * This is used in dumping the kernel table via sysctl().
1220 */
1221int
1222sysctl_dumpentry(struct radix_node *rn, void *vw)
1223{
1224 struct walkarg *w = vw;
1225 struct rtentry *rt = (struct rtentry *)rn;
1226 struct rt_addrinfo rtinfo;
1227 int error, msglen;
1228
1229 if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg))
1230 return 0;
1231
1232 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1233 rtinfo.rti_dst = rt_key(rt);
1234 rtinfo.rti_gateway = rt->rt_gateway;
1235 rtinfo.rti_netmask = rt_mask(rt);
1236 rtinfo.rti_genmask = rt->rt_genmask;
1237 if (rt->rt_ifp != NULL) {
1238 rtinfo.rti_ifpaddr =
1239 TAILQ_FIRST(&rt->rt_ifp->if_addrheads[mycpuid])->ifa->ifa_addr;
1240 rtinfo.rti_ifaaddr = rt->rt_ifa->ifa_addr;
1241 if (rt->rt_ifp->if_flags & IFF_POINTOPOINT)
1242 rtinfo.rti_bcastaddr = rt->rt_ifa->ifa_dstaddr;
1243 }
1244 msglen = rt_msgsize(RTM_GET, &rtinfo);
1245 if (w->w_tmemsize < msglen && resizewalkarg(w, msglen) != 0)
1246 return (ENOMEM);
1247 rt_msg_buffer(RTM_GET, &rtinfo, w->w_tmem, msglen);
1248 if (w->w_req != NULL) {
1249 struct rt_msghdr *rtm = w->w_tmem;
1250
1251 rtm->rtm_flags = rt->rt_flags;
1252 rtm->rtm_use = rt->rt_use;
1253 rtm->rtm_rmx = rt->rt_rmx;
1254 rtm->rtm_index = rt->rt_ifp->if_index;
1255 rtm->rtm_errno = rtm->rtm_pid = rtm->rtm_seq = 0;
1256 rtm->rtm_addrs = rtinfo.rti_addrs;
1257 error = SYSCTL_OUT(w->w_req, rtm, msglen);
1258 return (error);
1259 }
1260 return (0);
1261}
1262
1263static int
1264sysctl_iflist(int af, struct walkarg *w)
1265{
1266 struct ifnet *ifp;
1267 struct rt_addrinfo rtinfo;
1268 int msglen, error;
1269
1270 bzero(&rtinfo, sizeof(struct rt_addrinfo));
1271 TAILQ_FOREACH(ifp, &ifnet, if_link) {
1272 struct ifaddr_container *ifac;
1273 struct ifaddr *ifa;
1274
1275 if (w->w_arg && w->w_arg != ifp->if_index)
1276 continue;
1277 ifac = TAILQ_FIRST(&ifp->if_addrheads[mycpuid]);
1278 ifa = ifac->ifa;
1279 rtinfo.rti_ifpaddr = ifa->ifa_addr;
1280 msglen = rt_msgsize(RTM_IFINFO, &rtinfo);
1281 if (w->w_tmemsize < msglen && resizewalkarg(w, msglen) != 0)
1282 return (ENOMEM);
1283 rt_msg_buffer(RTM_IFINFO, &rtinfo, w->w_tmem, msglen);
1284 rtinfo.rti_ifpaddr = NULL;
1285 if (w->w_req != NULL && w->w_tmem != NULL) {
1286 struct if_msghdr *ifm = w->w_tmem;
1287
1288 ifm->ifm_index = ifp->if_index;
1289 ifm->ifm_flags = ifp->if_flags;
1290 ifm->ifm_data = ifp->if_data;
1291 ifm->ifm_addrs = rtinfo.rti_addrs;
1292 error = SYSCTL_OUT(w->w_req, ifm, msglen);
1293 if (error)
1294 return (error);
1295 }
1296 while ((ifac = TAILQ_NEXT(ifac, ifa_link)) != NULL) {
1297 ifa = ifac->ifa;
1298
1299 if (af && af != ifa->ifa_addr->sa_family)
1300 continue;
1301 if (curproc->p_ucred->cr_prison &&
1302 prison_if(curproc->p_ucred, ifa->ifa_addr))
1303 continue;
1304 rtinfo.rti_ifaaddr = ifa->ifa_addr;
1305 rtinfo.rti_netmask = ifa->ifa_netmask;
1306 rtinfo.rti_bcastaddr = ifa->ifa_dstaddr;
1307 msglen = rt_msgsize(RTM_NEWADDR, &rtinfo);
1308 if (w->w_tmemsize < msglen &&
1309 resizewalkarg(w, msglen) != 0)
1310 return (ENOMEM);
1311 rt_msg_buffer(RTM_NEWADDR, &rtinfo, w->w_tmem, msglen);
1312 if (w->w_req != NULL) {
1313 struct ifa_msghdr *ifam = w->w_tmem;
1314
1315 ifam->ifam_index = ifa->ifa_ifp->if_index;
1316 ifam->ifam_flags = ifa->ifa_flags;
1317 ifam->ifam_metric = ifa->ifa_metric;
1318 ifam->ifam_addrs = rtinfo.rti_addrs;
1319 error = SYSCTL_OUT(w->w_req, w->w_tmem, msglen);
1320 if (error)
1321 return (error);
1322 }
1323 }
1324 rtinfo.rti_netmask = NULL;
1325 rtinfo.rti_ifaaddr = NULL;
1326 rtinfo.rti_bcastaddr = NULL;
1327 }
1328 return (0);
1329}
1330
1331static int
1332sysctl_rtsock(SYSCTL_HANDLER_ARGS)
1333{
1334 int *name = (int *)arg1;
1335 u_int namelen = arg2;
1336 struct radix_node_head *rnh;
1337 int i, error = EINVAL;
1338 int origcpu;
1339 u_char af;
1340 struct walkarg w;
1341
1342 name ++;
1343 namelen--;
1344 if (req->newptr)
1345 return (EPERM);
1346 if (namelen != 3 && namelen != 4)
1347 return (EINVAL);
1348 af = name[0];
1349 bzero(&w, sizeof w);
1350 w.w_op = name[1];
1351 w.w_arg = name[2];
1352 w.w_req = req;
1353
1354 /*
1355 * Optional third argument specifies cpu, used primarily for
1356 * debugging the route table.
1357 */
1358 if (namelen == 4) {
1359 if (name[3] < 0 || name[3] >= ncpus)
1360 return (EINVAL);
1361 origcpu = mycpuid;
1362 lwkt_migratecpu(name[3]);
1363 } else {
1364 origcpu = -1;
1365 }
1366 crit_enter();
1367 switch (w.w_op) {
1368 case NET_RT_DUMP:
1369 case NET_RT_FLAGS:
1370 for (i = 1; i <= AF_MAX; i++)
1371 if ((rnh = rt_tables[mycpuid][i]) &&
1372 (af == 0 || af == i) &&
1373 (error = rnh->rnh_walktree(rnh,
1374 sysctl_dumpentry, &w)))
1375 break;
1376 break;
1377
1378 case NET_RT_IFLIST:
1379 error = sysctl_iflist(af, &w);
1380 }
1381 crit_exit();
1382 if (w.w_tmem != NULL)
1383 kfree(w.w_tmem, M_RTABLE);
1384 if (origcpu >= 0)
1385 lwkt_migratecpu(origcpu);
1386 return (error);
1387}
1388
1389SYSCTL_NODE(_net, PF_ROUTE, routetable, CTLFLAG_RD, sysctl_rtsock, "");
1390
1391/*
1392 * Definitions of protocols supported in the ROUTE domain.
1393 */
1394
1395static struct domain routedomain; /* or at least forward */
1396
1397static struct protosw routesw[] = {
1398{ SOCK_RAW, &routedomain, 0, PR_ATOMIC|PR_ADDR,
1399 0, route_output, raw_ctlinput, 0,
1400 cpu0_soport, cpu0_ctlport,
1401 raw_init, 0, 0, 0,
1402 &route_usrreqs
1403}
1404};
1405
1406static struct domain routedomain = {
1407 PF_ROUTE, "route", NULL, NULL, NULL,
1408 routesw, &routesw[(sizeof routesw)/(sizeof routesw[0])],
1409};
1410
1411DOMAIN_SET(route);
1412