rtsock.c revision 1.190 1 /* $NetBSD: rtsock.c,v 1.190 2016/06/16 02:38:40 ozaki-r Exp $ */
2
3 /*
4 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
5 * All rights reserved.
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 project nor the names of its contributors
16 * may be used to endorse or promote products derived from this software
17 * without specific prior written permission.
18 *
19 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
20 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
21 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
22 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
23 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
24 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
25 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
26 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
27 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
28 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
29 * SUCH DAMAGE.
30 */
31
32 /*
33 * Copyright (c) 1988, 1991, 1993
34 * The Regents of the University of California. All rights reserved.
35 *
36 * Redistribution and use in source and binary forms, with or without
37 * modification, are permitted provided that the following conditions
38 * are met:
39 * 1. Redistributions of source code must retain the above copyright
40 * notice, this list of conditions and the following disclaimer.
41 * 2. Redistributions in binary form must reproduce the above copyright
42 * notice, this list of conditions and the following disclaimer in the
43 * documentation and/or other materials provided with the distribution.
44 * 3. Neither the name of the University nor the names of its contributors
45 * may be used to endorse or promote products derived from this software
46 * without specific prior written permission.
47 *
48 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
49 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
50 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
51 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
52 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
53 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
54 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
55 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
56 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
57 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
58 * SUCH DAMAGE.
59 *
60 * @(#)rtsock.c 8.7 (Berkeley) 10/12/95
61 */
62
63 #include <sys/cdefs.h>
64 __KERNEL_RCSID(0, "$NetBSD: rtsock.c,v 1.190 2016/06/16 02:38:40 ozaki-r Exp $");
65
66 #ifdef _KERNEL_OPT
67 #include "opt_inet.h"
68 #include "opt_mpls.h"
69 #include "opt_compat_netbsd.h"
70 #include "opt_sctp.h"
71 #endif
72
73 #include <sys/param.h>
74 #include <sys/systm.h>
75 #include <sys/proc.h>
76 #include <sys/socket.h>
77 #include <sys/socketvar.h>
78 #include <sys/domain.h>
79 #include <sys/protosw.h>
80 #include <sys/sysctl.h>
81 #include <sys/kauth.h>
82 #include <sys/kmem.h>
83 #include <sys/intr.h>
84
85 #include <net/if.h>
86 #include <net/if_llatbl.h>
87 #include <net/if_types.h>
88 #include <net/route.h>
89 #include <net/raw_cb.h>
90
91 #include <netinet/in_var.h>
92 #include <netinet/if_inarp.h>
93
94 #include <netmpls/mpls.h>
95
96 #ifdef SCTP
97 extern void sctp_add_ip_address(struct ifaddr *);
98 extern void sctp_delete_ip_address(struct ifaddr *);
99 #endif
100
101 #if defined(COMPAT_14) || defined(COMPAT_50)
102 #include <compat/net/if.h>
103 #include <compat/net/route.h>
104 #endif
105 #ifdef COMPAT_RTSOCK
106 #define RTM_XVERSION RTM_OVERSION
107 #define RT_XADVANCE(a,b) RT_OADVANCE(a,b)
108 #define RT_XROUNDUP(n) RT_OROUNDUP(n)
109 #define PF_XROUTE PF_OROUTE
110 #define rt_xmsghdr rt_msghdr50
111 #define if_xmsghdr if_msghdr /* if_msghdr50 is for RTM_OIFINFO */
112 #define ifa_xmsghdr ifa_msghdr50
113 #define if_xannouncemsghdr if_announcemsghdr50
114 #define COMPATNAME(x) compat_50_ ## x
115 #define DOMAINNAME "oroute"
116 CTASSERT(sizeof(struct ifa_xmsghdr) == 20);
117 DOMAIN_DEFINE(compat_50_routedomain); /* forward declare and add to link set */
118 #else /* COMPAT_RTSOCK */
119 #define RTM_XVERSION RTM_VERSION
120 #define RT_XADVANCE(a,b) RT_ADVANCE(a,b)
121 #define RT_XROUNDUP(n) RT_ROUNDUP(n)
122 #define PF_XROUTE PF_ROUTE
123 #define rt_xmsghdr rt_msghdr
124 #define if_xmsghdr if_msghdr
125 #define ifa_xmsghdr ifa_msghdr
126 #define if_xannouncemsghdr if_announcemsghdr
127 #define COMPATNAME(x) x
128 #define DOMAINNAME "route"
129 CTASSERT(sizeof(struct ifa_xmsghdr) == 24);
130 #ifdef COMPAT_50
131 #define COMPATCALL(name, args) compat_50_ ## name args
132 #endif
133 DOMAIN_DEFINE(routedomain); /* forward declare and add to link set */
134 #undef COMPAT_50
135 #undef COMPAT_14
136 #endif /* COMPAT_RTSOCK */
137
138 #ifndef COMPATCALL
139 #define COMPATCALL(name, args) do { } while (/*CONSTCOND*/ 0)
140 #endif
141
142 #ifdef RTSOCK_DEBUG
143 #define RT_IN_PRINT(info, b, a) (in_print((b), sizeof(b), \
144 &((const struct sockaddr_in *)(info)->rti_info[(a)])->sin_addr), (b))
145 #endif /* RTSOCK_DEBUG */
146
147 struct route_info COMPATNAME(route_info) = {
148 .ri_dst = { .sa_len = 2, .sa_family = PF_XROUTE, },
149 .ri_src = { .sa_len = 2, .sa_family = PF_XROUTE, },
150 .ri_maxqlen = IFQ_MAXLEN,
151 };
152
153 #define PRESERVED_RTF (RTF_UP | RTF_GATEWAY | RTF_HOST | RTF_DONE | RTF_MASK)
154
155 static void COMPATNAME(route_init)(void);
156 static int COMPATNAME(route_output)(struct mbuf *, struct socket *);
157
158 static int rt_xaddrs(u_char, const char *, const char *, struct rt_addrinfo *);
159 static struct mbuf *rt_makeifannouncemsg(struct ifnet *, int, int,
160 struct rt_addrinfo *);
161 static int rt_msg2(int, struct rt_addrinfo *, void *, struct rt_walkarg *, int *);
162 static void rt_setmetrics(int, const struct rt_xmsghdr *, struct rtentry *);
163 static void rtm_setmetrics(const struct rtentry *, struct rt_xmsghdr *);
164 static void sysctl_net_route_setup(struct sysctllog **);
165 static int sysctl_dumpentry(struct rtentry *, void *);
166 static int sysctl_iflist(int, struct rt_walkarg *, int);
167 static int sysctl_rtable(SYSCTLFN_PROTO);
168 static void rt_adjustcount(int, int);
169
170 static const struct protosw COMPATNAME(route_protosw)[];
171
172 static void
173 rt_adjustcount(int af, int cnt)
174 {
175 struct route_cb * const cb = &COMPATNAME(route_info).ri_cb;
176
177 cb->any_count += cnt;
178
179 switch (af) {
180 case AF_INET:
181 cb->ip_count += cnt;
182 return;
183 #ifdef INET6
184 case AF_INET6:
185 cb->ip6_count += cnt;
186 return;
187 #endif
188 case AF_MPLS:
189 cb->mpls_count += cnt;
190 return;
191 }
192 }
193
194 static int
195 COMPATNAME(route_attach)(struct socket *so, int proto)
196 {
197 struct rawcb *rp;
198 int s, error;
199
200 KASSERT(sotorawcb(so) == NULL);
201 rp = kmem_zalloc(sizeof(*rp), KM_SLEEP);
202 rp->rcb_len = sizeof(*rp);
203 so->so_pcb = rp;
204
205 s = splsoftnet();
206 if ((error = raw_attach(so, proto)) == 0) {
207 rt_adjustcount(rp->rcb_proto.sp_protocol, 1);
208 rp->rcb_laddr = &COMPATNAME(route_info).ri_src;
209 rp->rcb_faddr = &COMPATNAME(route_info).ri_dst;
210 }
211 splx(s);
212
213 if (error) {
214 kmem_free(rp, sizeof(*rp));
215 so->so_pcb = NULL;
216 return error;
217 }
218
219 soisconnected(so);
220 so->so_options |= SO_USELOOPBACK;
221 KASSERT(solocked(so));
222
223 return error;
224 }
225
226 static void
227 COMPATNAME(route_detach)(struct socket *so)
228 {
229 struct rawcb *rp = sotorawcb(so);
230 int s;
231
232 KASSERT(rp != NULL);
233 KASSERT(solocked(so));
234
235 s = splsoftnet();
236 rt_adjustcount(rp->rcb_proto.sp_protocol, -1);
237 raw_detach(so);
238 splx(s);
239 }
240
241 static int
242 COMPATNAME(route_accept)(struct socket *so, struct sockaddr *nam)
243 {
244 KASSERT(solocked(so));
245
246 panic("route_accept");
247
248 return EOPNOTSUPP;
249 }
250
251 static int
252 COMPATNAME(route_bind)(struct socket *so, struct sockaddr *nam, struct lwp *l)
253 {
254 KASSERT(solocked(so));
255
256 return EOPNOTSUPP;
257 }
258
259 static int
260 COMPATNAME(route_listen)(struct socket *so, struct lwp *l)
261 {
262 KASSERT(solocked(so));
263
264 return EOPNOTSUPP;
265 }
266
267 static int
268 COMPATNAME(route_connect)(struct socket *so, struct sockaddr *nam, struct lwp *l)
269 {
270 KASSERT(solocked(so));
271
272 return EOPNOTSUPP;
273 }
274
275 static int
276 COMPATNAME(route_connect2)(struct socket *so, struct socket *so2)
277 {
278 KASSERT(solocked(so));
279
280 return EOPNOTSUPP;
281 }
282
283 static int
284 COMPATNAME(route_disconnect)(struct socket *so)
285 {
286 struct rawcb *rp = sotorawcb(so);
287 int s;
288
289 KASSERT(solocked(so));
290 KASSERT(rp != NULL);
291
292 s = splsoftnet();
293 soisdisconnected(so);
294 raw_disconnect(rp);
295 splx(s);
296
297 return 0;
298 }
299
300 static int
301 COMPATNAME(route_shutdown)(struct socket *so)
302 {
303 int s;
304
305 KASSERT(solocked(so));
306
307 /*
308 * Mark the connection as being incapable of further input.
309 */
310 s = splsoftnet();
311 socantsendmore(so);
312 splx(s);
313 return 0;
314 }
315
316 static int
317 COMPATNAME(route_abort)(struct socket *so)
318 {
319 KASSERT(solocked(so));
320
321 panic("route_abort");
322
323 return EOPNOTSUPP;
324 }
325
326 static int
327 COMPATNAME(route_ioctl)(struct socket *so, u_long cmd, void *nam,
328 struct ifnet * ifp)
329 {
330 return EOPNOTSUPP;
331 }
332
333 static int
334 COMPATNAME(route_stat)(struct socket *so, struct stat *ub)
335 {
336 KASSERT(solocked(so));
337
338 return 0;
339 }
340
341 static int
342 COMPATNAME(route_peeraddr)(struct socket *so, struct sockaddr *nam)
343 {
344 struct rawcb *rp = sotorawcb(so);
345
346 KASSERT(solocked(so));
347 KASSERT(rp != NULL);
348 KASSERT(nam != NULL);
349
350 if (rp->rcb_faddr == NULL)
351 return ENOTCONN;
352
353 raw_setpeeraddr(rp, nam);
354 return 0;
355 }
356
357 static int
358 COMPATNAME(route_sockaddr)(struct socket *so, struct sockaddr *nam)
359 {
360 struct rawcb *rp = sotorawcb(so);
361
362 KASSERT(solocked(so));
363 KASSERT(rp != NULL);
364 KASSERT(nam != NULL);
365
366 if (rp->rcb_faddr == NULL)
367 return ENOTCONN;
368
369 raw_setsockaddr(rp, nam);
370 return 0;
371 }
372
373 static int
374 COMPATNAME(route_rcvd)(struct socket *so, int flags, struct lwp *l)
375 {
376 KASSERT(solocked(so));
377
378 return EOPNOTSUPP;
379 }
380
381 static int
382 COMPATNAME(route_recvoob)(struct socket *so, struct mbuf *m, int flags)
383 {
384 KASSERT(solocked(so));
385
386 return EOPNOTSUPP;
387 }
388
389 static int
390 COMPATNAME(route_send)(struct socket *so, struct mbuf *m,
391 struct sockaddr *nam, struct mbuf *control, struct lwp *l)
392 {
393 int error = 0;
394 int s;
395
396 KASSERT(solocked(so));
397 KASSERT(so->so_proto == &COMPATNAME(route_protosw)[0]);
398
399 s = splsoftnet();
400 error = raw_send(so, m, nam, control, l, &COMPATNAME(route_output));
401 splx(s);
402
403 return error;
404 }
405
406 static int
407 COMPATNAME(route_sendoob)(struct socket *so, struct mbuf *m,
408 struct mbuf *control)
409 {
410 KASSERT(solocked(so));
411
412 m_freem(m);
413 m_freem(control);
414
415 return EOPNOTSUPP;
416 }
417 static int
418 COMPATNAME(route_purgeif)(struct socket *so, struct ifnet *ifp)
419 {
420
421 panic("route_purgeif");
422
423 return EOPNOTSUPP;
424 }
425
426 #ifdef INET
427 static int
428 route_get_sdl_index(struct rt_addrinfo *info, int *sdl_index)
429 {
430 struct rtentry *nrt;
431 int error;
432
433 error = rtrequest1(RTM_GET, info, &nrt);
434 if (error != 0)
435 return error;
436 /*
437 * nrt->rt_ifp->if_index may not be correct
438 * due to changing to ifplo0.
439 */
440 *sdl_index = satosdl(nrt->rt_gateway)->sdl_index;
441 rtfree(nrt);
442
443 return 0;
444 }
445 #endif /* INET */
446
447 static void
448 route_get_sdl(const struct ifnet *ifp, const struct sockaddr *dst,
449 struct sockaddr_dl *sdl, int *flags)
450 {
451 struct llentry *la;
452
453 KASSERT(ifp != NULL);
454
455 IF_AFDATA_RLOCK(ifp);
456 switch (dst->sa_family) {
457 case AF_INET:
458 la = lla_lookup(LLTABLE(ifp), 0, dst);
459 break;
460 case AF_INET6:
461 la = lla_lookup(LLTABLE6(ifp), 0, dst);
462 break;
463 default:
464 la = NULL;
465 KASSERTMSG(0, "Invalid AF=%d\n", dst->sa_family);
466 break;
467 }
468 IF_AFDATA_RUNLOCK(ifp);
469
470 void *a = (LLE_IS_VALID(la) && (la->la_flags & LLE_VALID) == LLE_VALID)
471 ? &la->ll_addr : NULL;
472
473 a = sockaddr_dl_init(sdl, sizeof(*sdl), ifp->if_index, ifp->if_type,
474 NULL, 0, a, ifp->if_addrlen);
475 KASSERT(a != NULL);
476
477 if (la != NULL) {
478 *flags = la->la_flags;
479 LLE_RUNLOCK(la);
480 }
481 }
482
483 static int
484 route_output_report(struct rtentry *rt, struct rt_addrinfo *info,
485 struct rt_xmsghdr *rtm, struct rt_xmsghdr **new_rtm)
486 {
487 int len;
488 struct ifnet *ifp;
489
490 if ((rtm->rtm_addrs & (RTA_IFP | RTA_IFA)) == 0)
491 ;
492 else if ((ifp = rt->rt_ifp) != NULL) {
493 const struct ifaddr *rtifa;
494 info->rti_info[RTAX_IFP] = ifp->if_dl->ifa_addr;
495 /* rtifa used to be simply rt->rt_ifa.
496 * If rt->rt_ifa != NULL, then
497 * rt_get_ifa() != NULL. So this
498 * ought to still be safe. --dyoung
499 */
500 rtifa = rt_get_ifa(rt);
501 info->rti_info[RTAX_IFA] = rtifa->ifa_addr;
502 #ifdef RTSOCK_DEBUG
503 if (info->rti_info[RTAX_IFA]->sa_family == AF_INET) {
504 char ibuf[INET_ADDRSTRLEN];
505 char abuf[INET_ADDRSTRLEN];
506 printf("%s: copying out RTAX_IFA %s "
507 "for info->rti_info[RTAX_DST] %s "
508 "ifa_getifa %p ifa_seqno %p\n",
509 __func__,
510 RT_IN_PRINT(info, ibuf, RTAX_IFA),
511 RT_IN_PRINT(info, abuf, RTAX_DST),
512 (void *)rtifa->ifa_getifa,
513 rtifa->ifa_seqno);
514 }
515 #endif /* RTSOCK_DEBUG */
516 if (ifp->if_flags & IFF_POINTOPOINT)
517 info->rti_info[RTAX_BRD] = rtifa->ifa_dstaddr;
518 else
519 info->rti_info[RTAX_BRD] = NULL;
520 rtm->rtm_index = ifp->if_index;
521 } else {
522 info->rti_info[RTAX_IFP] = NULL;
523 info->rti_info[RTAX_IFA] = NULL;
524 }
525 (void)rt_msg2(rtm->rtm_type, info, NULL, NULL, &len);
526 if (len > rtm->rtm_msglen) {
527 struct rt_xmsghdr *old_rtm = rtm;
528 R_Malloc(*new_rtm, struct rt_xmsghdr *, len);
529 if (*new_rtm == NULL)
530 return ENOBUFS;
531 (void)memcpy(*new_rtm, old_rtm, old_rtm->rtm_msglen);
532 rtm = *new_rtm;
533 }
534 (void)rt_msg2(rtm->rtm_type, info, rtm, NULL, 0);
535 rtm->rtm_flags = rt->rt_flags;
536 rtm_setmetrics(rt, rtm);
537 rtm->rtm_addrs = info->rti_addrs;
538
539 return 0;
540 }
541
542 /*ARGSUSED*/
543 int
544 COMPATNAME(route_output)(struct mbuf *m, struct socket *so)
545 {
546 struct sockproto proto = { .sp_family = PF_XROUTE, };
547 struct rt_xmsghdr *rtm = NULL;
548 struct rt_xmsghdr *old_rtm = NULL, *new_rtm = NULL;
549 struct rtentry *rt = NULL;
550 struct rtentry *saved_nrt = NULL;
551 struct rt_addrinfo info;
552 int len, error = 0;
553 struct ifnet *ifp = NULL;
554 struct ifaddr *ifa = NULL;
555 sa_family_t family;
556 struct sockaddr_dl sdl;
557
558 #define senderr(e) do { error = e; goto flush;} while (/*CONSTCOND*/ 0)
559 if (m == NULL || ((m->m_len < sizeof(int32_t)) &&
560 (m = m_pullup(m, sizeof(int32_t))) == NULL))
561 return ENOBUFS;
562 if ((m->m_flags & M_PKTHDR) == 0)
563 panic("%s", __func__);
564 len = m->m_pkthdr.len;
565 if (len < sizeof(*rtm) ||
566 len != mtod(m, struct rt_xmsghdr *)->rtm_msglen) {
567 info.rti_info[RTAX_DST] = NULL;
568 senderr(EINVAL);
569 }
570 R_Malloc(rtm, struct rt_xmsghdr *, len);
571 if (rtm == NULL) {
572 info.rti_info[RTAX_DST] = NULL;
573 senderr(ENOBUFS);
574 }
575 m_copydata(m, 0, len, rtm);
576 if (rtm->rtm_version != RTM_XVERSION) {
577 info.rti_info[RTAX_DST] = NULL;
578 senderr(EPROTONOSUPPORT);
579 }
580 rtm->rtm_pid = curproc->p_pid;
581 memset(&info, 0, sizeof(info));
582 info.rti_addrs = rtm->rtm_addrs;
583 if (rt_xaddrs(rtm->rtm_type, (const char *)(rtm + 1), len + (char *)rtm,
584 &info)) {
585 senderr(EINVAL);
586 }
587 info.rti_flags = rtm->rtm_flags;
588 #ifdef RTSOCK_DEBUG
589 if (info.rti_info[RTAX_DST]->sa_family == AF_INET) {
590 char abuf[INET_ADDRSTRLEN];
591 printf("%s: extracted info.rti_info[RTAX_DST] %s\n", __func__,
592 RT_IN_PRINT(&info, abuf, RTAX_DST));
593 }
594 #endif /* RTSOCK_DEBUG */
595 if (info.rti_info[RTAX_DST] == NULL ||
596 (info.rti_info[RTAX_DST]->sa_family >= AF_MAX)) {
597 senderr(EINVAL);
598 }
599 if (info.rti_info[RTAX_GATEWAY] != NULL &&
600 (info.rti_info[RTAX_GATEWAY]->sa_family >= AF_MAX)) {
601 senderr(EINVAL);
602 }
603
604 /*
605 * Verify that the caller has the appropriate privilege; RTM_GET
606 * is the only operation the non-superuser is allowed.
607 */
608 if (kauth_authorize_network(curlwp->l_cred, KAUTH_NETWORK_ROUTE,
609 0, rtm, NULL, NULL) != 0)
610 senderr(EACCES);
611
612 switch (rtm->rtm_type) {
613
614 case RTM_ADD:
615 if (info.rti_info[RTAX_GATEWAY] == NULL) {
616 senderr(EINVAL);
617 }
618 #ifdef INET
619 /* support for new ARP code with keeping backcompat */
620 if (info.rti_info[RTAX_GATEWAY]->sa_family == AF_LINK) {
621 const struct sockaddr_dl *sdlp =
622 satocsdl(info.rti_info[RTAX_GATEWAY]);
623
624 /* Allow routing requests by interface index */
625 if (sdlp->sdl_nlen == 0 && sdlp->sdl_alen == 0
626 && sdlp->sdl_slen == 0)
627 goto fallback;
628 /*
629 * Old arp binaries don't set the sdl_index
630 * so we have to complement it.
631 */
632 int sdl_index = sdlp->sdl_index;
633 if (sdl_index == 0) {
634 error = route_get_sdl_index(&info, &sdl_index);
635 if (error != 0)
636 goto fallback;
637 } else if (
638 info.rti_info[RTAX_DST]->sa_family == AF_INET) {
639 /*
640 * XXX workaround for SIN_PROXY case; proxy arp
641 * entry should be in an interface that has
642 * a network route including the destination,
643 * not a local (link) route that may not be a
644 * desired place, for example a tap.
645 */
646 const struct sockaddr_inarp *sina =
647 (const struct sockaddr_inarp *)
648 info.rti_info[RTAX_DST];
649 if (sina->sin_other & SIN_PROXY) {
650 error = route_get_sdl_index(&info,
651 &sdl_index);
652 if (error != 0)
653 goto fallback;
654 }
655 }
656 error = lla_rt_output(rtm->rtm_type, rtm->rtm_flags,
657 rtm->rtm_rmx.rmx_expire, &info, sdl_index);
658 break;
659 }
660 fallback:
661 #endif /* INET */
662 error = rtrequest1(rtm->rtm_type, &info, &saved_nrt);
663 if (error == 0) {
664 rt_setmetrics(rtm->rtm_inits, rtm, saved_nrt);
665 rtfree(saved_nrt);
666 }
667 break;
668
669 case RTM_DELETE:
670 #ifdef INET
671 /* support for new ARP code */
672 if (info.rti_info[RTAX_GATEWAY] &&
673 (info.rti_info[RTAX_GATEWAY]->sa_family == AF_LINK) &&
674 (rtm->rtm_flags & RTF_LLDATA) != 0) {
675 error = lla_rt_output(rtm->rtm_type, rtm->rtm_flags,
676 rtm->rtm_rmx.rmx_expire, &info, 0);
677 break;
678 }
679 #endif /* INET */
680 error = rtrequest1(rtm->rtm_type, &info, &saved_nrt);
681 if (error != 0)
682 break;
683
684 rt = saved_nrt;
685 info.rti_info[RTAX_DST] = rt_getkey(rt);
686 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
687 info.rti_info[RTAX_NETMASK] = rt_mask(rt);
688 info.rti_info[RTAX_TAG] = rt_gettag(rt);
689 error = route_output_report(rt, &info, rtm, &new_rtm);
690 if (error)
691 senderr(error);
692 if (new_rtm != NULL) {
693 old_rtm = rtm;
694 rtm = new_rtm;
695 }
696 break;
697
698 case RTM_GET:
699 case RTM_CHANGE:
700 case RTM_LOCK:
701 /* XXX This will mask info.rti_info[RTAX_DST] with
702 * info.rti_info[RTAX_NETMASK] before
703 * searching. It did not used to do that. --dyoung
704 */
705 rt = NULL;
706 error = rtrequest1(RTM_GET, &info, &rt);
707 if (error != 0)
708 senderr(error);
709 if (rtm->rtm_type != RTM_GET) {/* XXX: too grotty */
710 if (memcmp(info.rti_info[RTAX_DST], rt_getkey(rt),
711 info.rti_info[RTAX_DST]->sa_len) != 0)
712 senderr(ESRCH);
713 if (info.rti_info[RTAX_NETMASK] == NULL &&
714 rt_mask(rt) != NULL)
715 senderr(ETOOMANYREFS);
716 }
717
718 /*
719 * XXX if arp/ndp requests an L2 entry, we have to obtain
720 * it from lltable while for the route command we have to
721 * return a route as it is. How to distinguish them?
722 * For newer arp/ndp, RTF_LLDATA flag set by arp/ndp
723 * indicates an L2 entry is requested. For old arp/ndp
724 * binaries, we check RTF_UP flag is NOT set; it works
725 * by the fact that arp/ndp don't set it while the route
726 * command sets it.
727 */
728 if (((rtm->rtm_flags & RTF_LLDATA) != 0 ||
729 (rtm->rtm_flags & RTF_UP) == 0) &&
730 rtm->rtm_type == RTM_GET &&
731 sockaddr_cmp(rt_getkey(rt), info.rti_info[RTAX_DST]) != 0) {
732 int ll_flags = 0;
733 route_get_sdl(rt->rt_ifp, info.rti_info[RTAX_DST], &sdl,
734 &ll_flags);
735 info.rti_info[RTAX_GATEWAY] = sstocsa(&sdl);
736 error = route_output_report(rt, &info, rtm, &new_rtm);
737 if (error)
738 senderr(error);
739 if (new_rtm != NULL) {
740 old_rtm = rtm;
741 rtm = new_rtm;
742 }
743 rtm->rtm_flags |= RTF_LLDATA;
744 rtm->rtm_flags |= (ll_flags & LLE_STATIC) ? RTF_STATIC : 0;
745 break;
746 }
747
748 switch (rtm->rtm_type) {
749 case RTM_GET:
750 info.rti_info[RTAX_DST] = rt_getkey(rt);
751 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
752 info.rti_info[RTAX_NETMASK] = rt_mask(rt);
753 info.rti_info[RTAX_TAG] = rt_gettag(rt);
754 error = route_output_report(rt, &info, rtm, &new_rtm);
755 if (error)
756 senderr(error);
757 if (new_rtm != NULL) {
758 old_rtm = rtm;
759 rtm = new_rtm;
760 }
761 break;
762
763 case RTM_CHANGE:
764 /*
765 * new gateway could require new ifaddr, ifp;
766 * flags may also be different; ifp may be specified
767 * by ll sockaddr when protocol address is ambiguous
768 */
769 if ((error = rt_getifa(&info)) != 0)
770 senderr(error);
771 if (info.rti_info[RTAX_GATEWAY]) {
772 error = rt_setgate(rt,
773 info.rti_info[RTAX_GATEWAY]);
774 if (error != 0)
775 senderr(error);
776 }
777 if (info.rti_info[RTAX_TAG]) {
778 const struct sockaddr *tag;
779 tag = rt_settag(rt, info.rti_info[RTAX_TAG]);
780 if (tag == NULL)
781 senderr(ENOBUFS);
782 }
783 /* new gateway could require new ifaddr, ifp;
784 flags may also be different; ifp may be specified
785 by ll sockaddr when protocol address is ambiguous */
786 if (info.rti_info[RTAX_IFP] &&
787 (ifa = ifa_ifwithnet(info.rti_info[RTAX_IFP])) &&
788 (ifp = ifa->ifa_ifp) && (info.rti_info[RTAX_IFA] ||
789 info.rti_info[RTAX_GATEWAY])) {
790 if (info.rti_info[RTAX_IFA] == NULL ||
791 (ifa = ifa_ifwithaddr(
792 info.rti_info[RTAX_IFA])) == NULL)
793 ifa = ifaof_ifpforaddr(
794 info.rti_info[RTAX_IFA] ?
795 info.rti_info[RTAX_IFA] :
796 info.rti_info[RTAX_GATEWAY], ifp);
797 } else if ((info.rti_info[RTAX_IFA] &&
798 (ifa = ifa_ifwithaddr(info.rti_info[RTAX_IFA]))) ||
799 (info.rti_info[RTAX_GATEWAY] &&
800 (ifa = ifa_ifwithroute(rt->rt_flags,
801 rt_getkey(rt), info.rti_info[RTAX_GATEWAY])))) {
802 ifp = ifa->ifa_ifp;
803 }
804 if (ifa) {
805 struct ifaddr *oifa = rt->rt_ifa;
806 if (oifa != ifa) {
807 if (oifa && oifa->ifa_rtrequest) {
808 oifa->ifa_rtrequest(RTM_DELETE,
809 rt, &info);
810 }
811 rt_replace_ifa(rt, ifa);
812 rt->rt_ifp = ifp;
813 }
814 }
815 if (ifp && rt->rt_ifp != ifp)
816 rt->rt_ifp = ifp;
817 rt_setmetrics(rtm->rtm_inits, rtm, rt);
818 if (rt->rt_flags != info.rti_flags)
819 rt->rt_flags = (info.rti_flags & ~PRESERVED_RTF)
820 | (rt->rt_flags & PRESERVED_RTF);
821 if (rt->rt_ifa && rt->rt_ifa->ifa_rtrequest)
822 rt->rt_ifa->ifa_rtrequest(RTM_ADD, rt, &info);
823 /*FALLTHROUGH*/
824 case RTM_LOCK:
825 rt->rt_rmx.rmx_locks &= ~(rtm->rtm_inits);
826 rt->rt_rmx.rmx_locks |=
827 (rtm->rtm_inits & rtm->rtm_rmx.rmx_locks);
828 break;
829 }
830 break;
831
832 default:
833 senderr(EOPNOTSUPP);
834 }
835
836 flush:
837 if (rtm) {
838 if (error)
839 rtm->rtm_errno = error;
840 else
841 rtm->rtm_flags |= RTF_DONE;
842 }
843 family = info.rti_info[RTAX_DST] ? info.rti_info[RTAX_DST]->sa_family :
844 0;
845 /* We cannot free old_rtm until we have stopped using the
846 * pointers in info, some of which may point to sockaddrs
847 * in old_rtm.
848 */
849 if (old_rtm != NULL)
850 Free(old_rtm);
851 if (rt)
852 rtfree(rt);
853 {
854 struct rawcb *rp = NULL;
855 /*
856 * Check to see if we don't want our own messages.
857 */
858 if ((so->so_options & SO_USELOOPBACK) == 0) {
859 if (COMPATNAME(route_info).ri_cb.any_count <= 1) {
860 if (rtm)
861 Free(rtm);
862 m_freem(m);
863 return error;
864 }
865 /* There is another listener, so construct message */
866 rp = sotorawcb(so);
867 }
868 if (rtm) {
869 m_copyback(m, 0, rtm->rtm_msglen, rtm);
870 if (m->m_pkthdr.len < rtm->rtm_msglen) {
871 m_freem(m);
872 m = NULL;
873 } else if (m->m_pkthdr.len > rtm->rtm_msglen)
874 m_adj(m, rtm->rtm_msglen - m->m_pkthdr.len);
875 Free(rtm);
876 }
877 if (rp)
878 rp->rcb_proto.sp_family = 0; /* Avoid us */
879 if (family)
880 proto.sp_protocol = family;
881 if (m)
882 raw_input(m, &proto, &COMPATNAME(route_info).ri_src,
883 &COMPATNAME(route_info).ri_dst);
884 if (rp)
885 rp->rcb_proto.sp_family = PF_XROUTE;
886 }
887 return error;
888 }
889
890 static void
891 rt_setmetrics(int which, const struct rt_xmsghdr *in, struct rtentry *out)
892 {
893 #define metric(f, e) if (which & (f)) out->rt_rmx.e = in->rtm_rmx.e;
894 metric(RTV_RPIPE, rmx_recvpipe);
895 metric(RTV_SPIPE, rmx_sendpipe);
896 metric(RTV_SSTHRESH, rmx_ssthresh);
897 metric(RTV_RTT, rmx_rtt);
898 metric(RTV_RTTVAR, rmx_rttvar);
899 metric(RTV_HOPCOUNT, rmx_hopcount);
900 metric(RTV_MTU, rmx_mtu);
901 #undef metric
902 if (which & RTV_EXPIRE) {
903 out->rt_rmx.rmx_expire = in->rtm_rmx.rmx_expire ?
904 time_wall_to_mono(in->rtm_rmx.rmx_expire) : 0;
905 }
906 }
907
908 static void
909 rtm_setmetrics(const struct rtentry *in, struct rt_xmsghdr *out)
910 {
911 #define metric(e) out->rtm_rmx.e = in->rt_rmx.e;
912 metric(rmx_recvpipe);
913 metric(rmx_sendpipe);
914 metric(rmx_ssthresh);
915 metric(rmx_rtt);
916 metric(rmx_rttvar);
917 metric(rmx_hopcount);
918 metric(rmx_mtu);
919 #undef metric
920 out->rtm_rmx.rmx_expire = in->rt_rmx.rmx_expire ?
921 time_mono_to_wall(in->rt_rmx.rmx_expire) : 0;
922 }
923
924 static int
925 rt_xaddrs(u_char rtmtype, const char *cp, const char *cplim,
926 struct rt_addrinfo *rtinfo)
927 {
928 const struct sockaddr *sa = NULL; /* Quell compiler warning */
929 int i;
930
931 for (i = 0; i < RTAX_MAX && cp < cplim; i++) {
932 if ((rtinfo->rti_addrs & (1 << i)) == 0)
933 continue;
934 rtinfo->rti_info[i] = sa = (const struct sockaddr *)cp;
935 RT_XADVANCE(cp, sa);
936 }
937
938 /*
939 * Check for extra addresses specified, except RTM_GET asking
940 * for interface info.
941 */
942 if (rtmtype == RTM_GET) {
943 if (((rtinfo->rti_addrs &
944 (~((1 << RTAX_IFP) | (1 << RTAX_IFA)))) & (~0 << i)) != 0)
945 return 1;
946 } else if ((rtinfo->rti_addrs & (~0 << i)) != 0)
947 return 1;
948 /* Check for bad data length. */
949 if (cp != cplim) {
950 if (i == RTAX_NETMASK + 1 && sa != NULL &&
951 cp - RT_XROUNDUP(sa->sa_len) + sa->sa_len == cplim)
952 /*
953 * The last sockaddr was info.rti_info[RTAX_NETMASK].
954 * We accept this for now for the sake of old
955 * binaries or third party softwares.
956 */
957 ;
958 else
959 return 1;
960 }
961 return 0;
962 }
963
964 static int
965 rt_getlen(int type)
966 {
967 #ifndef COMPAT_RTSOCK
968 CTASSERT(__alignof(struct ifa_msghdr) >= sizeof(uint64_t));
969 CTASSERT(__alignof(struct if_msghdr) >= sizeof(uint64_t));
970 CTASSERT(__alignof(struct if_announcemsghdr) >= sizeof(uint64_t));
971 CTASSERT(__alignof(struct rt_msghdr) >= sizeof(uint64_t));
972 #endif
973
974 switch (type) {
975 case RTM_DELADDR:
976 case RTM_NEWADDR:
977 case RTM_CHGADDR:
978 return sizeof(struct ifa_xmsghdr);
979
980 case RTM_OOIFINFO:
981 #ifdef COMPAT_14
982 return sizeof(struct if_msghdr14);
983 #else
984 #ifdef DIAGNOSTIC
985 printf("RTM_OOIFINFO\n");
986 #endif
987 return -1;
988 #endif
989 case RTM_OIFINFO:
990 #ifdef COMPAT_50
991 return sizeof(struct if_msghdr50);
992 #else
993 #ifdef DIAGNOSTIC
994 printf("RTM_OIFINFO\n");
995 #endif
996 return -1;
997 #endif
998
999 case RTM_IFINFO:
1000 return sizeof(struct if_xmsghdr);
1001
1002 case RTM_IFANNOUNCE:
1003 case RTM_IEEE80211:
1004 return sizeof(struct if_xannouncemsghdr);
1005
1006 default:
1007 return sizeof(struct rt_xmsghdr);
1008 }
1009 }
1010
1011
1012 struct mbuf *
1013 COMPATNAME(rt_msg1)(int type, struct rt_addrinfo *rtinfo, void *data, int datalen)
1014 {
1015 struct rt_xmsghdr *rtm;
1016 struct mbuf *m;
1017 int i;
1018 const struct sockaddr *sa;
1019 int len, dlen;
1020
1021 m = m_gethdr(M_DONTWAIT, MT_DATA);
1022 if (m == NULL)
1023 return m;
1024 MCLAIM(m, &COMPATNAME(routedomain).dom_mowner);
1025
1026 if ((len = rt_getlen(type)) == -1)
1027 goto out;
1028 if (len > MHLEN + MLEN)
1029 panic("%s: message too long", __func__);
1030 else if (len > MHLEN) {
1031 m->m_next = m_get(M_DONTWAIT, MT_DATA);
1032 if (m->m_next == NULL)
1033 goto out;
1034 MCLAIM(m->m_next, m->m_owner);
1035 m->m_pkthdr.len = len;
1036 m->m_len = MHLEN;
1037 m->m_next->m_len = len - MHLEN;
1038 } else {
1039 m->m_pkthdr.len = m->m_len = len;
1040 }
1041 m_reset_rcvif(m);
1042 m_copyback(m, 0, datalen, data);
1043 if (len > datalen)
1044 (void)memset(mtod(m, char *) + datalen, 0, len - datalen);
1045 rtm = mtod(m, struct rt_xmsghdr *);
1046 for (i = 0; i < RTAX_MAX; i++) {
1047 if ((sa = rtinfo->rti_info[i]) == NULL)
1048 continue;
1049 rtinfo->rti_addrs |= (1 << i);
1050 dlen = RT_XROUNDUP(sa->sa_len);
1051 m_copyback(m, len, sa->sa_len, sa);
1052 if (dlen != sa->sa_len) {
1053 /*
1054 * Up to 6 + 1 nul's since roundup is to
1055 * sizeof(uint64_t) (8 bytes)
1056 */
1057 m_copyback(m, len + sa->sa_len,
1058 dlen - sa->sa_len, "\0\0\0\0\0\0");
1059 }
1060 len += dlen;
1061 }
1062 if (m->m_pkthdr.len != len)
1063 goto out;
1064 rtm->rtm_msglen = len;
1065 rtm->rtm_version = RTM_XVERSION;
1066 rtm->rtm_type = type;
1067 return m;
1068 out:
1069 m_freem(m);
1070 return NULL;
1071 }
1072
1073 /*
1074 * rt_msg2
1075 *
1076 * fills 'cp' or 'w'.w_tmem with the routing socket message and
1077 * returns the length of the message in 'lenp'.
1078 *
1079 * if walkarg is 0, cp is expected to be 0 or a buffer large enough to hold
1080 * the message
1081 * otherwise walkarg's w_needed is updated and if the user buffer is
1082 * specified and w_needed indicates space exists the information is copied
1083 * into the temp space (w_tmem). w_tmem is [re]allocated if necessary,
1084 * if the allocation fails ENOBUFS is returned.
1085 */
1086 static int
1087 rt_msg2(int type, struct rt_addrinfo *rtinfo, void *cpv, struct rt_walkarg *w,
1088 int *lenp)
1089 {
1090 int i;
1091 int len, dlen, second_time = 0;
1092 char *cp0, *cp = cpv;
1093
1094 rtinfo->rti_addrs = 0;
1095 again:
1096 if ((len = rt_getlen(type)) == -1)
1097 return EINVAL;
1098
1099 if ((cp0 = cp) != NULL)
1100 cp += len;
1101 for (i = 0; i < RTAX_MAX; i++) {
1102 const struct sockaddr *sa;
1103
1104 if ((sa = rtinfo->rti_info[i]) == NULL)
1105 continue;
1106 rtinfo->rti_addrs |= (1 << i);
1107 dlen = RT_XROUNDUP(sa->sa_len);
1108 if (cp) {
1109 int diff = dlen - sa->sa_len;
1110 (void)memcpy(cp, sa, (size_t)sa->sa_len);
1111 cp += sa->sa_len;
1112 if (diff > 0) {
1113 (void)memset(cp, 0, (size_t)diff);
1114 cp += diff;
1115 }
1116 }
1117 len += dlen;
1118 }
1119 if (cp == NULL && w != NULL && !second_time) {
1120 struct rt_walkarg *rw = w;
1121
1122 rw->w_needed += len;
1123 if (rw->w_needed <= 0 && rw->w_where) {
1124 if (rw->w_tmemsize < len) {
1125 if (rw->w_tmem)
1126 free(rw->w_tmem, M_RTABLE);
1127 rw->w_tmem = malloc(len, M_RTABLE, M_NOWAIT);
1128 if (rw->w_tmem)
1129 rw->w_tmemsize = len;
1130 else
1131 rw->w_tmemsize = 0;
1132 }
1133 if (rw->w_tmem) {
1134 cp = rw->w_tmem;
1135 second_time = 1;
1136 goto again;
1137 } else {
1138 rw->w_tmemneeded = len;
1139 return ENOBUFS;
1140 }
1141 }
1142 }
1143 if (cp) {
1144 struct rt_xmsghdr *rtm = (struct rt_xmsghdr *)cp0;
1145
1146 rtm->rtm_version = RTM_XVERSION;
1147 rtm->rtm_type = type;
1148 rtm->rtm_msglen = len;
1149 }
1150 if (lenp)
1151 *lenp = len;
1152 return 0;
1153 }
1154
1155 #ifndef COMPAT_RTSOCK
1156 int
1157 rt_msg3(int type, struct rt_addrinfo *rtinfo, void *cpv, struct rt_walkarg *w,
1158 int *lenp)
1159 {
1160 return rt_msg2(type, rtinfo, cpv, w, lenp);
1161 }
1162 #endif
1163
1164 /*
1165 * This routine is called to generate a message from the routing
1166 * socket indicating that a redirect has occurred, a routing lookup
1167 * has failed, or that a protocol has detected timeouts to a particular
1168 * destination.
1169 */
1170 void
1171 COMPATNAME(rt_missmsg)(int type, const struct rt_addrinfo *rtinfo, int flags,
1172 int error)
1173 {
1174 struct rt_xmsghdr rtm;
1175 struct mbuf *m;
1176 const struct sockaddr *sa = rtinfo->rti_info[RTAX_DST];
1177 struct rt_addrinfo info = *rtinfo;
1178
1179 COMPATCALL(rt_missmsg, (type, rtinfo, flags, error));
1180 if (COMPATNAME(route_info).ri_cb.any_count == 0)
1181 return;
1182 memset(&rtm, 0, sizeof(rtm));
1183 rtm.rtm_pid = curproc->p_pid;
1184 rtm.rtm_flags = RTF_DONE | flags;
1185 rtm.rtm_errno = error;
1186 m = COMPATNAME(rt_msg1)(type, &info, &rtm, sizeof(rtm));
1187 if (m == NULL)
1188 return;
1189 mtod(m, struct rt_xmsghdr *)->rtm_addrs = info.rti_addrs;
1190 COMPATNAME(route_enqueue)(m, sa ? sa->sa_family : 0);
1191 }
1192
1193 /*
1194 * This routine is called to generate a message from the routing
1195 * socket indicating that the status of a network interface has changed.
1196 */
1197 void
1198 COMPATNAME(rt_ifmsg)(struct ifnet *ifp)
1199 {
1200 struct if_xmsghdr ifm;
1201 struct mbuf *m;
1202 struct rt_addrinfo info;
1203
1204 COMPATCALL(rt_ifmsg, (ifp));
1205 if (COMPATNAME(route_info).ri_cb.any_count == 0)
1206 return;
1207 (void)memset(&info, 0, sizeof(info));
1208 (void)memset(&ifm, 0, sizeof(ifm));
1209 ifm.ifm_index = ifp->if_index;
1210 ifm.ifm_flags = ifp->if_flags;
1211 ifm.ifm_data = ifp->if_data;
1212 ifm.ifm_addrs = 0;
1213 m = COMPATNAME(rt_msg1)(RTM_IFINFO, &info, &ifm, sizeof(ifm));
1214 if (m == NULL)
1215 return;
1216 COMPATNAME(route_enqueue)(m, 0);
1217 #ifdef COMPAT_14
1218 compat_14_rt_oifmsg(ifp);
1219 #endif
1220 #ifdef COMPAT_50
1221 compat_50_rt_oifmsg(ifp);
1222 #endif
1223 }
1224
1225
1226 /*
1227 * This is called to generate messages from the routing socket
1228 * indicating a network interface has had addresses associated with it.
1229 * if we ever reverse the logic and replace messages TO the routing
1230 * socket indicate a request to configure interfaces, then it will
1231 * be unnecessary as the routing socket will automatically generate
1232 * copies of it.
1233 */
1234 void
1235 COMPATNAME(rt_newaddrmsg)(int cmd, struct ifaddr *ifa, int error,
1236 struct rtentry *rt)
1237 {
1238 #define cmdpass(__cmd, __pass) (((__cmd) << 2) | (__pass))
1239 struct rt_addrinfo info;
1240 const struct sockaddr *sa;
1241 int pass;
1242 struct mbuf *m;
1243 struct ifnet *ifp;
1244 struct rt_xmsghdr rtm;
1245 struct ifa_xmsghdr ifam;
1246 int ncmd;
1247
1248 KASSERT(ifa != NULL);
1249 ifp = ifa->ifa_ifp;
1250 #ifdef SCTP
1251 if (cmd == RTM_ADD) {
1252 sctp_add_ip_address(ifa);
1253 } else if (cmd == RTM_DELETE) {
1254 sctp_delete_ip_address(ifa);
1255 }
1256 #endif
1257
1258 COMPATCALL(rt_newaddrmsg, (cmd, ifa, error, rt));
1259 if (COMPATNAME(route_info).ri_cb.any_count == 0)
1260 return;
1261 for (pass = 1; pass < 3; pass++) {
1262 memset(&info, 0, sizeof(info));
1263 switch (cmdpass(cmd, pass)) {
1264 case cmdpass(RTM_ADD, 1):
1265 case cmdpass(RTM_CHANGE, 1):
1266 case cmdpass(RTM_DELETE, 2):
1267 case cmdpass(RTM_NEWADDR, 1):
1268 case cmdpass(RTM_DELADDR, 1):
1269 case cmdpass(RTM_CHGADDR, 1):
1270 switch (cmd) {
1271 case RTM_ADD:
1272 ncmd = RTM_NEWADDR;
1273 break;
1274 case RTM_DELETE:
1275 ncmd = RTM_DELADDR;
1276 break;
1277 case RTM_CHANGE:
1278 ncmd = RTM_CHGADDR;
1279 break;
1280 default:
1281 ncmd = cmd;
1282 }
1283 info.rti_info[RTAX_IFA] = sa = ifa->ifa_addr;
1284 KASSERT(ifp->if_dl != NULL);
1285 info.rti_info[RTAX_IFP] = ifp->if_dl->ifa_addr;
1286 info.rti_info[RTAX_NETMASK] = ifa->ifa_netmask;
1287 info.rti_info[RTAX_BRD] = ifa->ifa_dstaddr;
1288 memset(&ifam, 0, sizeof(ifam));
1289 ifam.ifam_index = ifp->if_index;
1290 ifam.ifam_metric = ifa->ifa_metric;
1291 ifam.ifam_flags = ifa->ifa_flags;
1292 m = COMPATNAME(rt_msg1)(ncmd, &info, &ifam, sizeof(ifam));
1293 if (m == NULL)
1294 continue;
1295 mtod(m, struct ifa_xmsghdr *)->ifam_addrs =
1296 info.rti_addrs;
1297 break;
1298 case cmdpass(RTM_ADD, 2):
1299 case cmdpass(RTM_CHANGE, 2):
1300 case cmdpass(RTM_DELETE, 1):
1301 if (rt == NULL)
1302 continue;
1303 info.rti_info[RTAX_NETMASK] = rt_mask(rt);
1304 info.rti_info[RTAX_DST] = sa = rt_getkey(rt);
1305 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
1306 memset(&rtm, 0, sizeof(rtm));
1307 rtm.rtm_pid = curproc->p_pid;
1308 rtm.rtm_index = ifp->if_index;
1309 rtm.rtm_flags |= rt->rt_flags;
1310 rtm.rtm_errno = error;
1311 m = COMPATNAME(rt_msg1)(cmd, &info, &rtm, sizeof(rtm));
1312 if (m == NULL)
1313 continue;
1314 mtod(m, struct rt_xmsghdr *)->rtm_addrs = info.rti_addrs;
1315 break;
1316 default:
1317 continue;
1318 }
1319 #ifdef DIAGNOSTIC
1320 if (m == NULL)
1321 panic("%s: called with wrong command", __func__);
1322 #endif
1323 COMPATNAME(route_enqueue)(m, sa ? sa->sa_family : 0);
1324 }
1325 #undef cmdpass
1326 }
1327
1328 static struct mbuf *
1329 rt_makeifannouncemsg(struct ifnet *ifp, int type, int what,
1330 struct rt_addrinfo *info)
1331 {
1332 struct if_xannouncemsghdr ifan;
1333
1334 memset(info, 0, sizeof(*info));
1335 memset(&ifan, 0, sizeof(ifan));
1336 ifan.ifan_index = ifp->if_index;
1337 strlcpy(ifan.ifan_name, ifp->if_xname, sizeof(ifan.ifan_name));
1338 ifan.ifan_what = what;
1339 return COMPATNAME(rt_msg1)(type, info, &ifan, sizeof(ifan));
1340 }
1341
1342 /*
1343 * This is called to generate routing socket messages indicating
1344 * network interface arrival and departure.
1345 */
1346 void
1347 COMPATNAME(rt_ifannouncemsg)(struct ifnet *ifp, int what)
1348 {
1349 struct mbuf *m;
1350 struct rt_addrinfo info;
1351
1352 COMPATCALL(rt_ifannouncemsg, (ifp, what));
1353 if (COMPATNAME(route_info).ri_cb.any_count == 0)
1354 return;
1355 m = rt_makeifannouncemsg(ifp, RTM_IFANNOUNCE, what, &info);
1356 if (m == NULL)
1357 return;
1358 COMPATNAME(route_enqueue)(m, 0);
1359 }
1360
1361 /*
1362 * This is called to generate routing socket messages indicating
1363 * IEEE80211 wireless events.
1364 * XXX we piggyback on the RTM_IFANNOUNCE msg format in a clumsy way.
1365 */
1366 void
1367 COMPATNAME(rt_ieee80211msg)(struct ifnet *ifp, int what, void *data,
1368 size_t data_len)
1369 {
1370 struct mbuf *m;
1371 struct rt_addrinfo info;
1372
1373 COMPATCALL(rt_ieee80211msg, (ifp, what, data, data_len));
1374 if (COMPATNAME(route_info).ri_cb.any_count == 0)
1375 return;
1376 m = rt_makeifannouncemsg(ifp, RTM_IEEE80211, what, &info);
1377 if (m == NULL)
1378 return;
1379 /*
1380 * Append the ieee80211 data. Try to stick it in the
1381 * mbuf containing the ifannounce msg; otherwise allocate
1382 * a new mbuf and append.
1383 *
1384 * NB: we assume m is a single mbuf.
1385 */
1386 if (data_len > M_TRAILINGSPACE(m)) {
1387 struct mbuf *n = m_get(M_NOWAIT, MT_DATA);
1388 if (n == NULL) {
1389 m_freem(m);
1390 return;
1391 }
1392 (void)memcpy(mtod(n, void *), data, data_len);
1393 n->m_len = data_len;
1394 m->m_next = n;
1395 } else if (data_len > 0) {
1396 (void)memcpy(mtod(m, uint8_t *) + m->m_len, data, data_len);
1397 m->m_len += data_len;
1398 }
1399 if (m->m_flags & M_PKTHDR)
1400 m->m_pkthdr.len += data_len;
1401 mtod(m, struct if_xannouncemsghdr *)->ifan_msglen += data_len;
1402 COMPATNAME(route_enqueue)(m, 0);
1403 }
1404
1405 /*
1406 * This is used in dumping the kernel table via sysctl().
1407 */
1408 static int
1409 sysctl_dumpentry(struct rtentry *rt, void *v)
1410 {
1411 struct rt_walkarg *w = v;
1412 int error = 0, size;
1413 struct rt_addrinfo info;
1414
1415 if (w->w_op == NET_RT_FLAGS && !(rt->rt_flags & w->w_arg))
1416 return 0;
1417 memset(&info, 0, sizeof(info));
1418 info.rti_info[RTAX_DST] = rt_getkey(rt);
1419 info.rti_info[RTAX_GATEWAY] = rt->rt_gateway;
1420 info.rti_info[RTAX_NETMASK] = rt_mask(rt);
1421 info.rti_info[RTAX_TAG] = rt_gettag(rt);
1422 if (rt->rt_ifp) {
1423 const struct ifaddr *rtifa;
1424 info.rti_info[RTAX_IFP] = rt->rt_ifp->if_dl->ifa_addr;
1425 /* rtifa used to be simply rt->rt_ifa. If rt->rt_ifa != NULL,
1426 * then rt_get_ifa() != NULL. So this ought to still be safe.
1427 * --dyoung
1428 */
1429 rtifa = rt_get_ifa(rt);
1430 info.rti_info[RTAX_IFA] = rtifa->ifa_addr;
1431 if (rt->rt_ifp->if_flags & IFF_POINTOPOINT)
1432 info.rti_info[RTAX_BRD] = rtifa->ifa_dstaddr;
1433 }
1434 if ((error = rt_msg2(RTM_GET, &info, 0, w, &size)))
1435 return error;
1436 if (w->w_where && w->w_tmem && w->w_needed <= 0) {
1437 struct rt_xmsghdr *rtm = (struct rt_xmsghdr *)w->w_tmem;
1438
1439 rtm->rtm_flags = rt->rt_flags;
1440 rtm->rtm_use = rt->rt_use;
1441 rtm_setmetrics(rt, rtm);
1442 KASSERT(rt->rt_ifp != NULL);
1443 rtm->rtm_index = rt->rt_ifp->if_index;
1444 rtm->rtm_errno = rtm->rtm_pid = rtm->rtm_seq = 0;
1445 rtm->rtm_addrs = info.rti_addrs;
1446 if ((error = copyout(rtm, w->w_where, size)) != 0)
1447 w->w_where = NULL;
1448 else
1449 w->w_where = (char *)w->w_where + size;
1450 }
1451 return error;
1452 }
1453
1454 static int
1455 sysctl_iflist(int af, struct rt_walkarg *w, int type)
1456 {
1457 struct ifnet *ifp;
1458 struct ifaddr *ifa;
1459 struct rt_addrinfo info;
1460 int len, error = 0;
1461 int s;
1462 struct psref psref;
1463 int bound = curlwp_bind();
1464
1465 memset(&info, 0, sizeof(info));
1466
1467 s = pserialize_read_enter();
1468 IFNET_READER_FOREACH(ifp) {
1469 if (w->w_arg && w->w_arg != ifp->if_index)
1470 continue;
1471 if (IFADDR_EMPTY(ifp))
1472 continue;
1473
1474 psref_acquire(&psref, &ifp->if_psref, ifnet_psref_class);
1475 pserialize_read_exit(s);
1476
1477 info.rti_info[RTAX_IFP] = ifp->if_dl->ifa_addr;
1478 switch (type) {
1479 case NET_RT_IFLIST:
1480 error = rt_msg2(RTM_IFINFO, &info, NULL, w, &len);
1481 break;
1482 #ifdef COMPAT_14
1483 case NET_RT_OOIFLIST:
1484 error = rt_msg2(RTM_OOIFINFO, &info, NULL, w, &len);
1485 break;
1486 #endif
1487 #ifdef COMPAT_50
1488 case NET_RT_OIFLIST:
1489 error = rt_msg2(RTM_OIFINFO, &info, NULL, w, &len);
1490 break;
1491 #endif
1492 default:
1493 panic("sysctl_iflist(1)");
1494 }
1495 if (error)
1496 goto release_exit;
1497 info.rti_info[RTAX_IFP] = NULL;
1498 if (w->w_where && w->w_tmem && w->w_needed <= 0) {
1499 switch (type) {
1500 case NET_RT_IFLIST: {
1501 struct if_xmsghdr *ifm;
1502
1503 ifm = (struct if_xmsghdr *)w->w_tmem;
1504 ifm->ifm_index = ifp->if_index;
1505 ifm->ifm_flags = ifp->if_flags;
1506 ifm->ifm_data = ifp->if_data;
1507 ifm->ifm_addrs = info.rti_addrs;
1508 error = copyout(ifm, w->w_where, len);
1509 if (error)
1510 goto release_exit;
1511 w->w_where = (char *)w->w_where + len;
1512 break;
1513 }
1514
1515 #ifdef COMPAT_14
1516 case NET_RT_OOIFLIST:
1517 error = compat_14_iflist(ifp, w, &info, len);
1518 if (error)
1519 goto release_exit;
1520 break;
1521 #endif
1522 #ifdef COMPAT_50
1523 case NET_RT_OIFLIST:
1524 error = compat_50_iflist(ifp, w, &info, len);
1525 if (error)
1526 goto release_exit;
1527 break;
1528 #endif
1529 default:
1530 panic("sysctl_iflist(2)");
1531 }
1532 }
1533 IFADDR_FOREACH(ifa, ifp) {
1534 if (af && af != ifa->ifa_addr->sa_family)
1535 continue;
1536 info.rti_info[RTAX_IFA] = ifa->ifa_addr;
1537 info.rti_info[RTAX_NETMASK] = ifa->ifa_netmask;
1538 info.rti_info[RTAX_BRD] = ifa->ifa_dstaddr;
1539 if ((error = rt_msg2(RTM_NEWADDR, &info, 0, w, &len)))
1540 goto release_exit;
1541 if (w->w_where && w->w_tmem && w->w_needed <= 0) {
1542 struct ifa_xmsghdr *ifam;
1543
1544 ifam = (struct ifa_xmsghdr *)w->w_tmem;
1545 ifam->ifam_index = ifa->ifa_ifp->if_index;
1546 ifam->ifam_flags = ifa->ifa_flags;
1547 ifam->ifam_metric = ifa->ifa_metric;
1548 ifam->ifam_addrs = info.rti_addrs;
1549 error = copyout(w->w_tmem, w->w_where, len);
1550 if (error)
1551 goto release_exit;
1552 w->w_where = (char *)w->w_where + len;
1553 }
1554 }
1555 info.rti_info[RTAX_IFA] = info.rti_info[RTAX_NETMASK] =
1556 info.rti_info[RTAX_BRD] = NULL;
1557
1558 s = pserialize_read_enter();
1559 psref_release(&psref, &ifp->if_psref, ifnet_psref_class);
1560 }
1561 pserialize_read_exit(s);
1562 curlwp_bindx(bound);
1563
1564 return 0;
1565
1566 release_exit:
1567 psref_release(&psref, &ifp->if_psref, ifnet_psref_class);
1568 curlwp_bindx(bound);
1569 return error;
1570 }
1571
1572 static int
1573 sysctl_rtable(SYSCTLFN_ARGS)
1574 {
1575 void *where = oldp;
1576 size_t *given = oldlenp;
1577 int i, s, error = EINVAL;
1578 u_char af;
1579 struct rt_walkarg w;
1580
1581 if (namelen == 1 && name[0] == CTL_QUERY)
1582 return sysctl_query(SYSCTLFN_CALL(rnode));
1583
1584 if (newp)
1585 return EPERM;
1586 if (namelen != 3)
1587 return EINVAL;
1588 af = name[0];
1589 w.w_tmemneeded = 0;
1590 w.w_tmemsize = 0;
1591 w.w_tmem = NULL;
1592 again:
1593 /* we may return here if a later [re]alloc of the t_mem buffer fails */
1594 if (w.w_tmemneeded) {
1595 w.w_tmem = malloc(w.w_tmemneeded, M_RTABLE, M_WAITOK);
1596 w.w_tmemsize = w.w_tmemneeded;
1597 w.w_tmemneeded = 0;
1598 }
1599 w.w_op = name[1];
1600 w.w_arg = name[2];
1601 w.w_given = *given;
1602 w.w_needed = 0 - w.w_given;
1603 w.w_where = where;
1604
1605 s = splsoftnet();
1606 switch (w.w_op) {
1607
1608 case NET_RT_DUMP:
1609 case NET_RT_FLAGS:
1610 #ifdef INET
1611 /*
1612 * take care of llinfo entries, the caller must
1613 * specify an AF
1614 */
1615 if (w.w_op == NET_RT_FLAGS &&
1616 (w.w_arg == 0 || w.w_arg & RTF_LLDATA)) {
1617 if (af != 0)
1618 error = lltable_sysctl_dumparp(af, &w);
1619 else
1620 error = EINVAL;
1621 break;
1622 }
1623 #endif /* INET */
1624
1625 for (i = 1; i <= AF_MAX; i++)
1626 if ((af == 0 || af == i) &&
1627 (error = rt_walktree(i, sysctl_dumpentry, &w)))
1628 break;
1629 break;
1630
1631 #ifdef COMPAT_14
1632 case NET_RT_OOIFLIST:
1633 error = sysctl_iflist(af, &w, w.w_op);
1634 break;
1635 #endif
1636 #ifdef COMPAT_50
1637 case NET_RT_OIFLIST:
1638 error = sysctl_iflist(af, &w, w.w_op);
1639 break;
1640 #endif
1641 case NET_RT_IFLIST:
1642 error = sysctl_iflist(af, &w, w.w_op);
1643 break;
1644 }
1645 splx(s);
1646
1647 /* check to see if we couldn't allocate memory with NOWAIT */
1648 if (error == ENOBUFS && w.w_tmem == 0 && w.w_tmemneeded)
1649 goto again;
1650
1651 if (w.w_tmem)
1652 free(w.w_tmem, M_RTABLE);
1653 w.w_needed += w.w_given;
1654 if (where) {
1655 *given = (char *)w.w_where - (char *)where;
1656 if (*given < w.w_needed)
1657 return ENOMEM;
1658 } else {
1659 *given = (11 * w.w_needed) / 10;
1660 }
1661 return error;
1662 }
1663
1664 /*
1665 * Routing message software interrupt routine
1666 */
1667 static void
1668 COMPATNAME(route_intr)(void *cookie)
1669 {
1670 struct sockproto proto = { .sp_family = PF_XROUTE, };
1671 struct route_info * const ri = &COMPATNAME(route_info);
1672 struct mbuf *m;
1673 int s;
1674
1675 mutex_enter(softnet_lock);
1676 KERNEL_LOCK(1, NULL);
1677 while (!IF_IS_EMPTY(&ri->ri_intrq)) {
1678 s = splnet();
1679 IF_DEQUEUE(&ri->ri_intrq, m);
1680 splx(s);
1681 if (m == NULL)
1682 break;
1683 proto.sp_protocol = M_GETCTX(m, uintptr_t);
1684 raw_input(m, &proto, &ri->ri_src, &ri->ri_dst);
1685 }
1686 KERNEL_UNLOCK_ONE(NULL);
1687 mutex_exit(softnet_lock);
1688 }
1689
1690 /*
1691 * Enqueue a message to the software interrupt routine.
1692 */
1693 void
1694 COMPATNAME(route_enqueue)(struct mbuf *m, int family)
1695 {
1696 struct route_info * const ri = &COMPATNAME(route_info);
1697 int s, wasempty;
1698
1699 s = splnet();
1700 if (IF_QFULL(&ri->ri_intrq)) {
1701 IF_DROP(&ri->ri_intrq);
1702 m_freem(m);
1703 } else {
1704 wasempty = IF_IS_EMPTY(&ri->ri_intrq);
1705 M_SETCTX(m, (uintptr_t)family);
1706 IF_ENQUEUE(&ri->ri_intrq, m);
1707 if (wasempty)
1708 softint_schedule(ri->ri_sih);
1709 }
1710 splx(s);
1711 }
1712
1713 static void
1714 COMPATNAME(route_init)(void)
1715 {
1716 struct route_info * const ri = &COMPATNAME(route_info);
1717
1718 #ifndef COMPAT_RTSOCK
1719 rt_init();
1720 #endif
1721
1722 sysctl_net_route_setup(NULL);
1723 ri->ri_intrq.ifq_maxlen = ri->ri_maxqlen;
1724 ri->ri_sih = softint_establish(SOFTINT_NET | SOFTINT_MPSAFE,
1725 COMPATNAME(route_intr), NULL);
1726 }
1727
1728 /*
1729 * Definitions of protocols supported in the ROUTE domain.
1730 */
1731 #ifndef COMPAT_RTSOCK
1732 PR_WRAP_USRREQS(route);
1733 #else
1734 PR_WRAP_USRREQS(compat_50_route);
1735 #endif
1736
1737 static const struct pr_usrreqs route_usrreqs = {
1738 .pr_attach = COMPATNAME(route_attach_wrapper),
1739 .pr_detach = COMPATNAME(route_detach_wrapper),
1740 .pr_accept = COMPATNAME(route_accept_wrapper),
1741 .pr_bind = COMPATNAME(route_bind_wrapper),
1742 .pr_listen = COMPATNAME(route_listen_wrapper),
1743 .pr_connect = COMPATNAME(route_connect_wrapper),
1744 .pr_connect2 = COMPATNAME(route_connect2_wrapper),
1745 .pr_disconnect = COMPATNAME(route_disconnect_wrapper),
1746 .pr_shutdown = COMPATNAME(route_shutdown_wrapper),
1747 .pr_abort = COMPATNAME(route_abort_wrapper),
1748 .pr_ioctl = COMPATNAME(route_ioctl_wrapper),
1749 .pr_stat = COMPATNAME(route_stat_wrapper),
1750 .pr_peeraddr = COMPATNAME(route_peeraddr_wrapper),
1751 .pr_sockaddr = COMPATNAME(route_sockaddr_wrapper),
1752 .pr_rcvd = COMPATNAME(route_rcvd_wrapper),
1753 .pr_recvoob = COMPATNAME(route_recvoob_wrapper),
1754 .pr_send = COMPATNAME(route_send_wrapper),
1755 .pr_sendoob = COMPATNAME(route_sendoob_wrapper),
1756 .pr_purgeif = COMPATNAME(route_purgeif_wrapper),
1757 };
1758
1759 static const struct protosw COMPATNAME(route_protosw)[] = {
1760 {
1761 .pr_type = SOCK_RAW,
1762 .pr_domain = &COMPATNAME(routedomain),
1763 .pr_flags = PR_ATOMIC|PR_ADDR,
1764 .pr_input = raw_input,
1765 .pr_ctlinput = raw_ctlinput,
1766 .pr_usrreqs = &route_usrreqs,
1767 .pr_init = raw_init,
1768 },
1769 };
1770
1771 struct domain COMPATNAME(routedomain) = {
1772 .dom_family = PF_XROUTE,
1773 .dom_name = DOMAINNAME,
1774 .dom_init = COMPATNAME(route_init),
1775 .dom_protosw = COMPATNAME(route_protosw),
1776 .dom_protoswNPROTOSW =
1777 &COMPATNAME(route_protosw)[__arraycount(COMPATNAME(route_protosw))],
1778 };
1779
1780 static void
1781 sysctl_net_route_setup(struct sysctllog **clog)
1782 {
1783 const struct sysctlnode *rnode = NULL;
1784
1785 sysctl_createv(clog, 0, NULL, &rnode,
1786 CTLFLAG_PERMANENT,
1787 CTLTYPE_NODE, DOMAINNAME,
1788 SYSCTL_DESCR("PF_ROUTE information"),
1789 NULL, 0, NULL, 0,
1790 CTL_NET, PF_XROUTE, CTL_EOL);
1791
1792 sysctl_createv(clog, 0, NULL, NULL,
1793 CTLFLAG_PERMANENT,
1794 CTLTYPE_NODE, "rtable",
1795 SYSCTL_DESCR("Routing table information"),
1796 sysctl_rtable, 0, NULL, 0,
1797 CTL_NET, PF_XROUTE, 0 /* any protocol */, CTL_EOL);
1798
1799 sysctl_createv(clog, 0, &rnode, NULL,
1800 CTLFLAG_PERMANENT,
1801 CTLTYPE_STRUCT, "stats",
1802 SYSCTL_DESCR("Routing statistics"),
1803 NULL, 0, &rtstat, sizeof(rtstat),
1804 CTL_CREATE, CTL_EOL);
1805 }
1806