if.c revision 1.209 1 /* $NetBSD: if.c,v 1.209 2007/12/20 19:53:30 dyoung Exp $ */
2
3 /*-
4 * Copyright (c) 1999, 2000, 2001 The NetBSD Foundation, Inc.
5 * All rights reserved.
6 *
7 * This code is derived from software contributed to The NetBSD Foundation
8 * by William Studenmund and Jason R. Thorpe.
9 *
10 * Redistribution and use in source and binary forms, with or without
11 * modification, are permitted provided that the following conditions
12 * are met:
13 * 1. Redistributions of source code must retain the above copyright
14 * notice, this list of conditions and the following disclaimer.
15 * 2. Redistributions in binary form must reproduce the above copyright
16 * notice, this list of conditions and the following disclaimer in the
17 * documentation and/or other materials provided with the distribution.
18 * 3. All advertising materials mentioning features or use of this software
19 * must display the following acknowledgement:
20 * This product includes software developed by the NetBSD
21 * Foundation, Inc. and its contributors.
22 * 4. Neither the name of The NetBSD Foundation nor the names of its
23 * contributors may be used to endorse or promote products derived
24 * from this software without specific prior written permission.
25 *
26 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
27 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
28 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
29 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
30 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
31 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
32 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
33 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
34 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
35 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
36 * POSSIBILITY OF SUCH DAMAGE.
37 */
38
39 /*
40 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
41 * All rights reserved.
42 *
43 * Redistribution and use in source and binary forms, with or without
44 * modification, are permitted provided that the following conditions
45 * are met:
46 * 1. Redistributions of source code must retain the above copyright
47 * notice, this list of conditions and the following disclaimer.
48 * 2. Redistributions in binary form must reproduce the above copyright
49 * notice, this list of conditions and the following disclaimer in the
50 * documentation and/or other materials provided with the distribution.
51 * 3. Neither the name of the project nor the names of its contributors
52 * may be used to endorse or promote products derived from this software
53 * without specific prior written permission.
54 *
55 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
56 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
57 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
58 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
59 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
60 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
61 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
62 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
63 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
64 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
65 * SUCH DAMAGE.
66 */
67
68 /*
69 * Copyright (c) 1980, 1986, 1993
70 * The Regents of the University of California. All rights reserved.
71 *
72 * Redistribution and use in source and binary forms, with or without
73 * modification, are permitted provided that the following conditions
74 * are met:
75 * 1. Redistributions of source code must retain the above copyright
76 * notice, this list of conditions and the following disclaimer.
77 * 2. Redistributions in binary form must reproduce the above copyright
78 * notice, this list of conditions and the following disclaimer in the
79 * documentation and/or other materials provided with the distribution.
80 * 3. Neither the name of the University nor the names of its contributors
81 * may be used to endorse or promote products derived from this software
82 * without specific prior written permission.
83 *
84 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
85 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
86 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
87 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
88 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
89 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
90 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
91 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
92 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
93 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
94 * SUCH DAMAGE.
95 *
96 * @(#)if.c 8.5 (Berkeley) 1/9/95
97 */
98
99 #include <sys/cdefs.h>
100 __KERNEL_RCSID(0, "$NetBSD: if.c,v 1.209 2007/12/20 19:53:30 dyoung Exp $");
101
102 #include "opt_inet.h"
103
104 #include "opt_atalk.h"
105 #include "opt_natm.h"
106 #include "opt_pfil_hooks.h"
107
108 #include <sys/param.h>
109 #include <sys/mbuf.h>
110 #include <sys/systm.h>
111 #include <sys/callout.h>
112 #include <sys/proc.h>
113 #include <sys/socket.h>
114 #include <sys/socketvar.h>
115 #include <sys/domain.h>
116 #include <sys/protosw.h>
117 #include <sys/kernel.h>
118 #include <sys/ioctl.h>
119 #include <sys/sysctl.h>
120 #include <sys/syslog.h>
121 #include <sys/kauth.h>
122
123 #include <net/if.h>
124 #include <net/if_dl.h>
125 #include <net/if_ether.h>
126 #include <net/if_media.h>
127 #include <net80211/ieee80211.h>
128 #include <net80211/ieee80211_ioctl.h>
129 #include <net/if_types.h>
130 #include <net/radix.h>
131 #include <net/route.h>
132 #include <net/netisr.h>
133 #ifdef NETATALK
134 #include <netatalk/at_extern.h>
135 #include <netatalk/at.h>
136 #endif
137 #include <net/pfil.h>
138
139 #ifdef INET6
140 #include <netinet/in.h>
141 #include <netinet6/in6_var.h>
142 #include <netinet6/nd6.h>
143 #endif
144
145 #include "carp.h"
146 #if NCARP > 0
147 #include <netinet/ip_carp.h>
148 #endif
149
150 #include <compat/sys/sockio.h>
151 #include <compat/sys/socket.h>
152
153 MALLOC_DEFINE(M_IFADDR, "ifaddr", "interface address");
154 MALLOC_DEFINE(M_IFMADDR, "ether_multi", "link-level multicast address");
155
156 int ifqmaxlen = IFQ_MAXLEN;
157 callout_t if_slowtimo_ch;
158
159 int netisr; /* scheduling bits for network */
160
161 static int if_rt_walktree(struct rtentry *, void *);
162
163 static struct if_clone *if_clone_lookup(const char *, int *);
164 static int if_clone_list(struct if_clonereq *);
165
166 static LIST_HEAD(, if_clone) if_cloners = LIST_HEAD_INITIALIZER(if_cloners);
167 static int if_cloners_count;
168
169 #ifdef PFIL_HOOKS
170 struct pfil_head if_pfil; /* packet filtering hook for interfaces */
171 #endif
172
173 static void if_detach_queues(struct ifnet *, struct ifqueue *);
174
175 /*
176 * Network interface utility routines.
177 *
178 * Routines with ifa_ifwith* names take sockaddr *'s as
179 * parameters.
180 */
181 void
182 ifinit(void)
183 {
184
185 callout_init(&if_slowtimo_ch, 0);
186 if_slowtimo(NULL);
187 #ifdef PFIL_HOOKS
188 if_pfil.ph_type = PFIL_TYPE_IFNET;
189 if_pfil.ph_ifnet = NULL;
190 if (pfil_head_register(&if_pfil) != 0)
191 printf("WARNING: unable to register pfil hook\n");
192 #endif
193 }
194
195 /*
196 * Null routines used while an interface is going away. These routines
197 * just return an error.
198 */
199
200 int
201 if_nulloutput(struct ifnet *ifp, struct mbuf *m,
202 const struct sockaddr *so, struct rtentry *rt)
203 {
204
205 return ENXIO;
206 }
207
208 void
209 if_nullinput(struct ifnet *ifp, struct mbuf *m)
210 {
211
212 /* Nothing. */
213 }
214
215 void
216 if_nullstart(struct ifnet *ifp)
217 {
218
219 /* Nothing. */
220 }
221
222 int
223 if_nullioctl(struct ifnet *ifp, u_long cmd, void *data)
224 {
225
226 return ENXIO;
227 }
228
229 int
230 if_nullinit(struct ifnet *ifp)
231 {
232
233 return ENXIO;
234 }
235
236 void
237 if_nullstop(struct ifnet *ifp, int disable)
238 {
239
240 /* Nothing. */
241 }
242
243 void
244 if_nullwatchdog(struct ifnet *ifp)
245 {
246
247 /* Nothing. */
248 }
249
250 void
251 if_nulldrain(struct ifnet *ifp)
252 {
253
254 /* Nothing. */
255 }
256
257 static u_int if_index = 1;
258 struct ifnet_head ifnet;
259 size_t if_indexlim = 0;
260 struct ifaddr **ifnet_addrs = NULL;
261 struct ifnet **ifindex2ifnet = NULL;
262 struct ifnet *lo0ifp;
263
264 /*
265 * Allocate the link level name for the specified interface. This
266 * is an attachment helper. It must be called after ifp->if_addrlen
267 * is initialized, which may not be the case when if_attach() is
268 * called.
269 */
270 void
271 if_alloc_sadl(struct ifnet *ifp)
272 {
273 unsigned socksize, ifasize;
274 int addrlen, namelen;
275 struct sockaddr_dl *mask, *sdl;
276 struct ifaddr *ifa;
277
278 /*
279 * If the interface already has a link name, release it
280 * now. This is useful for interfaces that can change
281 * link types, and thus switch link names often.
282 */
283 if (ifp->if_sadl != NULL)
284 if_free_sadl(ifp);
285
286 namelen = strlen(ifp->if_xname);
287 addrlen = ifp->if_addrlen;
288 socksize = roundup(sockaddr_dl_measure(namelen, addrlen), sizeof(long));
289 ifasize = sizeof(*ifa) + 2 * socksize;
290 ifa = (struct ifaddr *)malloc(ifasize, M_IFADDR, M_WAITOK|M_ZERO);
291
292 sdl = (struct sockaddr_dl *)(ifa + 1);
293 mask = (struct sockaddr_dl *)(socksize + (char *)sdl);
294
295 sockaddr_dl_init(sdl, socksize, ifp->if_index, ifp->if_type,
296 ifp->if_xname, namelen, NULL, addrlen);
297 mask->sdl_len = sockaddr_dl_measure(namelen, 0);
298 memset(&mask->sdl_data[0], 0xff, namelen);
299
300 ifnet_addrs[ifp->if_index] = ifa;
301 IFAREF(ifa);
302 ifa_insert(ifp, ifa);
303 ifp->if_dl = ifa;
304 IFAREF(ifa);
305 ifa->ifa_rtrequest = link_rtrequest;
306 ifa->ifa_addr = (struct sockaddr *)sdl;
307 ifp->if_sadl = sdl;
308 ifa->ifa_netmask = (struct sockaddr *)mask;
309 }
310
311 /*
312 * Free the link level name for the specified interface. This is
313 * a detach helper. This is called from if_detach() or from
314 * link layer type specific detach functions.
315 */
316 void
317 if_free_sadl(struct ifnet *ifp)
318 {
319 struct ifaddr *ifa;
320 int s;
321
322 ifa = ifnet_addrs[ifp->if_index];
323 if (ifa == NULL) {
324 KASSERT(ifp->if_sadl == NULL);
325 KASSERT(ifp->if_dl == NULL);
326 return;
327 }
328
329 KASSERT(ifp->if_sadl != NULL);
330 KASSERT(ifp->if_dl != NULL);
331
332 s = splnet();
333 rtinit(ifa, RTM_DELETE, 0);
334 ifa_remove(ifp, ifa);
335
336 ifp->if_sadl = NULL;
337
338 ifnet_addrs[ifp->if_index] = NULL;
339 IFAFREE(ifa);
340 ifp->if_dl = NULL;
341 IFAFREE(ifa);
342 splx(s);
343 }
344
345 /*
346 * Attach an interface to the
347 * list of "active" interfaces.
348 */
349 void
350 if_attach(struct ifnet *ifp)
351 {
352 int indexlim = 0;
353
354 if (if_indexlim == 0) {
355 TAILQ_INIT(&ifnet);
356 if_indexlim = 8;
357 }
358 TAILQ_INIT(&ifp->if_addrlist);
359 TAILQ_INSERT_TAIL(&ifnet, ifp, if_list);
360 ifp->if_index = if_index;
361 if (ifindex2ifnet == NULL)
362 if_index++;
363 else
364 while (ifp->if_index < if_indexlim &&
365 ifindex2ifnet[ifp->if_index] != NULL) {
366 ++if_index;
367 if (if_index == 0)
368 if_index = 1;
369 /*
370 * If we hit USHRT_MAX, we skip back to 0 since
371 * there are a number of places where the value
372 * of if_index or if_index itself is compared
373 * to or stored in an unsigned short. By
374 * jumping back, we won't botch those assignments
375 * or comparisons.
376 */
377 else if (if_index == USHRT_MAX) {
378 /*
379 * However, if we have to jump back to
380 * zero *twice* without finding an empty
381 * slot in ifindex2ifnet[], then there
382 * there are too many (>65535) interfaces.
383 */
384 if (indexlim++)
385 panic("too many interfaces");
386 else
387 if_index = 1;
388 }
389 ifp->if_index = if_index;
390 }
391
392 /*
393 * We have some arrays that should be indexed by if_index.
394 * since if_index will grow dynamically, they should grow too.
395 * struct ifadd **ifnet_addrs
396 * struct ifnet **ifindex2ifnet
397 */
398 if (ifnet_addrs == NULL || ifindex2ifnet == NULL ||
399 ifp->if_index >= if_indexlim) {
400 size_t m, n, oldlim;
401 void *q;
402
403 oldlim = if_indexlim;
404 while (ifp->if_index >= if_indexlim)
405 if_indexlim <<= 1;
406
407 /* grow ifnet_addrs */
408 m = oldlim * sizeof(struct ifaddr *);
409 n = if_indexlim * sizeof(struct ifaddr *);
410 q = (void *)malloc(n, M_IFADDR, M_WAITOK|M_ZERO);
411 if (ifnet_addrs != NULL) {
412 memcpy(q, ifnet_addrs, m);
413 free((void *)ifnet_addrs, M_IFADDR);
414 }
415 ifnet_addrs = (struct ifaddr **)q;
416
417 /* grow ifindex2ifnet */
418 m = oldlim * sizeof(struct ifnet *);
419 n = if_indexlim * sizeof(struct ifnet *);
420 q = (void *)malloc(n, M_IFADDR, M_WAITOK|M_ZERO);
421 if (ifindex2ifnet != NULL) {
422 memcpy(q, (void *)ifindex2ifnet, m);
423 free((void *)ifindex2ifnet, M_IFADDR);
424 }
425 ifindex2ifnet = (struct ifnet **)q;
426 }
427
428 ifindex2ifnet[ifp->if_index] = ifp;
429
430 /*
431 * Link level name is allocated later by a separate call to
432 * if_alloc_sadl().
433 */
434
435 if (ifp->if_snd.ifq_maxlen == 0)
436 ifp->if_snd.ifq_maxlen = ifqmaxlen;
437 ifp->if_broadcastaddr = 0; /* reliably crash if used uninitialized */
438
439 ifp->if_link_state = LINK_STATE_UNKNOWN;
440
441 ifp->if_capenable = 0;
442 ifp->if_csum_flags_tx = 0;
443 ifp->if_csum_flags_rx = 0;
444
445 #ifdef ALTQ
446 ifp->if_snd.altq_type = 0;
447 ifp->if_snd.altq_disc = NULL;
448 ifp->if_snd.altq_flags &= ALTQF_CANTCHANGE;
449 ifp->if_snd.altq_tbr = NULL;
450 ifp->if_snd.altq_ifp = ifp;
451 #endif
452
453 #ifdef PFIL_HOOKS
454 ifp->if_pfil.ph_type = PFIL_TYPE_IFNET;
455 ifp->if_pfil.ph_ifnet = ifp;
456 if (pfil_head_register(&ifp->if_pfil) != 0)
457 printf("%s: WARNING: unable to register pfil hook\n",
458 ifp->if_xname);
459 (void)pfil_run_hooks(&if_pfil,
460 (struct mbuf **)PFIL_IFNET_ATTACH, ifp, PFIL_IFNET);
461 #endif
462
463 if (!STAILQ_EMPTY(&domains))
464 if_attachdomain1(ifp);
465
466 /* Announce the interface. */
467 rt_ifannouncemsg(ifp, IFAN_ARRIVAL);
468 }
469
470 void
471 if_attachdomain(void)
472 {
473 struct ifnet *ifp;
474 int s;
475
476 s = splnet();
477 IFNET_FOREACH(ifp)
478 if_attachdomain1(ifp);
479 splx(s);
480 }
481
482 void
483 if_attachdomain1(struct ifnet *ifp)
484 {
485 struct domain *dp;
486 int s;
487
488 s = splnet();
489
490 /* address family dependent data region */
491 memset(ifp->if_afdata, 0, sizeof(ifp->if_afdata));
492 DOMAIN_FOREACH(dp) {
493 if (dp->dom_ifattach != NULL)
494 ifp->if_afdata[dp->dom_family] =
495 (*dp->dom_ifattach)(ifp);
496 }
497
498 splx(s);
499 }
500
501 /*
502 * Deactivate an interface. This points all of the procedure
503 * handles at error stubs. May be called from interrupt context.
504 */
505 void
506 if_deactivate(struct ifnet *ifp)
507 {
508 int s;
509
510 s = splnet();
511
512 ifp->if_output = if_nulloutput;
513 ifp->if_input = if_nullinput;
514 ifp->if_start = if_nullstart;
515 ifp->if_ioctl = if_nullioctl;
516 ifp->if_init = if_nullinit;
517 ifp->if_stop = if_nullstop;
518 ifp->if_watchdog = if_nullwatchdog;
519 ifp->if_drain = if_nulldrain;
520
521 /* No more packets may be enqueued. */
522 ifp->if_snd.ifq_maxlen = 0;
523
524 splx(s);
525 }
526
527 void
528 if_purgeaddrs(struct ifnet *ifp, int family,
529 void (*purgeaddr)(struct ifaddr *))
530 {
531 struct ifaddr *ifa, *nifa;
532
533 for (ifa = IFADDR_FIRST(ifp); ifa != NULL; ifa = nifa) {
534 nifa = IFADDR_NEXT(ifa);
535 if (ifa->ifa_addr->sa_family != family)
536 continue;
537 (*purgeaddr)(ifa);
538 }
539 }
540
541 /*
542 * Detach an interface from the list of "active" interfaces,
543 * freeing any resources as we go along.
544 *
545 * NOTE: This routine must be called with a valid thread context,
546 * as it may block.
547 */
548 void
549 if_detach(struct ifnet *ifp)
550 {
551 struct socket so;
552 struct ifaddr *ifa;
553 #ifdef IFAREF_DEBUG
554 struct ifaddr *last_ifa = NULL;
555 #endif
556 struct domain *dp;
557 const struct protosw *pr;
558 int s, i, family, purged;
559
560 /*
561 * XXX It's kind of lame that we have to have the
562 * XXX socket structure...
563 */
564 memset(&so, 0, sizeof(so));
565
566 s = splnet();
567
568 /*
569 * Do an if_down() to give protocols a chance to do something.
570 */
571 if_down(ifp);
572
573 #ifdef ALTQ
574 if (ALTQ_IS_ENABLED(&ifp->if_snd))
575 altq_disable(&ifp->if_snd);
576 if (ALTQ_IS_ATTACHED(&ifp->if_snd))
577 altq_detach(&ifp->if_snd);
578 #endif
579
580
581 #if NCARP > 0
582 /* Remove the interface from any carp group it is a part of. */
583 if (ifp->if_carp != NULL && ifp->if_type != IFT_CARP)
584 carp_ifdetach(ifp);
585 #endif
586
587 /*
588 * Rip all the addresses off the interface. This should make
589 * all of the routes go away.
590 *
591 * pr_usrreq calls can remove an arbitrary number of ifaddrs
592 * from the list, including our "cursor", ifa. For safety,
593 * and to honor the TAILQ abstraction, I just restart the
594 * loop after each removal. Note that the loop will exit
595 * when all of the remaining ifaddrs belong to the AF_LINK
596 * family. I am counting on the historical fact that at
597 * least one pr_usrreq in each address domain removes at
598 * least one ifaddr.
599 */
600 again:
601 IFADDR_FOREACH(ifa, ifp) {
602 family = ifa->ifa_addr->sa_family;
603 #ifdef IFAREF_DEBUG
604 printf("if_detach: ifaddr %p, family %d, refcnt %d\n",
605 ifa, family, ifa->ifa_refcnt);
606 if (last_ifa != NULL && ifa == last_ifa)
607 panic("if_detach: loop detected");
608 last_ifa = ifa;
609 #endif
610 if (family == AF_LINK)
611 continue;
612 dp = pffinddomain(family);
613 #ifdef DIAGNOSTIC
614 if (dp == NULL)
615 panic("if_detach: no domain for AF %d",
616 family);
617 #endif
618 /*
619 * XXX These PURGEIF calls are redundant with the
620 * purge-all-families calls below, but are left in for
621 * now both to make a smaller change, and to avoid
622 * unplanned interactions with clearing of
623 * ifp->if_addrlist.
624 */
625 purged = 0;
626 for (pr = dp->dom_protosw;
627 pr < dp->dom_protoswNPROTOSW; pr++) {
628 so.so_proto = pr;
629 if (pr->pr_usrreq != NULL) {
630 (void) (*pr->pr_usrreq)(&so,
631 PRU_PURGEIF, NULL, NULL,
632 (struct mbuf *) ifp, curlwp);
633 purged = 1;
634 }
635 }
636 if (purged == 0) {
637 /*
638 * XXX What's really the best thing to do
639 * XXX here? --thorpej (at) NetBSD.org
640 */
641 printf("if_detach: WARNING: AF %d not purged\n",
642 family);
643 ifa_remove(ifp, ifa);
644 }
645 goto again;
646 }
647
648 if_free_sadl(ifp);
649
650 /* Walk the routing table looking for stragglers. */
651 for (i = 0; i <= AF_MAX; i++)
652 (void)rt_walktree(i, if_rt_walktree, ifp);
653
654 DOMAIN_FOREACH(dp) {
655 if (dp->dom_ifdetach != NULL && ifp->if_afdata[dp->dom_family])
656 (*dp->dom_ifdetach)(ifp,
657 ifp->if_afdata[dp->dom_family]);
658
659 /*
660 * One would expect multicast memberships (INET and
661 * INET6) on UDP sockets to be purged by the PURGEIF
662 * calls above, but if all addresses were removed from
663 * the interface prior to destruction, the calls will
664 * not be made (e.g. ppp, for which pppd(8) generally
665 * removes addresses before destroying the interface).
666 * Because there is no invariant that multicast
667 * memberships only exist for interfaces with IPv4
668 * addresses, we must call PURGEIF regardless of
669 * addresses. (Protocols which might store ifnet
670 * pointers are marked with PR_PURGEIF.)
671 */
672 for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++) {
673 so.so_proto = pr;
674 if (pr->pr_usrreq != NULL && pr->pr_flags & PR_PURGEIF)
675 (void)(*pr->pr_usrreq)(&so, PRU_PURGEIF, NULL,
676 NULL, (struct mbuf *)ifp, curlwp);
677 }
678 }
679
680 #ifdef PFIL_HOOKS
681 (void)pfil_run_hooks(&if_pfil,
682 (struct mbuf **)PFIL_IFNET_DETACH, ifp, PFIL_IFNET);
683 (void)pfil_head_unregister(&ifp->if_pfil);
684 #endif
685
686 /* Announce that the interface is gone. */
687 rt_ifannouncemsg(ifp, IFAN_DEPARTURE);
688
689 ifindex2ifnet[ifp->if_index] = NULL;
690
691 TAILQ_REMOVE(&ifnet, ifp, if_list);
692
693 /*
694 * remove packets that came from ifp, from software interrupt queues.
695 */
696 DOMAIN_FOREACH(dp) {
697 for (i = 0; i < __arraycount(dp->dom_ifqueues); i++) {
698 if (dp->dom_ifqueues[i] == NULL)
699 break;
700 if_detach_queues(ifp, dp->dom_ifqueues[i]);
701 }
702 }
703
704 splx(s);
705 }
706
707 static void
708 if_detach_queues(struct ifnet *ifp, struct ifqueue *q)
709 {
710 struct mbuf *m, *prev, *next;
711
712 prev = NULL;
713 for (m = q->ifq_head; m != NULL; m = next) {
714 next = m->m_nextpkt;
715 #ifdef DIAGNOSTIC
716 if ((m->m_flags & M_PKTHDR) == 0) {
717 prev = m;
718 continue;
719 }
720 #endif
721 if (m->m_pkthdr.rcvif != ifp) {
722 prev = m;
723 continue;
724 }
725
726 if (prev != NULL)
727 prev->m_nextpkt = m->m_nextpkt;
728 else
729 q->ifq_head = m->m_nextpkt;
730 if (q->ifq_tail == m)
731 q->ifq_tail = prev;
732 q->ifq_len--;
733
734 m->m_nextpkt = NULL;
735 m_freem(m);
736 IF_DROP(q);
737 }
738 }
739
740 /*
741 * Callback for a radix tree walk to delete all references to an
742 * ifnet.
743 */
744 static int
745 if_rt_walktree(struct rtentry *rt, void *v)
746 {
747 struct ifnet *ifp = (struct ifnet *)v;
748 int error;
749
750 if (rt->rt_ifp != ifp)
751 return 0;
752
753 /* Delete the entry. */
754 ++rt->rt_refcnt;
755 error = rtrequest(RTM_DELETE, rt_getkey(rt), rt->rt_gateway,
756 rt_mask(rt), rt->rt_flags, NULL);
757 KASSERT((rt->rt_flags & RTF_UP) == 0);
758 rt->rt_ifp = NULL;
759 RTFREE(rt);
760 if (error != 0)
761 printf("%s: warning: unable to delete rtentry @ %p, "
762 "error = %d\n", ifp->if_xname, rt, error);
763 return 0;
764 }
765
766 /*
767 * Create a clone network interface.
768 */
769 int
770 if_clone_create(const char *name)
771 {
772 struct if_clone *ifc;
773 int unit;
774
775 ifc = if_clone_lookup(name, &unit);
776 if (ifc == NULL)
777 return EINVAL;
778
779 if (ifunit(name) != NULL)
780 return EEXIST;
781
782 return (*ifc->ifc_create)(ifc, unit);
783 }
784
785 /*
786 * Destroy a clone network interface.
787 */
788 int
789 if_clone_destroy(const char *name)
790 {
791 struct if_clone *ifc;
792 struct ifnet *ifp;
793
794 ifc = if_clone_lookup(name, NULL);
795 if (ifc == NULL)
796 return EINVAL;
797
798 ifp = ifunit(name);
799 if (ifp == NULL)
800 return ENXIO;
801
802 if (ifc->ifc_destroy == NULL)
803 return EOPNOTSUPP;
804
805 return (*ifc->ifc_destroy)(ifp);
806 }
807
808 /*
809 * Look up a network interface cloner.
810 */
811 static struct if_clone *
812 if_clone_lookup(const char *name, int *unitp)
813 {
814 struct if_clone *ifc;
815 const char *cp;
816 int unit;
817
818 /* separate interface name from unit */
819 for (cp = name;
820 cp - name < IFNAMSIZ && *cp && (*cp < '0' || *cp > '9');
821 cp++)
822 continue;
823
824 if (cp == name || cp - name == IFNAMSIZ || !*cp)
825 return NULL; /* No name or unit number */
826
827 LIST_FOREACH(ifc, &if_cloners, ifc_list) {
828 if (strlen(ifc->ifc_name) == cp - name &&
829 strncmp(name, ifc->ifc_name, cp - name) == 0)
830 break;
831 }
832
833 if (ifc == NULL)
834 return NULL;
835
836 unit = 0;
837 while (cp - name < IFNAMSIZ && *cp) {
838 if (*cp < '0' || *cp > '9' || unit > INT_MAX / 10) {
839 /* Bogus unit number. */
840 return NULL;
841 }
842 unit = (unit * 10) + (*cp++ - '0');
843 }
844
845 if (unitp != NULL)
846 *unitp = unit;
847 return ifc;
848 }
849
850 /*
851 * Register a network interface cloner.
852 */
853 void
854 if_clone_attach(struct if_clone *ifc)
855 {
856
857 LIST_INSERT_HEAD(&if_cloners, ifc, ifc_list);
858 if_cloners_count++;
859 }
860
861 /*
862 * Unregister a network interface cloner.
863 */
864 void
865 if_clone_detach(struct if_clone *ifc)
866 {
867
868 LIST_REMOVE(ifc, ifc_list);
869 if_cloners_count--;
870 }
871
872 /*
873 * Provide list of interface cloners to userspace.
874 */
875 static int
876 if_clone_list(struct if_clonereq *ifcr)
877 {
878 char outbuf[IFNAMSIZ], *dst;
879 struct if_clone *ifc;
880 int count, error = 0;
881
882 ifcr->ifcr_total = if_cloners_count;
883 if ((dst = ifcr->ifcr_buffer) == NULL) {
884 /* Just asking how many there are. */
885 return 0;
886 }
887
888 if (ifcr->ifcr_count < 0)
889 return EINVAL;
890
891 count = (if_cloners_count < ifcr->ifcr_count) ?
892 if_cloners_count : ifcr->ifcr_count;
893
894 for (ifc = LIST_FIRST(&if_cloners); ifc != NULL && count != 0;
895 ifc = LIST_NEXT(ifc, ifc_list), count--, dst += IFNAMSIZ) {
896 (void)strncpy(outbuf, ifc->ifc_name, sizeof(outbuf));
897 if (outbuf[sizeof(outbuf) - 1] != '\0')
898 return ENAMETOOLONG;
899 error = copyout(outbuf, dst, sizeof(outbuf));
900 if (error != 0)
901 break;
902 }
903
904 return error;
905 }
906
907 void
908 ifa_insert(struct ifnet *ifp, struct ifaddr *ifa)
909 {
910 ifa->ifa_ifp = ifp;
911 TAILQ_INSERT_TAIL(&ifp->if_addrlist, ifa, ifa_list);
912 IFAREF(ifa);
913 }
914
915 void
916 ifa_remove(struct ifnet *ifp, struct ifaddr *ifa)
917 {
918 KASSERT(ifa->ifa_ifp == ifp);
919 TAILQ_REMOVE(&ifp->if_addrlist, ifa, ifa_list);
920 IFAFREE(ifa);
921 }
922
923 static inline int
924 equal(const struct sockaddr *sa1, const struct sockaddr *sa2)
925 {
926 return sockaddr_cmp(sa1, sa2) == 0;
927 }
928
929 /*
930 * Locate an interface based on a complete address.
931 */
932 /*ARGSUSED*/
933 struct ifaddr *
934 ifa_ifwithaddr(const struct sockaddr *addr)
935 {
936 struct ifnet *ifp;
937 struct ifaddr *ifa;
938
939 IFNET_FOREACH(ifp) {
940 if (ifp->if_output == if_nulloutput)
941 continue;
942 IFADDR_FOREACH(ifa, ifp) {
943 if (ifa->ifa_addr->sa_family != addr->sa_family)
944 continue;
945 if (equal(addr, ifa->ifa_addr))
946 return ifa;
947 if ((ifp->if_flags & IFF_BROADCAST) &&
948 ifa->ifa_broadaddr &&
949 /* IP6 doesn't have broadcast */
950 ifa->ifa_broadaddr->sa_len != 0 &&
951 equal(ifa->ifa_broadaddr, addr))
952 return ifa;
953 }
954 }
955 return NULL;
956 }
957
958 /*
959 * Locate the point to point interface with a given destination address.
960 */
961 /*ARGSUSED*/
962 struct ifaddr *
963 ifa_ifwithdstaddr(const struct sockaddr *addr)
964 {
965 struct ifnet *ifp;
966 struct ifaddr *ifa;
967
968 IFNET_FOREACH(ifp) {
969 if (ifp->if_output == if_nulloutput)
970 continue;
971 if ((ifp->if_flags & IFF_POINTOPOINT) == 0)
972 continue;
973 IFADDR_FOREACH(ifa, ifp) {
974 if (ifa->ifa_addr->sa_family != addr->sa_family ||
975 ifa->ifa_dstaddr == NULL)
976 continue;
977 if (equal(addr, ifa->ifa_dstaddr))
978 return ifa;
979 }
980 }
981 return NULL;
982 }
983
984 /*
985 * Find an interface on a specific network. If many, choice
986 * is most specific found.
987 */
988 struct ifaddr *
989 ifa_ifwithnet(const struct sockaddr *addr)
990 {
991 struct ifnet *ifp;
992 struct ifaddr *ifa;
993 const struct sockaddr_dl *sdl;
994 struct ifaddr *ifa_maybe = 0;
995 u_int af = addr->sa_family;
996 const char *addr_data = addr->sa_data, *cplim;
997
998 if (af == AF_LINK) {
999 sdl = satocsdl(addr);
1000 if (sdl->sdl_index && sdl->sdl_index < if_indexlim &&
1001 ifindex2ifnet[sdl->sdl_index] &&
1002 ifindex2ifnet[sdl->sdl_index]->if_output != if_nulloutput)
1003 return ifnet_addrs[sdl->sdl_index];
1004 }
1005 #ifdef NETATALK
1006 if (af == AF_APPLETALK) {
1007 const struct sockaddr_at *sat, *sat2;
1008 sat = (const struct sockaddr_at *)addr;
1009 IFNET_FOREACH(ifp) {
1010 if (ifp->if_output == if_nulloutput)
1011 continue;
1012 ifa = at_ifawithnet((const struct sockaddr_at *)addr, ifp);
1013 if (ifa == NULL)
1014 continue;
1015 sat2 = (struct sockaddr_at *)ifa->ifa_addr;
1016 if (sat2->sat_addr.s_net == sat->sat_addr.s_net)
1017 return ifa; /* exact match */
1018 if (ifa_maybe == NULL) {
1019 /* else keep the if with the right range */
1020 ifa_maybe = ifa;
1021 }
1022 }
1023 return ifa_maybe;
1024 }
1025 #endif
1026 IFNET_FOREACH(ifp) {
1027 if (ifp->if_output == if_nulloutput)
1028 continue;
1029 IFADDR_FOREACH(ifa, ifp) {
1030 const char *cp, *cp2, *cp3;
1031
1032 if (ifa->ifa_addr->sa_family != af ||
1033 ifa->ifa_netmask == NULL)
1034 next: continue;
1035 cp = addr_data;
1036 cp2 = ifa->ifa_addr->sa_data;
1037 cp3 = ifa->ifa_netmask->sa_data;
1038 cplim = (const char *)ifa->ifa_netmask +
1039 ifa->ifa_netmask->sa_len;
1040 while (cp3 < cplim) {
1041 if ((*cp++ ^ *cp2++) & *cp3++) {
1042 /* want to continue for() loop */
1043 goto next;
1044 }
1045 }
1046 if (ifa_maybe == NULL ||
1047 rn_refines((void *)ifa->ifa_netmask,
1048 (void *)ifa_maybe->ifa_netmask))
1049 ifa_maybe = ifa;
1050 }
1051 }
1052 return ifa_maybe;
1053 }
1054
1055 /*
1056 * Find the interface of the addresss.
1057 */
1058 struct ifaddr *
1059 ifa_ifwithladdr(const struct sockaddr *addr)
1060 {
1061 struct ifaddr *ia;
1062
1063 if ((ia = ifa_ifwithaddr(addr)) || (ia = ifa_ifwithdstaddr(addr)) ||
1064 (ia = ifa_ifwithnet(addr)))
1065 return ia;
1066 return NULL;
1067 }
1068
1069 /*
1070 * Find an interface using a specific address family
1071 */
1072 struct ifaddr *
1073 ifa_ifwithaf(int af)
1074 {
1075 struct ifnet *ifp;
1076 struct ifaddr *ifa;
1077
1078 IFNET_FOREACH(ifp) {
1079 if (ifp->if_output == if_nulloutput)
1080 continue;
1081 IFADDR_FOREACH(ifa, ifp) {
1082 if (ifa->ifa_addr->sa_family == af)
1083 return ifa;
1084 }
1085 }
1086 return NULL;
1087 }
1088
1089 /*
1090 * Find an interface address specific to an interface best matching
1091 * a given address.
1092 */
1093 struct ifaddr *
1094 ifaof_ifpforaddr(const struct sockaddr *addr, struct ifnet *ifp)
1095 {
1096 struct ifaddr *ifa;
1097 const char *cp, *cp2, *cp3;
1098 const char *cplim;
1099 struct ifaddr *ifa_maybe = 0;
1100 u_int af = addr->sa_family;
1101
1102 if (ifp->if_output == if_nulloutput)
1103 return NULL;
1104
1105 if (af >= AF_MAX)
1106 return NULL;
1107
1108 IFADDR_FOREACH(ifa, ifp) {
1109 if (ifa->ifa_addr->sa_family != af)
1110 continue;
1111 ifa_maybe = ifa;
1112 if (ifa->ifa_netmask == NULL) {
1113 if (equal(addr, ifa->ifa_addr) ||
1114 (ifa->ifa_dstaddr &&
1115 equal(addr, ifa->ifa_dstaddr)))
1116 return ifa;
1117 continue;
1118 }
1119 cp = addr->sa_data;
1120 cp2 = ifa->ifa_addr->sa_data;
1121 cp3 = ifa->ifa_netmask->sa_data;
1122 cplim = ifa->ifa_netmask->sa_len + (char *)ifa->ifa_netmask;
1123 for (; cp3 < cplim; cp3++) {
1124 if ((*cp++ ^ *cp2++) & *cp3)
1125 break;
1126 }
1127 if (cp3 == cplim)
1128 return ifa;
1129 }
1130 return ifa_maybe;
1131 }
1132
1133 /*
1134 * Default action when installing a route with a Link Level gateway.
1135 * Lookup an appropriate real ifa to point to.
1136 * This should be moved to /sys/net/link.c eventually.
1137 */
1138 void
1139 link_rtrequest(int cmd, struct rtentry *rt, struct rt_addrinfo *info)
1140 {
1141 struct ifaddr *ifa;
1142 const struct sockaddr *dst;
1143 struct ifnet *ifp;
1144
1145 if (cmd != RTM_ADD || ((ifa = rt->rt_ifa) == NULL) ||
1146 ((ifp = ifa->ifa_ifp) == NULL) || ((dst = rt_getkey(rt)) == NULL))
1147 return;
1148 if ((ifa = ifaof_ifpforaddr(dst, ifp)) != NULL) {
1149 rt_replace_ifa(rt, ifa);
1150 if (ifa->ifa_rtrequest && ifa->ifa_rtrequest != link_rtrequest)
1151 ifa->ifa_rtrequest(cmd, rt, info);
1152 }
1153 }
1154
1155 /*
1156 * Handle a change in the interface link state.
1157 */
1158 void
1159 if_link_state_change(struct ifnet *ifp, int link_state)
1160 {
1161 if (ifp->if_link_state == link_state)
1162 return;
1163 ifp->if_link_state = link_state;
1164 /* Notify that the link state has changed. */
1165 rt_ifmsg(ifp);
1166 #if NCARP > 0
1167 if (ifp->if_carp)
1168 carp_carpdev_state(ifp);
1169 #endif
1170 }
1171
1172 /*
1173 * Mark an interface down and notify protocols of
1174 * the transition.
1175 * NOTE: must be called at splsoftnet or equivalent.
1176 */
1177 void
1178 if_down(struct ifnet *ifp)
1179 {
1180 struct ifaddr *ifa;
1181
1182 ifp->if_flags &= ~IFF_UP;
1183 microtime(&ifp->if_lastchange);
1184 IFADDR_FOREACH(ifa, ifp)
1185 pfctlinput(PRC_IFDOWN, ifa->ifa_addr);
1186 IFQ_PURGE(&ifp->if_snd);
1187 #if NCARP > 0
1188 if (ifp->if_carp)
1189 carp_carpdev_state(ifp);
1190 #endif
1191 rt_ifmsg(ifp);
1192 }
1193
1194 /*
1195 * Mark an interface up and notify protocols of
1196 * the transition.
1197 * NOTE: must be called at splsoftnet or equivalent.
1198 */
1199 void
1200 if_up(struct ifnet *ifp)
1201 {
1202 #ifdef notyet
1203 struct ifaddr *ifa;
1204 #endif
1205
1206 ifp->if_flags |= IFF_UP;
1207 microtime(&ifp->if_lastchange);
1208 #ifdef notyet
1209 /* this has no effect on IP, and will kill all ISO connections XXX */
1210 IFADDR_FOREACH(ifa, ifp)
1211 pfctlinput(PRC_IFUP, ifa->ifa_addr);
1212 #endif
1213 #if NCARP > 0
1214 if (ifp->if_carp)
1215 carp_carpdev_state(ifp);
1216 #endif
1217 rt_ifmsg(ifp);
1218 #ifdef INET6
1219 in6_if_up(ifp);
1220 #endif
1221 }
1222
1223 /*
1224 * Handle interface watchdog timer routines. Called
1225 * from softclock, we decrement timers (if set) and
1226 * call the appropriate interface routine on expiration.
1227 */
1228 void
1229 if_slowtimo(void *arg)
1230 {
1231 struct ifnet *ifp;
1232 int s = splnet();
1233
1234 IFNET_FOREACH(ifp) {
1235 if (ifp->if_timer == 0 || --ifp->if_timer)
1236 continue;
1237 if (ifp->if_watchdog != NULL)
1238 (*ifp->if_watchdog)(ifp);
1239 }
1240 splx(s);
1241 callout_reset(&if_slowtimo_ch, hz / IFNET_SLOWHZ, if_slowtimo, NULL);
1242 }
1243
1244 /*
1245 * Set/clear promiscuous mode on interface ifp based on the truth value
1246 * of pswitch. The calls are reference counted so that only the first
1247 * "on" request actually has an effect, as does the final "off" request.
1248 * Results are undefined if the "off" and "on" requests are not matched.
1249 */
1250 int
1251 ifpromisc(struct ifnet *ifp, int pswitch)
1252 {
1253 int pcount, ret;
1254 short flags;
1255 struct ifreq ifr;
1256
1257 pcount = ifp->if_pcount;
1258 flags = ifp->if_flags;
1259 if (pswitch) {
1260 /*
1261 * Allow the device to be "placed" into promiscuous
1262 * mode even if it is not configured up. It will
1263 * consult IFF_PROMISC when it is is brought up.
1264 */
1265 if (ifp->if_pcount++ != 0)
1266 return 0;
1267 ifp->if_flags |= IFF_PROMISC;
1268 if ((ifp->if_flags & IFF_UP) == 0)
1269 return 0;
1270 } else {
1271 if (--ifp->if_pcount > 0)
1272 return 0;
1273 ifp->if_flags &= ~IFF_PROMISC;
1274 /*
1275 * If the device is not configured up, we should not need to
1276 * turn off promiscuous mode (device should have turned it
1277 * off when interface went down; and will look at IFF_PROMISC
1278 * again next time interface comes up).
1279 */
1280 if ((ifp->if_flags & IFF_UP) == 0)
1281 return 0;
1282 }
1283 memset(&ifr, 0, sizeof(ifr));
1284 ifr.ifr_flags = ifp->if_flags;
1285 ret = (*ifp->if_ioctl)(ifp, SIOCSIFFLAGS, (void *) &ifr);
1286 /* Restore interface state if not successful. */
1287 if (ret != 0) {
1288 ifp->if_pcount = pcount;
1289 ifp->if_flags = flags;
1290 }
1291 return ret;
1292 }
1293
1294 /*
1295 * Map interface name to
1296 * interface structure pointer.
1297 */
1298 struct ifnet *
1299 ifunit(const char *name)
1300 {
1301 struct ifnet *ifp;
1302 const char *cp = name;
1303 u_int unit = 0;
1304 u_int i;
1305
1306 /*
1307 * If the entire name is a number, treat it as an ifindex.
1308 */
1309 for (i = 0; i < IFNAMSIZ && *cp >= '0' && *cp <= '9'; i++, cp++) {
1310 unit = unit * 10 + (*cp - '0');
1311 }
1312
1313 /*
1314 * If the number took all of the name, then it's a valid ifindex.
1315 */
1316 if (i == IFNAMSIZ || (cp != name && *cp == '\0')) {
1317 if (unit >= if_indexlim)
1318 return NULL;
1319 ifp = ifindex2ifnet[unit];
1320 if (ifp == NULL || ifp->if_output == if_nulloutput)
1321 return NULL;
1322 return ifp;
1323 }
1324
1325 IFNET_FOREACH(ifp) {
1326 if (ifp->if_output == if_nulloutput)
1327 continue;
1328 if (strcmp(ifp->if_xname, name) == 0)
1329 return ifp;
1330 }
1331 return NULL;
1332 }
1333
1334 /*
1335 * Interface ioctls.
1336 */
1337 int
1338 ifioctl(struct socket *so, u_long cmd, void *data, struct lwp *l)
1339 {
1340 struct ifnet *ifp;
1341 struct ifreq *ifr;
1342 struct ifcapreq *ifcr;
1343 struct ifdatareq *ifdr;
1344 int s, error = 0;
1345 #if defined(COMPAT_OSOCK) || defined(COMPAT_OIFREQ)
1346 u_long ocmd = cmd;
1347 #endif
1348 short oif_flags;
1349 #ifdef COMPAT_OIFREQ
1350 struct ifreq ifrb;
1351 struct oifreq *oifr = NULL;
1352 #endif
1353
1354 switch (cmd) {
1355 #ifdef COMPAT_OIFREQ
1356 case OSIOCGIFCONF:
1357 case OOSIOCGIFCONF:
1358 return compat_ifconf(cmd, data);
1359 #endif
1360 case SIOCGIFCONF:
1361 return ifconf(cmd, data);
1362 }
1363
1364 #ifdef COMPAT_OIFREQ
1365 cmd = compat_cvtcmd(cmd);
1366 if (cmd != ocmd) {
1367 oifr = data;
1368 data = ifr = &ifrb;
1369 ifreqo2n(oifr, ifr);
1370 } else
1371 #endif
1372 ifr = data;
1373 ifcr = data;
1374 ifdr = data;
1375
1376 ifp = ifunit(ifr->ifr_name);
1377
1378 switch (cmd) {
1379 case SIOCIFCREATE:
1380 case SIOCIFDESTROY:
1381 if (l != NULL) {
1382 error = kauth_authorize_network(l->l_cred,
1383 KAUTH_NETWORK_INTERFACE,
1384 KAUTH_REQ_NETWORK_INTERFACE_SETPRIV, ifp,
1385 (void *)cmd, NULL);
1386 if (error != 0)
1387 return error;
1388 }
1389 return (cmd == SIOCIFCREATE) ?
1390 if_clone_create(ifr->ifr_name) :
1391 if_clone_destroy(ifr->ifr_name);
1392
1393 case SIOCIFGCLONERS:
1394 return if_clone_list((struct if_clonereq *)data);
1395 }
1396
1397 if (ifp == NULL)
1398 return ENXIO;
1399
1400 switch (cmd) {
1401 case SIOCSIFFLAGS:
1402 case SIOCSIFCAP:
1403 case SIOCSIFMETRIC:
1404 case SIOCZIFDATA:
1405 case SIOCSIFMTU:
1406 case SIOCSIFPHYADDR:
1407 case SIOCDIFPHYADDR:
1408 #ifdef INET6
1409 case SIOCSIFPHYADDR_IN6:
1410 #endif
1411 case SIOCSLIFPHYADDR:
1412 case SIOCADDMULTI:
1413 case SIOCDELMULTI:
1414 case SIOCSIFMEDIA:
1415 case SIOCSDRVSPEC:
1416 case SIOCG80211:
1417 case SIOCS80211:
1418 case SIOCS80211NWID:
1419 case SIOCS80211NWKEY:
1420 case SIOCS80211POWER:
1421 case SIOCS80211BSSID:
1422 case SIOCS80211CHANNEL:
1423 if (l != NULL) {
1424 error = kauth_authorize_network(l->l_cred,
1425 KAUTH_NETWORK_INTERFACE,
1426 KAUTH_REQ_NETWORK_INTERFACE_SETPRIV, ifp,
1427 (void *)cmd, NULL);
1428 if (error != 0)
1429 return error;
1430 }
1431 }
1432
1433 oif_flags = ifp->if_flags;
1434 switch (cmd) {
1435
1436 case SIOCGIFFLAGS:
1437 ifr->ifr_flags = ifp->if_flags;
1438 break;
1439
1440 case SIOCGIFMETRIC:
1441 ifr->ifr_metric = ifp->if_metric;
1442 break;
1443
1444 case SIOCGIFMTU:
1445 ifr->ifr_mtu = ifp->if_mtu;
1446 break;
1447
1448 case SIOCGIFDLT:
1449 ifr->ifr_dlt = ifp->if_dlt;
1450 break;
1451
1452 case SIOCSIFFLAGS:
1453 if (ifp->if_flags & IFF_UP && (ifr->ifr_flags & IFF_UP) == 0) {
1454 s = splnet();
1455 if_down(ifp);
1456 splx(s);
1457 }
1458 if (ifr->ifr_flags & IFF_UP && (ifp->if_flags & IFF_UP) == 0) {
1459 s = splnet();
1460 if_up(ifp);
1461 splx(s);
1462 }
1463 ifp->if_flags = (ifp->if_flags & IFF_CANTCHANGE) |
1464 (ifr->ifr_flags &~ IFF_CANTCHANGE);
1465 if (ifp->if_ioctl)
1466 (void)(*ifp->if_ioctl)(ifp, cmd, data);
1467 break;
1468
1469 case SIOCGIFCAP:
1470 ifcr->ifcr_capabilities = ifp->if_capabilities;
1471 ifcr->ifcr_capenable = ifp->if_capenable;
1472 break;
1473
1474 case SIOCSIFCAP:
1475 if ((ifcr->ifcr_capenable & ~ifp->if_capabilities) != 0)
1476 return EINVAL;
1477 if (ifp->if_ioctl == NULL)
1478 return EOPNOTSUPP;
1479
1480 /* Must prevent race with packet reception here. */
1481 s = splnet();
1482 if (ifcr->ifcr_capenable != ifp->if_capenable) {
1483 struct ifreq ifrq;
1484
1485 ifrq.ifr_flags = ifp->if_flags;
1486 ifp->if_capenable = ifcr->ifcr_capenable;
1487
1488 /* Pre-compute the checksum flags mask. */
1489 ifp->if_csum_flags_tx = 0;
1490 ifp->if_csum_flags_rx = 0;
1491 if (ifp->if_capenable & IFCAP_CSUM_IPv4_Tx) {
1492 ifp->if_csum_flags_tx |= M_CSUM_IPv4;
1493 }
1494 if (ifp->if_capenable & IFCAP_CSUM_IPv4_Rx) {
1495 ifp->if_csum_flags_rx |= M_CSUM_IPv4;
1496 }
1497
1498 if (ifp->if_capenable & IFCAP_CSUM_TCPv4_Tx) {
1499 ifp->if_csum_flags_tx |= M_CSUM_TCPv4;
1500 }
1501 if (ifp->if_capenable & IFCAP_CSUM_TCPv4_Rx) {
1502 ifp->if_csum_flags_rx |= M_CSUM_TCPv4;
1503 }
1504
1505 if (ifp->if_capenable & IFCAP_CSUM_UDPv4_Tx) {
1506 ifp->if_csum_flags_tx |= M_CSUM_UDPv4;
1507 }
1508 if (ifp->if_capenable & IFCAP_CSUM_UDPv4_Rx) {
1509 ifp->if_csum_flags_rx |= M_CSUM_UDPv4;
1510 }
1511
1512 if (ifp->if_capenable & IFCAP_CSUM_TCPv6_Tx) {
1513 ifp->if_csum_flags_tx |= M_CSUM_TCPv6;
1514 }
1515 if (ifp->if_capenable & IFCAP_CSUM_TCPv6_Rx) {
1516 ifp->if_csum_flags_rx |= M_CSUM_TCPv6;
1517 }
1518
1519 if (ifp->if_capenable & IFCAP_CSUM_UDPv6_Tx) {
1520 ifp->if_csum_flags_tx |= M_CSUM_UDPv6;
1521 }
1522 if (ifp->if_capenable & IFCAP_CSUM_UDPv6_Rx) {
1523 ifp->if_csum_flags_rx |= M_CSUM_UDPv6;
1524 }
1525
1526 /*
1527 * Only kick the interface if it's up. If it's
1528 * not up now, it will notice the cap enables
1529 * when it is brought up later.
1530 */
1531 if (ifp->if_flags & IFF_UP)
1532 (void)(*ifp->if_ioctl)(ifp, SIOCSIFFLAGS,
1533 (void *)&ifrq);
1534 }
1535 splx(s);
1536 break;
1537
1538 case SIOCSIFMETRIC:
1539 ifp->if_metric = ifr->ifr_metric;
1540 break;
1541
1542 case SIOCGIFDATA:
1543 ifdr->ifdr_data = ifp->if_data;
1544 break;
1545
1546 case SIOCZIFDATA:
1547 ifdr->ifdr_data = ifp->if_data;
1548 /*
1549 * Assumes that the volatile counters that can be
1550 * zero'ed are at the end of if_data.
1551 */
1552 memset(&ifp->if_data.ifi_ipackets, 0, sizeof(ifp->if_data) -
1553 offsetof(struct if_data, ifi_ipackets));
1554 break;
1555
1556 case SIOCSIFMTU:
1557 {
1558 u_long oldmtu = ifp->if_mtu;
1559
1560 if (ifp->if_ioctl == NULL)
1561 return EOPNOTSUPP;
1562 error = (*ifp->if_ioctl)(ifp, cmd, data);
1563
1564 /*
1565 * If the link MTU changed, do network layer specific procedure.
1566 */
1567 if (ifp->if_mtu != oldmtu) {
1568 #ifdef INET6
1569 nd6_setmtu(ifp);
1570 #endif
1571 }
1572 break;
1573 }
1574 case SIOCSIFPHYADDR:
1575 case SIOCDIFPHYADDR:
1576 #ifdef INET6
1577 case SIOCSIFPHYADDR_IN6:
1578 #endif
1579 case SIOCSLIFPHYADDR:
1580 case SIOCADDMULTI:
1581 case SIOCDELMULTI:
1582 case SIOCSIFMEDIA:
1583 case SIOCGIFPSRCADDR:
1584 case SIOCGIFPDSTADDR:
1585 case SIOCGLIFPHYADDR:
1586 case SIOCGIFMEDIA:
1587 case SIOCG80211:
1588 case SIOCS80211:
1589 case SIOCS80211NWID:
1590 case SIOCS80211NWKEY:
1591 case SIOCS80211POWER:
1592 case SIOCS80211BSSID:
1593 case SIOCS80211CHANNEL:
1594 if (ifp->if_ioctl == NULL)
1595 return EOPNOTSUPP;
1596 error = (*ifp->if_ioctl)(ifp, cmd, data);
1597 break;
1598
1599 case SIOCSDRVSPEC:
1600 default:
1601 if (so->so_proto == NULL)
1602 return EOPNOTSUPP;
1603 #ifdef COMPAT_OSOCK
1604 error = compat_ifioctl(so, ocmd, cmd, data, l);
1605 #else
1606 error = ((*so->so_proto->pr_usrreq)(so, PRU_CONTROL,
1607 (struct mbuf *)cmd, (struct mbuf *)data,
1608 (struct mbuf *)ifp, l));
1609 #endif
1610 break;
1611 }
1612
1613 if (((oif_flags ^ ifp->if_flags) & IFF_UP) != 0) {
1614 #ifdef INET6
1615 if ((ifp->if_flags & IFF_UP) != 0) {
1616 s = splnet();
1617 in6_if_up(ifp);
1618 splx(s);
1619 }
1620 #endif
1621 }
1622 #ifdef COMPAT_OIFREQ
1623 if (cmd != ocmd)
1624 ifreqn2o(oifr, ifr);
1625 #endif
1626
1627 return error;
1628 }
1629
1630 /*
1631 * Return interface configuration
1632 * of system. List may be used
1633 * in later ioctl's (above) to get
1634 * other information.
1635 *
1636 * Each record is a struct ifreq. Before the addition of
1637 * sockaddr_storage, the API rule was that sockaddr flavors that did
1638 * not fit would extend beyond the struct ifreq, with the next struct
1639 * ifreq starting sa_len beyond the struct sockaddr. Because the
1640 * union in struct ifreq includes struct sockaddr_storage, every kind
1641 * of sockaddr must fit. Thus, there are no longer any overlength
1642 * records.
1643 *
1644 * Records are added to the user buffer if they fit, and ifc_len is
1645 * adjusted to the length that was written. Thus, the user is only
1646 * assured of getting the complete list if ifc_len on return is at
1647 * least sizeof(struct ifreq) less than it was on entry.
1648 *
1649 * If the user buffer pointer is NULL, this routine copies no data and
1650 * returns the amount of space that would be needed.
1651 *
1652 * Invariants:
1653 * ifrp points to the next part of the user's buffer to be used. If
1654 * ifrp != NULL, space holds the number of bytes remaining that we may
1655 * write at ifrp. Otherwise, space holds the number of bytes that
1656 * would have been written had there been adequate space.
1657 */
1658 /*ARGSUSED*/
1659 int
1660 ifconf(u_long cmd, void *data)
1661 {
1662 struct ifconf *ifc = (struct ifconf *)data;
1663 struct ifnet *ifp;
1664 struct ifaddr *ifa;
1665 struct ifreq ifr, *ifrp;
1666 int space, error = 0;
1667 const int sz = (int)sizeof(struct ifreq);
1668
1669 if ((ifrp = ifc->ifc_req) == NULL)
1670 space = 0;
1671 else
1672 space = ifc->ifc_len;
1673 IFNET_FOREACH(ifp) {
1674 (void)strncpy(ifr.ifr_name, ifp->if_xname,
1675 sizeof(ifr.ifr_name));
1676 if (ifr.ifr_name[sizeof(ifr.ifr_name) - 1] != '\0')
1677 return ENAMETOOLONG;
1678 if (IFADDR_EMPTY(ifp)) {
1679 /* Interface with no addresses - send zero sockaddr. */
1680 memset(&ifr.ifr_addr, 0, sizeof(ifr.ifr_addr));
1681 if (ifrp != NULL)
1682 {
1683 if (space >= sz) {
1684 error = copyout(&ifr, ifrp, sz);
1685 if (error != 0)
1686 return (error);
1687 ifrp++; space -= sz;
1688 }
1689 }
1690 else
1691 space += sz;
1692 continue;
1693 }
1694
1695 IFADDR_FOREACH(ifa, ifp) {
1696 struct sockaddr *sa = ifa->ifa_addr;
1697 /* all sockaddrs must fit in sockaddr_storage */
1698 KASSERT(sa->sa_len <= sizeof(ifr.ifr_ifru));
1699
1700 if (ifrp != NULL)
1701 {
1702 memcpy(&ifr.ifr_space, sa, sa->sa_len);
1703 if (space >= sz) {
1704 error = copyout(&ifr, ifrp, sz);
1705 if (error != 0)
1706 return (error);
1707 ifrp++; space -= sz;
1708 }
1709 }
1710 else
1711 space += sz;
1712 }
1713 }
1714 if (ifrp != NULL)
1715 {
1716 KASSERT(0 <= space && space <= ifc->ifc_len);
1717 ifc->ifc_len -= space;
1718 }
1719 else
1720 {
1721 KASSERT(space >= 0);
1722 ifc->ifc_len = space;
1723 }
1724 return (0);
1725 }
1726
1727 int
1728 ifreq_setaddr(const u_long cmd, struct ifreq *ifr, const struct sockaddr *sa)
1729 {
1730 uint8_t len;
1731 u_long ncmd;
1732 const uint8_t osockspace = sizeof(ifr->ifr_addr);
1733 const uint8_t sockspace = sizeof(ifr->ifr_ifru.ifru_space);
1734
1735 #ifdef INET6
1736 if (cmd == SIOCGIFPSRCADDR_IN6 || cmd == SIOCGIFPDSTADDR_IN6)
1737 len = MIN(sizeof(struct sockaddr_in6), sa->sa_len);
1738 else
1739 #endif /* INET6 */
1740 if ((ncmd = compat_cvtcmd(cmd)) != cmd)
1741 len = MIN(osockspace, sa->sa_len);
1742 else
1743 len = MIN(sockspace, sa->sa_len);
1744 if (len < sa->sa_len)
1745 return EFBIG;
1746 sockaddr_copy(&ifr->ifr_addr, len, sa);
1747 return 0;
1748 }
1749
1750 /*
1751 * Queue message on interface, and start output if interface
1752 * not yet active.
1753 */
1754 int
1755 ifq_enqueue(struct ifnet *ifp, struct mbuf *m
1756 ALTQ_COMMA ALTQ_DECL(struct altq_pktattr *pktattr))
1757 {
1758 int len = m->m_pkthdr.len;
1759 int mflags = m->m_flags;
1760 int s = splnet();
1761 int error;
1762
1763 IFQ_ENQUEUE(&ifp->if_snd, m, pktattr, error);
1764 if (error != 0)
1765 goto out;
1766 ifp->if_obytes += len;
1767 if (mflags & M_MCAST)
1768 ifp->if_omcasts++;
1769 if ((ifp->if_flags & IFF_OACTIVE) == 0)
1770 (*ifp->if_start)(ifp);
1771 out:
1772 splx(s);
1773 return error;
1774 }
1775
1776 /*
1777 * Queue message on interface, possibly using a second fast queue
1778 */
1779 int
1780 ifq_enqueue2(struct ifnet *ifp, struct ifqueue *ifq, struct mbuf *m
1781 ALTQ_COMMA ALTQ_DECL(struct altq_pktattr *pktattr))
1782 {
1783 int error = 0;
1784
1785 if (ifq != NULL
1786 #ifdef ALTQ
1787 && ALTQ_IS_ENABLED(&ifp->if_snd) == 0
1788 #endif
1789 ) {
1790 if (IF_QFULL(ifq)) {
1791 IF_DROP(&ifp->if_snd);
1792 m_freem(m);
1793 if (error == 0)
1794 error = ENOBUFS;
1795 } else
1796 IF_ENQUEUE(ifq, m);
1797 } else
1798 IFQ_ENQUEUE(&ifp->if_snd, m, pktattr, error);
1799 if (error != 0) {
1800 ++ifp->if_oerrors;
1801 return error;
1802 }
1803 return 0;
1804 }
1805
1806
1807 #if defined(INET) || defined(INET6)
1808 static void
1809 sysctl_net_ifq_setup(struct sysctllog **clog,
1810 int pf, const char *pfname,
1811 int ipn, const char *ipname,
1812 int qid, struct ifqueue *ifq)
1813 {
1814
1815 sysctl_createv(clog, 0, NULL, NULL,
1816 CTLFLAG_PERMANENT,
1817 CTLTYPE_NODE, "net", NULL,
1818 NULL, 0, NULL, 0,
1819 CTL_NET, CTL_EOL);
1820 sysctl_createv(clog, 0, NULL, NULL,
1821 CTLFLAG_PERMANENT,
1822 CTLTYPE_NODE, pfname, NULL,
1823 NULL, 0, NULL, 0,
1824 CTL_NET, pf, CTL_EOL);
1825 sysctl_createv(clog, 0, NULL, NULL,
1826 CTLFLAG_PERMANENT,
1827 CTLTYPE_NODE, ipname, NULL,
1828 NULL, 0, NULL, 0,
1829 CTL_NET, pf, ipn, CTL_EOL);
1830 sysctl_createv(clog, 0, NULL, NULL,
1831 CTLFLAG_PERMANENT,
1832 CTLTYPE_NODE, "ifq",
1833 SYSCTL_DESCR("Protocol input queue controls"),
1834 NULL, 0, NULL, 0,
1835 CTL_NET, pf, ipn, qid, CTL_EOL);
1836
1837 sysctl_createv(clog, 0, NULL, NULL,
1838 CTLFLAG_PERMANENT,
1839 CTLTYPE_INT, "len",
1840 SYSCTL_DESCR("Current input queue length"),
1841 NULL, 0, &ifq->ifq_len, 0,
1842 CTL_NET, pf, ipn, qid, IFQCTL_LEN, CTL_EOL);
1843 sysctl_createv(clog, 0, NULL, NULL,
1844 CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1845 CTLTYPE_INT, "maxlen",
1846 SYSCTL_DESCR("Maximum allowed input queue length"),
1847 NULL, 0, &ifq->ifq_maxlen, 0,
1848 CTL_NET, pf, ipn, qid, IFQCTL_MAXLEN, CTL_EOL);
1849 #ifdef notyet
1850 sysctl_createv(clog, 0, NULL, NULL,
1851 CTLFLAG_PERMANENT,
1852 CTLTYPE_INT, "peak",
1853 SYSCTL_DESCR("Highest input queue length"),
1854 NULL, 0, &ifq->ifq_peak, 0,
1855 CTL_NET, pf, ipn, qid, IFQCTL_PEAK, CTL_EOL);
1856 #endif
1857 sysctl_createv(clog, 0, NULL, NULL,
1858 CTLFLAG_PERMANENT,
1859 CTLTYPE_INT, "drops",
1860 SYSCTL_DESCR("Packets dropped due to full input queue"),
1861 NULL, 0, &ifq->ifq_drops, 0,
1862 CTL_NET, pf, ipn, qid, IFQCTL_DROPS, CTL_EOL);
1863 }
1864
1865 #ifdef INET
1866 SYSCTL_SETUP(sysctl_net_inet_ip_ifq_setup,
1867 "sysctl net.inet.ip.ifq subtree setup")
1868 {
1869 extern struct ifqueue ipintrq;
1870
1871 sysctl_net_ifq_setup(clog, PF_INET, "inet", IPPROTO_IP, "ip",
1872 IPCTL_IFQ, &ipintrq);
1873 }
1874 #endif /* INET */
1875
1876 #ifdef INET6
1877 SYSCTL_SETUP(sysctl_net_inet6_ip6_ifq_setup,
1878 "sysctl net.inet6.ip6.ifq subtree setup")
1879 {
1880 extern struct ifqueue ip6intrq;
1881
1882 sysctl_net_ifq_setup(clog, PF_INET6, "inet6", IPPROTO_IPV6, "ip6",
1883 IPV6CTL_IFQ, &ip6intrq);
1884 }
1885 #endif /* INET6 */
1886 #endif /* INET || INET6 */
1887