FreeBSD/Linux Kernel Cross Reference
sys/net/if.c
1 /* $NetBSD: if.c,v 1.230.4.4 2011/02/16 20:37:47 bouyer Exp $ */
2
3 /*-
4 * Copyright (c) 1999, 2000, 2001, 2008 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 *
19 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
20 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
21 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
22 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
23 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
24 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
25 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
26 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
27 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
28 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
29 * POSSIBILITY OF SUCH DAMAGE.
30 */
31
32 /*
33 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
34 * 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 project 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 PROJECT 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 PROJECT 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
61 /*
62 * Copyright (c) 1980, 1986, 1993
63 * The Regents of the University of California. All rights reserved.
64 *
65 * Redistribution and use in source and binary forms, with or without
66 * modification, are permitted provided that the following conditions
67 * are met:
68 * 1. Redistributions of source code must retain the above copyright
69 * notice, this list of conditions and the following disclaimer.
70 * 2. Redistributions in binary form must reproduce the above copyright
71 * notice, this list of conditions and the following disclaimer in the
72 * documentation and/or other materials provided with the distribution.
73 * 3. Neither the name of the University nor the names of its contributors
74 * may be used to endorse or promote products derived from this software
75 * without specific prior written permission.
76 *
77 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
78 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
79 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
80 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
81 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
82 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
83 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
84 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
85 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
86 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
87 * SUCH DAMAGE.
88 *
89 * @(#)if.c 8.5 (Berkeley) 1/9/95
90 */
91
92 #include <sys/cdefs.h>
93 __KERNEL_RCSID(0, "$NetBSD: if.c,v 1.230.4.4 2011/02/16 20:37:47 bouyer Exp $");
94
95 #include "opt_inet.h"
96
97 #include "opt_atalk.h"
98 #include "opt_natm.h"
99 #include "opt_pfil_hooks.h"
100
101 #include <sys/param.h>
102 #include <sys/mbuf.h>
103 #include <sys/systm.h>
104 #include <sys/callout.h>
105 #include <sys/proc.h>
106 #include <sys/socket.h>
107 #include <sys/socketvar.h>
108 #include <sys/domain.h>
109 #include <sys/protosw.h>
110 #include <sys/kernel.h>
111 #include <sys/ioctl.h>
112 #include <sys/sysctl.h>
113 #include <sys/syslog.h>
114 #include <sys/kauth.h>
115
116 #include <net/if.h>
117 #include <net/if_dl.h>
118 #include <net/if_ether.h>
119 #include <net/if_media.h>
120 #include <net80211/ieee80211.h>
121 #include <net80211/ieee80211_ioctl.h>
122 #include <net/if_types.h>
123 #include <net/radix.h>
124 #include <net/route.h>
125 #include <net/netisr.h>
126 #ifdef NETATALK
127 #include <netatalk/at_extern.h>
128 #include <netatalk/at.h>
129 #endif
130 #include <net/pfil.h>
131
132 #ifdef INET6
133 #include <netinet/in.h>
134 #include <netinet6/in6_var.h>
135 #include <netinet6/nd6.h>
136 #endif
137
138 #include "carp.h"
139 #if NCARP > 0
140 #include <netinet/ip_carp.h>
141 #endif
142
143 #include <compat/sys/sockio.h>
144 #include <compat/sys/socket.h>
145
146 MALLOC_DEFINE(M_IFADDR, "ifaddr", "interface address");
147 MALLOC_DEFINE(M_IFMADDR, "ether_multi", "link-level multicast address");
148
149 int ifqmaxlen = IFQ_MAXLEN;
150 callout_t if_slowtimo_ch;
151
152 int netisr; /* scheduling bits for network */
153
154 static int if_rt_walktree(struct rtentry *, void *);
155
156 static struct if_clone *if_clone_lookup(const char *, int *);
157 static int if_clone_list(struct if_clonereq *);
158
159 static LIST_HEAD(, if_clone) if_cloners = LIST_HEAD_INITIALIZER(if_cloners);
160 static int if_cloners_count;
161
162 #ifdef PFIL_HOOKS
163 struct pfil_head if_pfil; /* packet filtering hook for interfaces */
164 #endif
165
166 static void if_detach_queues(struct ifnet *, struct ifqueue *);
167
168 /*
169 * Network interface utility routines.
170 *
171 * Routines with ifa_ifwith* names take sockaddr *'s as
172 * parameters.
173 */
174 void
175 ifinit(void)
176 {
177
178 callout_init(&if_slowtimo_ch, 0);
179 if_slowtimo(NULL);
180 }
181
182 /*
183 * XXX Initialization before configure().
184 * XXX hack to get pfil_add_hook working in autoconf.
185 */
186 void
187 ifinit1(void)
188 {
189
190 #ifdef PFIL_HOOKS
191 if_pfil.ph_type = PFIL_TYPE_IFNET;
192 if_pfil.ph_ifnet = NULL;
193 if (pfil_head_register(&if_pfil) != 0)
194 printf("WARNING: unable to register pfil hook\n");
195 #endif
196 }
197
198 struct ifnet *
199 if_alloc(u_char type)
200 {
201 return malloc(sizeof(struct ifnet), M_DEVBUF, M_WAITOK|M_ZERO);
202 }
203
204 void
205 if_initname(struct ifnet *ifp, const char *name, int unit)
206 {
207 (void)snprintf(ifp->if_xname, sizeof(ifp->if_xname),
208 "%s%d", name, unit);
209 }
210
211 /*
212 * Null routines used while an interface is going away. These routines
213 * just return an error.
214 */
215
216 int
217 if_nulloutput(struct ifnet *ifp, struct mbuf *m,
218 const struct sockaddr *so, struct rtentry *rt)
219 {
220
221 return ENXIO;
222 }
223
224 void
225 if_nullinput(struct ifnet *ifp, struct mbuf *m)
226 {
227
228 /* Nothing. */
229 }
230
231 void
232 if_nullstart(struct ifnet *ifp)
233 {
234
235 /* Nothing. */
236 }
237
238 int
239 if_nullioctl(struct ifnet *ifp, u_long cmd, void *data)
240 {
241
242 return ENXIO;
243 }
244
245 int
246 if_nullinit(struct ifnet *ifp)
247 {
248
249 return ENXIO;
250 }
251
252 void
253 if_nullstop(struct ifnet *ifp, int disable)
254 {
255
256 /* Nothing. */
257 }
258
259 void
260 if_nullwatchdog(struct ifnet *ifp)
261 {
262
263 /* Nothing. */
264 }
265
266 void
267 if_nulldrain(struct ifnet *ifp)
268 {
269
270 /* Nothing. */
271 }
272
273 static u_int if_index = 1;
274 struct ifnet_head ifnet;
275 size_t if_indexlim = 0;
276 struct ifaddr **ifnet_addrs = NULL;
277 struct ifnet **ifindex2ifnet = NULL;
278 struct ifnet *lo0ifp;
279
280 void
281 if_set_sadl(struct ifnet *ifp, const void *lla, u_char addrlen)
282 {
283 struct ifaddr *ifa;
284 struct sockaddr_dl *sdl;
285
286 ifp->if_addrlen = addrlen;
287 if_alloc_sadl(ifp);
288 ifa = ifp->if_dl;
289 sdl = satosdl(ifa->ifa_addr);
290
291 (void)sockaddr_dl_setaddr(sdl, sdl->sdl_len, lla, ifp->if_addrlen);
292 /* TBD routing socket */
293 }
294
295 struct ifaddr *
296 if_dl_create(const struct ifnet *ifp, const struct sockaddr_dl **sdlp)
297 {
298 unsigned socksize, ifasize;
299 int addrlen, namelen;
300 struct sockaddr_dl *mask, *sdl;
301 struct ifaddr *ifa;
302
303 namelen = strlen(ifp->if_xname);
304 addrlen = ifp->if_addrlen;
305 socksize = roundup(sockaddr_dl_measure(namelen, addrlen), sizeof(long));
306 ifasize = sizeof(*ifa) + 2 * socksize;
307 ifa = (struct ifaddr *)malloc(ifasize, M_IFADDR, M_WAITOK|M_ZERO);
308
309 sdl = (struct sockaddr_dl *)(ifa + 1);
310 mask = (struct sockaddr_dl *)(socksize + (char *)sdl);
311
312 sockaddr_dl_init(sdl, socksize, ifp->if_index, ifp->if_type,
313 ifp->if_xname, namelen, NULL, addrlen);
314 mask->sdl_len = sockaddr_dl_measure(namelen, 0);
315 memset(&mask->sdl_data[0], 0xff, namelen);
316 ifa->ifa_rtrequest = link_rtrequest;
317 ifa->ifa_addr = (struct sockaddr *)sdl;
318 ifa->ifa_netmask = (struct sockaddr *)mask;
319
320 *sdlp = sdl;
321
322 return ifa;
323 }
324
325 static void
326 if_sadl_setrefs(struct ifnet *ifp, struct ifaddr *ifa)
327 {
328 const struct sockaddr_dl *sdl;
329 ifnet_addrs[ifp->if_index] = ifa;
330 IFAREF(ifa);
331 ifp->if_dl = ifa;
332 IFAREF(ifa);
333 sdl = satosdl(ifa->ifa_addr);
334 ifp->if_sadl = sdl;
335 }
336
337 /*
338 * Allocate the link level name for the specified interface. This
339 * is an attachment helper. It must be called after ifp->if_addrlen
340 * is initialized, which may not be the case when if_attach() is
341 * called.
342 */
343 void
344 if_alloc_sadl(struct ifnet *ifp)
345 {
346 struct ifaddr *ifa;
347 const struct sockaddr_dl *sdl;
348
349 /*
350 * If the interface already has a link name, release it
351 * now. This is useful for interfaces that can change
352 * link types, and thus switch link names often.
353 */
354 if (ifp->if_sadl != NULL)
355 if_free_sadl(ifp);
356
357 ifa = if_dl_create(ifp, &sdl);
358
359 ifa_insert(ifp, ifa);
360 if_sadl_setrefs(ifp, ifa);
361 }
362
363 static void
364 if_deactivate_sadl(struct ifnet *ifp)
365 {
366 struct ifaddr *ifa;
367
368 KASSERT(ifp->if_dl != NULL);
369
370 ifa = ifp->if_dl;
371
372 ifp->if_sadl = NULL;
373
374 ifnet_addrs[ifp->if_index] = NULL;
375 IFAFREE(ifa);
376 ifp->if_dl = NULL;
377 IFAFREE(ifa);
378 }
379
380 void
381 if_activate_sadl(struct ifnet *ifp, struct ifaddr *ifa,
382 const struct sockaddr_dl *sdl)
383 {
384 int s;
385
386 s = splnet();
387
388 if_deactivate_sadl(ifp);
389
390 if_sadl_setrefs(ifp, ifa);
391 splx(s);
392 rt_ifmsg(ifp);
393 }
394
395 /*
396 * Free the link level name for the specified interface. This is
397 * a detach helper. This is called from if_detach() or from
398 * link layer type specific detach functions.
399 */
400 void
401 if_free_sadl(struct ifnet *ifp)
402 {
403 struct ifaddr *ifa;
404 int s;
405
406 ifa = ifnet_addrs[ifp->if_index];
407 if (ifa == NULL) {
408 KASSERT(ifp->if_sadl == NULL);
409 KASSERT(ifp->if_dl == NULL);
410 return;
411 }
412
413 KASSERT(ifp->if_sadl != NULL);
414 KASSERT(ifp->if_dl != NULL);
415
416 s = splnet();
417 rtinit(ifa, RTM_DELETE, 0);
418 ifa_remove(ifp, ifa);
419
420 if_deactivate_sadl(ifp);
421 splx(s);
422 }
423
424 /*
425 * Attach an interface to the
426 * list of "active" interfaces.
427 */
428 void
429 if_attach(struct ifnet *ifp)
430 {
431 int indexlim = 0;
432
433 if (if_indexlim == 0) {
434 TAILQ_INIT(&ifnet);
435 if_indexlim = 8;
436 }
437 TAILQ_INIT(&ifp->if_addrlist);
438 TAILQ_INSERT_TAIL(&ifnet, ifp, if_list);
439 ifp->if_index = if_index;
440 if (ifindex2ifnet == NULL)
441 if_index++;
442 else
443 while (ifp->if_index < if_indexlim &&
444 ifindex2ifnet[ifp->if_index] != NULL) {
445 ++if_index;
446 if (if_index == 0)
447 if_index = 1;
448 /*
449 * If we hit USHRT_MAX, we skip back to 0 since
450 * there are a number of places where the value
451 * of if_index or if_index itself is compared
452 * to or stored in an unsigned short. By
453 * jumping back, we won't botch those assignments
454 * or comparisons.
455 */
456 else if (if_index == USHRT_MAX) {
457 /*
458 * However, if we have to jump back to
459 * zero *twice* without finding an empty
460 * slot in ifindex2ifnet[], then there
461 * there are too many (>65535) interfaces.
462 */
463 if (indexlim++)
464 panic("too many interfaces");
465 else
466 if_index = 1;
467 }
468 ifp->if_index = if_index;
469 }
470
471 /*
472 * We have some arrays that should be indexed by if_index.
473 * since if_index will grow dynamically, they should grow too.
474 * struct ifadd **ifnet_addrs
475 * struct ifnet **ifindex2ifnet
476 */
477 if (ifnet_addrs == NULL || ifindex2ifnet == NULL ||
478 ifp->if_index >= if_indexlim) {
479 size_t m, n, oldlim;
480 void *q;
481
482 oldlim = if_indexlim;
483 while (ifp->if_index >= if_indexlim)
484 if_indexlim <<= 1;
485
486 /* grow ifnet_addrs */
487 m = oldlim * sizeof(struct ifaddr *);
488 n = if_indexlim * sizeof(struct ifaddr *);
489 q = malloc(n, M_IFADDR, M_WAITOK|M_ZERO);
490 if (ifnet_addrs != NULL) {
491 memcpy(q, ifnet_addrs, m);
492 free(ifnet_addrs, M_IFADDR);
493 }
494 ifnet_addrs = (struct ifaddr **)q;
495
496 /* grow ifindex2ifnet */
497 m = oldlim * sizeof(struct ifnet *);
498 n = if_indexlim * sizeof(struct ifnet *);
499 q = malloc(n, M_IFADDR, M_WAITOK|M_ZERO);
500 if (ifindex2ifnet != NULL) {
501 memcpy(q, ifindex2ifnet, m);
502 free(ifindex2ifnet, M_IFADDR);
503 }
504 ifindex2ifnet = (struct ifnet **)q;
505 }
506
507 ifindex2ifnet[ifp->if_index] = ifp;
508
509 /*
510 * Link level name is allocated later by a separate call to
511 * if_alloc_sadl().
512 */
513
514 if (ifp->if_snd.ifq_maxlen == 0)
515 ifp->if_snd.ifq_maxlen = ifqmaxlen;
516 ifp->if_broadcastaddr = 0; /* reliably crash if used uninitialized */
517
518 ifp->if_link_state = LINK_STATE_UNKNOWN;
519
520 ifp->if_capenable = 0;
521 ifp->if_csum_flags_tx = 0;
522 ifp->if_csum_flags_rx = 0;
523
524 #ifdef ALTQ
525 ifp->if_snd.altq_type = 0;
526 ifp->if_snd.altq_disc = NULL;
527 ifp->if_snd.altq_flags &= ALTQF_CANTCHANGE;
528 ifp->if_snd.altq_tbr = NULL;
529 ifp->if_snd.altq_ifp = ifp;
530 #endif
531
532 #ifdef PFIL_HOOKS
533 ifp->if_pfil.ph_type = PFIL_TYPE_IFNET;
534 ifp->if_pfil.ph_ifnet = ifp;
535 if (pfil_head_register(&ifp->if_pfil) != 0)
536 printf("%s: WARNING: unable to register pfil hook\n",
537 ifp->if_xname);
538 (void)pfil_run_hooks(&if_pfil,
539 (struct mbuf **)PFIL_IFNET_ATTACH, ifp, PFIL_IFNET);
540 #endif
541
542 if (!STAILQ_EMPTY(&domains))
543 if_attachdomain1(ifp);
544
545 /* Announce the interface. */
546 rt_ifannouncemsg(ifp, IFAN_ARRIVAL);
547 }
548
549 void
550 if_attachdomain(void)
551 {
552 struct ifnet *ifp;
553 int s;
554
555 s = splnet();
556 IFNET_FOREACH(ifp)
557 if_attachdomain1(ifp);
558 splx(s);
559 }
560
561 void
562 if_attachdomain1(struct ifnet *ifp)
563 {
564 struct domain *dp;
565 int s;
566
567 s = splnet();
568
569 /* address family dependent data region */
570 memset(ifp->if_afdata, 0, sizeof(ifp->if_afdata));
571 DOMAIN_FOREACH(dp) {
572 if (dp->dom_ifattach != NULL)
573 ifp->if_afdata[dp->dom_family] =
574 (*dp->dom_ifattach)(ifp);
575 }
576
577 splx(s);
578 }
579
580 /*
581 * Deactivate an interface. This points all of the procedure
582 * handles at error stubs. May be called from interrupt context.
583 */
584 void
585 if_deactivate(struct ifnet *ifp)
586 {
587 int s;
588
589 s = splnet();
590
591 ifp->if_output = if_nulloutput;
592 ifp->if_input = if_nullinput;
593 ifp->if_start = if_nullstart;
594 ifp->if_ioctl = if_nullioctl;
595 ifp->if_init = if_nullinit;
596 ifp->if_stop = if_nullstop;
597 ifp->if_watchdog = if_nullwatchdog;
598 ifp->if_drain = if_nulldrain;
599
600 /* No more packets may be enqueued. */
601 ifp->if_snd.ifq_maxlen = 0;
602
603 splx(s);
604 }
605
606 void
607 if_purgeaddrs(struct ifnet *ifp, int family, void (*purgeaddr)(struct ifaddr *))
608 {
609 struct ifaddr *ifa, *nifa;
610
611 for (ifa = IFADDR_FIRST(ifp); ifa != NULL; ifa = nifa) {
612 nifa = IFADDR_NEXT(ifa);
613 if (ifa->ifa_addr->sa_family != family)
614 continue;
615 (*purgeaddr)(ifa);
616 }
617 }
618
619 /*
620 * Detach an interface from the list of "active" interfaces,
621 * freeing any resources as we go along.
622 *
623 * NOTE: This routine must be called with a valid thread context,
624 * as it may block.
625 */
626 void
627 if_detach(struct ifnet *ifp)
628 {
629 struct socket so;
630 struct ifaddr *ifa;
631 #ifdef IFAREF_DEBUG
632 struct ifaddr *last_ifa = NULL;
633 #endif
634 struct domain *dp;
635 const struct protosw *pr;
636 int s, i, family, purged;
637
638 /*
639 * XXX It's kind of lame that we have to have the
640 * XXX socket structure...
641 */
642 memset(&so, 0, sizeof(so));
643
644 s = splnet();
645
646 /*
647 * Do an if_down() to give protocols a chance to do something.
648 */
649 if_down(ifp);
650
651 #ifdef ALTQ
652 if (ALTQ_IS_ENABLED(&ifp->if_snd))
653 altq_disable(&ifp->if_snd);
654 if (ALTQ_IS_ATTACHED(&ifp->if_snd))
655 altq_detach(&ifp->if_snd);
656 #endif
657
658
659 #if NCARP > 0
660 /* Remove the interface from any carp group it is a part of. */
661 if (ifp->if_carp != NULL && ifp->if_type != IFT_CARP)
662 carp_ifdetach(ifp);
663 #endif
664
665 /*
666 * Rip all the addresses off the interface. This should make
667 * all of the routes go away.
668 *
669 * pr_usrreq calls can remove an arbitrary number of ifaddrs
670 * from the list, including our "cursor", ifa. For safety,
671 * and to honor the TAILQ abstraction, I just restart the
672 * loop after each removal. Note that the loop will exit
673 * when all of the remaining ifaddrs belong to the AF_LINK
674 * family. I am counting on the historical fact that at
675 * least one pr_usrreq in each address domain removes at
676 * least one ifaddr.
677 */
678 again:
679 IFADDR_FOREACH(ifa, ifp) {
680 family = ifa->ifa_addr->sa_family;
681 #ifdef IFAREF_DEBUG
682 printf("if_detach: ifaddr %p, family %d, refcnt %d\n",
683 ifa, family, ifa->ifa_refcnt);
684 if (last_ifa != NULL && ifa == last_ifa)
685 panic("if_detach: loop detected");
686 last_ifa = ifa;
687 #endif
688 if (family == AF_LINK)
689 continue;
690 dp = pffinddomain(family);
691 #ifdef DIAGNOSTIC
692 if (dp == NULL)
693 panic("if_detach: no domain for AF %d",
694 family);
695 #endif
696 /*
697 * XXX These PURGEIF calls are redundant with the
698 * purge-all-families calls below, but are left in for
699 * now both to make a smaller change, and to avoid
700 * unplanned interactions with clearing of
701 * ifp->if_addrlist.
702 */
703 purged = 0;
704 for (pr = dp->dom_protosw;
705 pr < dp->dom_protoswNPROTOSW; pr++) {
706 so.so_proto = pr;
707 if (pr->pr_usrreq != NULL) {
708 (void) (*pr->pr_usrreq)(&so,
709 PRU_PURGEIF, NULL, NULL,
710 (struct mbuf *) ifp, curlwp);
711 purged = 1;
712 }
713 }
714 if (purged == 0) {
715 /*
716 * XXX What's really the best thing to do
717 * XXX here? --thorpej@NetBSD.org
718 */
719 printf("if_detach: WARNING: AF %d not purged\n",
720 family);
721 ifa_remove(ifp, ifa);
722 }
723 goto again;
724 }
725
726 if_free_sadl(ifp);
727
728 /* Walk the routing table looking for stragglers. */
729 for (i = 0; i <= AF_MAX; i++)
730 (void)rt_walktree(i, if_rt_walktree, ifp);
731
732 DOMAIN_FOREACH(dp) {
733 if (dp->dom_ifdetach != NULL && ifp->if_afdata[dp->dom_family])
734 (*dp->dom_ifdetach)(ifp,
735 ifp->if_afdata[dp->dom_family]);
736
737 /*
738 * One would expect multicast memberships (INET and
739 * INET6) on UDP sockets to be purged by the PURGEIF
740 * calls above, but if all addresses were removed from
741 * the interface prior to destruction, the calls will
742 * not be made (e.g. ppp, for which pppd(8) generally
743 * removes addresses before destroying the interface).
744 * Because there is no invariant that multicast
745 * memberships only exist for interfaces with IPv4
746 * addresses, we must call PURGEIF regardless of
747 * addresses. (Protocols which might store ifnet
748 * pointers are marked with PR_PURGEIF.)
749 */
750 for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++) {
751 so.so_proto = pr;
752 if (pr->pr_usrreq != NULL && pr->pr_flags & PR_PURGEIF)
753 (void)(*pr->pr_usrreq)(&so, PRU_PURGEIF, NULL,
754 NULL, (struct mbuf *)ifp, curlwp);
755 }
756 }
757
758 #ifdef PFIL_HOOKS
759 (void)pfil_run_hooks(&if_pfil,
760 (struct mbuf **)PFIL_IFNET_DETACH, ifp, PFIL_IFNET);
761 (void)pfil_head_unregister(&ifp->if_pfil);
762 #endif
763
764 /* Announce that the interface is gone. */
765 rt_ifannouncemsg(ifp, IFAN_DEPARTURE);
766
767 ifindex2ifnet[ifp->if_index] = NULL;
768
769 TAILQ_REMOVE(&ifnet, ifp, if_list);
770
771 /*
772 * remove packets that came from ifp, from software interrupt queues.
773 */
774 DOMAIN_FOREACH(dp) {
775 for (i = 0; i < __arraycount(dp->dom_ifqueues); i++) {
776 if (dp->dom_ifqueues[i] == NULL)
777 break;
778 if_detach_queues(ifp, dp->dom_ifqueues[i]);
779 }
780 }
781
782 splx(s);
783 }
784
785 static void
786 if_detach_queues(struct ifnet *ifp, struct ifqueue *q)
787 {
788 struct mbuf *m, *prev, *next;
789
790 prev = NULL;
791 for (m = q->ifq_head; m != NULL; m = next) {
792 next = m->m_nextpkt;
793 #ifdef DIAGNOSTIC
794 if ((m->m_flags & M_PKTHDR) == 0) {
795 prev = m;
796 continue;
797 }
798 #endif
799 if (m->m_pkthdr.rcvif != ifp) {
800 prev = m;
801 continue;
802 }
803
804 if (prev != NULL)
805 prev->m_nextpkt = m->m_nextpkt;
806 else
807 q->ifq_head = m->m_nextpkt;
808 if (q->ifq_tail == m)
809 q->ifq_tail = prev;
810 q->ifq_len--;
811
812 m->m_nextpkt = NULL;
813 m_freem(m);
814 IF_DROP(q);
815 }
816 }
817
818 /*
819 * Callback for a radix tree walk to delete all references to an
820 * ifnet.
821 */
822 static int
823 if_rt_walktree(struct rtentry *rt, void *v)
824 {
825 struct ifnet *ifp = (struct ifnet *)v;
826 int error;
827
828 if (rt->rt_ifp != ifp)
829 return 0;
830
831 /* Delete the entry. */
832 ++rt->rt_refcnt;
833 error = rtrequest(RTM_DELETE, rt_getkey(rt), rt->rt_gateway,
834 rt_mask(rt), rt->rt_flags, NULL);
835 KASSERT((rt->rt_flags & RTF_UP) == 0);
836 rt->rt_ifp = NULL;
837 RTFREE(rt);
838 if (error != 0)
839 printf("%s: warning: unable to delete rtentry @ %p, "
840 "error = %d\n", ifp->if_xname, rt, error);
841 return 0;
842 }
843
844 /*
845 * Create a clone network interface.
846 */
847 int
848 if_clone_create(const char *name)
849 {
850 struct if_clone *ifc;
851 int unit;
852
853 ifc = if_clone_lookup(name, &unit);
854 if (ifc == NULL)
855 return EINVAL;
856
857 if (ifunit(name) != NULL)
858 return EEXIST;
859
860 return (*ifc->ifc_create)(ifc, unit);
861 }
862
863 /*
864 * Destroy a clone network interface.
865 */
866 int
867 if_clone_destroy(const char *name)
868 {
869 struct if_clone *ifc;
870 struct ifnet *ifp;
871
872 ifc = if_clone_lookup(name, NULL);
873 if (ifc == NULL)
874 return EINVAL;
875
876 ifp = ifunit(name);
877 if (ifp == NULL)
878 return ENXIO;
879
880 if (ifc->ifc_destroy == NULL)
881 return EOPNOTSUPP;
882
883 return (*ifc->ifc_destroy)(ifp);
884 }
885
886 /*
887 * Look up a network interface cloner.
888 */
889 static struct if_clone *
890 if_clone_lookup(const char *name, int *unitp)
891 {
892 struct if_clone *ifc;
893 const char *cp;
894 int unit;
895
896 /* separate interface name from unit */
897 for (cp = name;
898 cp - name < IFNAMSIZ && *cp && (*cp < '' || *cp > '9');
899 cp++)
900 continue;
901
902 if (cp == name || cp - name == IFNAMSIZ || !*cp)
903 return NULL; /* No name or unit number */
904
905 LIST_FOREACH(ifc, &if_cloners, ifc_list) {
906 if (strlen(ifc->ifc_name) == cp - name &&
907 strncmp(name, ifc->ifc_name, cp - name) == 0)
908 break;
909 }
910
911 if (ifc == NULL)
912 return NULL;
913
914 unit = 0;
915 while (cp - name < IFNAMSIZ && *cp) {
916 if (*cp < '' || *cp > '9' || unit > INT_MAX / 10) {
917 /* Bogus unit number. */
918 return NULL;
919 }
920 unit = (unit * 10) + (*cp++ - '');
921 }
922
923 if (unitp != NULL)
924 *unitp = unit;
925 return ifc;
926 }
927
928 /*
929 * Register a network interface cloner.
930 */
931 void
932 if_clone_attach(struct if_clone *ifc)
933 {
934
935 LIST_INSERT_HEAD(&if_cloners, ifc, ifc_list);
936 if_cloners_count++;
937 }
938
939 /*
940 * Unregister a network interface cloner.
941 */
942 void
943 if_clone_detach(struct if_clone *ifc)
944 {
945
946 LIST_REMOVE(ifc, ifc_list);
947 if_cloners_count--;
948 }
949
950 /*
951 * Provide list of interface cloners to userspace.
952 */
953 static int
954 if_clone_list(struct if_clonereq *ifcr)
955 {
956 char outbuf[IFNAMSIZ], *dst;
957 struct if_clone *ifc;
958 int count, error = 0;
959
960 ifcr->ifcr_total = if_cloners_count;
961 if ((dst = ifcr->ifcr_buffer) == NULL) {
962 /* Just asking how many there are. */
963 return 0;
964 }
965
966 if (ifcr->ifcr_count < 0)
967 return EINVAL;
968
969 count = (if_cloners_count < ifcr->ifcr_count) ?
970 if_cloners_count : ifcr->ifcr_count;
971
972 for (ifc = LIST_FIRST(&if_cloners); ifc != NULL && count != 0;
973 ifc = LIST_NEXT(ifc, ifc_list), count--, dst += IFNAMSIZ) {
974 (void)strncpy(outbuf, ifc->ifc_name, sizeof(outbuf));
975 if (outbuf[sizeof(outbuf) - 1] != '\0')
976 return ENAMETOOLONG;
977 error = copyout(outbuf, dst, sizeof(outbuf));
978 if (error != 0)
979 break;
980 }
981
982 return error;
983 }
984
985 void
986 ifa_insert(struct ifnet *ifp, struct ifaddr *ifa)
987 {
988 ifa->ifa_ifp = ifp;
989 TAILQ_INSERT_TAIL(&ifp->if_addrlist, ifa, ifa_list);
990 IFAREF(ifa);
991 }
992
993 void
994 ifa_remove(struct ifnet *ifp, struct ifaddr *ifa)
995 {
996 KASSERT(ifa->ifa_ifp == ifp);
997 TAILQ_REMOVE(&ifp->if_addrlist, ifa, ifa_list);
998 IFAFREE(ifa);
999 }
1000
1001 static inline int
1002 equal(const struct sockaddr *sa1, const struct sockaddr *sa2)
1003 {
1004 return sockaddr_cmp(sa1, sa2) == 0;
1005 }
1006
1007 /*
1008 * Locate an interface based on a complete address.
1009 */
1010 /*ARGSUSED*/
1011 struct ifaddr *
1012 ifa_ifwithaddr(const struct sockaddr *addr)
1013 {
1014 struct ifnet *ifp;
1015 struct ifaddr *ifa;
1016
1017 IFNET_FOREACH(ifp) {
1018 if (ifp->if_output == if_nulloutput)
1019 continue;
1020 IFADDR_FOREACH(ifa, ifp) {
1021 if (ifa->ifa_addr->sa_family != addr->sa_family)
1022 continue;
1023 if (equal(addr, ifa->ifa_addr))
1024 return ifa;
1025 if ((ifp->if_flags & IFF_BROADCAST) &&
1026 ifa->ifa_broadaddr &&
1027 /* IP6 doesn't have broadcast */
1028 ifa->ifa_broadaddr->sa_len != 0 &&
1029 equal(ifa->ifa_broadaddr, addr))
1030 return ifa;
1031 }
1032 }
1033 return NULL;
1034 }
1035
1036 /*
1037 * Locate the point to point interface with a given destination address.
1038 */
1039 /*ARGSUSED*/
1040 struct ifaddr *
1041 ifa_ifwithdstaddr(const struct sockaddr *addr)
1042 {
1043 struct ifnet *ifp;
1044 struct ifaddr *ifa;
1045
1046 IFNET_FOREACH(ifp) {
1047 if (ifp->if_output == if_nulloutput)
1048 continue;
1049 if ((ifp->if_flags & IFF_POINTOPOINT) == 0)
1050 continue;
1051 IFADDR_FOREACH(ifa, ifp) {
1052 if (ifa->ifa_addr->sa_family != addr->sa_family ||
1053 ifa->ifa_dstaddr == NULL)
1054 continue;
1055 if (equal(addr, ifa->ifa_dstaddr))
1056 return ifa;
1057 }
1058 }
1059 return NULL;
1060 }
1061
1062 /*
1063 * Find an interface on a specific network. If many, choice
1064 * is most specific found.
1065 */
1066 struct ifaddr *
1067 ifa_ifwithnet(const struct sockaddr *addr)
1068 {
1069 struct ifnet *ifp;
1070 struct ifaddr *ifa;
1071 const struct sockaddr_dl *sdl;
1072 struct ifaddr *ifa_maybe = 0;
1073 u_int af = addr->sa_family;
1074 const char *addr_data = addr->sa_data, *cplim;
1075
1076 if (af == AF_LINK) {
1077 sdl = satocsdl(addr);
1078 if (sdl->sdl_index && sdl->sdl_index < if_indexlim &&
1079 ifindex2ifnet[sdl->sdl_index] &&
1080 ifindex2ifnet[sdl->sdl_index]->if_output != if_nulloutput)
1081 return ifnet_addrs[sdl->sdl_index];
1082 }
1083 #ifdef NETATALK
1084 if (af == AF_APPLETALK) {
1085 const struct sockaddr_at *sat, *sat2;
1086 sat = (const struct sockaddr_at *)addr;
1087 IFNET_FOREACH(ifp) {
1088 if (ifp->if_output == if_nulloutput)
1089 continue;
1090 ifa = at_ifawithnet((const struct sockaddr_at *)addr, ifp);
1091 if (ifa == NULL)
1092 continue;
1093 sat2 = (struct sockaddr_at *)ifa->ifa_addr;
1094 if (sat2->sat_addr.s_net == sat->sat_addr.s_net)
1095 return ifa; /* exact match */
1096 if (ifa_maybe == NULL) {
1097 /* else keep the if with the right range */
1098 ifa_maybe = ifa;
1099 }
1100 }
1101 return ifa_maybe;
1102 }
1103 #endif
1104 IFNET_FOREACH(ifp) {
1105 if (ifp->if_output == if_nulloutput)
1106 continue;
1107 IFADDR_FOREACH(ifa, ifp) {
1108 const char *cp, *cp2, *cp3;
1109
1110 if (ifa->ifa_addr->sa_family != af ||
1111 ifa->ifa_netmask == NULL)
1112 next: continue;
1113 cp = addr_data;
1114 cp2 = ifa->ifa_addr->sa_data;
1115 cp3 = ifa->ifa_netmask->sa_data;
1116 cplim = (const char *)ifa->ifa_netmask +
1117 ifa->ifa_netmask->sa_len;
1118 while (cp3 < cplim) {
1119 if ((*cp++ ^ *cp2++) & *cp3++) {
1120 /* want to continue for() loop */
1121 goto next;
1122 }
1123 }
1124 if (ifa_maybe == NULL ||
1125 rn_refines((void *)ifa->ifa_netmask,
1126 (void *)ifa_maybe->ifa_netmask))
1127 ifa_maybe = ifa;
1128 }
1129 }
1130 return ifa_maybe;
1131 }
1132
1133 /*
1134 * Find the interface of the addresss.
1135 */
1136 struct ifaddr *
1137 ifa_ifwithladdr(const struct sockaddr *addr)
1138 {
1139 struct ifaddr *ia;
1140
1141 if ((ia = ifa_ifwithaddr(addr)) || (ia = ifa_ifwithdstaddr(addr)) ||
1142 (ia = ifa_ifwithnet(addr)))
1143 return ia;
1144 return NULL;
1145 }
1146
1147 /*
1148 * Find an interface using a specific address family
1149 */
1150 struct ifaddr *
1151 ifa_ifwithaf(int af)
1152 {
1153 struct ifnet *ifp;
1154 struct ifaddr *ifa;
1155
1156 IFNET_FOREACH(ifp) {
1157 if (ifp->if_output == if_nulloutput)
1158 continue;
1159 IFADDR_FOREACH(ifa, ifp) {
1160 if (ifa->ifa_addr->sa_family == af)
1161 return ifa;
1162 }
1163 }
1164 return NULL;
1165 }
1166
1167 /*
1168 * Find an interface address specific to an interface best matching
1169 * a given address.
1170 */
1171 struct ifaddr *
1172 ifaof_ifpforaddr(const struct sockaddr *addr, struct ifnet *ifp)
1173 {
1174 struct ifaddr *ifa;
1175 const char *cp, *cp2, *cp3;
1176 const char *cplim;
1177 struct ifaddr *ifa_maybe = 0;
1178 u_int af = addr->sa_family;
1179
1180 if (ifp->if_output == if_nulloutput)
1181 return NULL;
1182
1183 if (af >= AF_MAX)
1184 return NULL;
1185
1186 IFADDR_FOREACH(ifa, ifp) {
1187 if (ifa->ifa_addr->sa_family != af)
1188 continue;
1189 ifa_maybe = ifa;
1190 if (ifa->ifa_netmask == NULL) {
1191 if (equal(addr, ifa->ifa_addr) ||
1192 (ifa->ifa_dstaddr &&
1193 equal(addr, ifa->ifa_dstaddr)))
1194 return ifa;
1195 continue;
1196 }
1197 cp = addr->sa_data;
1198 cp2 = ifa->ifa_addr->sa_data;
1199 cp3 = ifa->ifa_netmask->sa_data;
1200 cplim = ifa->ifa_netmask->sa_len + (char *)ifa->ifa_netmask;
1201 for (; cp3 < cplim; cp3++) {
1202 if ((*cp++ ^ *cp2++) & *cp3)
1203 break;
1204 }
1205 if (cp3 == cplim)
1206 return ifa;
1207 }
1208 return ifa_maybe;
1209 }
1210
1211 /*
1212 * Default action when installing a route with a Link Level gateway.
1213 * Lookup an appropriate real ifa to point to.
1214 * This should be moved to /sys/net/link.c eventually.
1215 */
1216 void
1217 link_rtrequest(int cmd, struct rtentry *rt, const struct rt_addrinfo *info)
1218 {
1219 struct ifaddr *ifa;
1220 const struct sockaddr *dst;
1221 struct ifnet *ifp;
1222
1223 if (cmd != RTM_ADD || (ifa = rt->rt_ifa) == NULL ||
1224 (ifp = ifa->ifa_ifp) == NULL || (dst = rt_getkey(rt)) == NULL)
1225 return;
1226 if ((ifa = ifaof_ifpforaddr(dst, ifp)) != NULL) {
1227 rt_replace_ifa(rt, ifa);
1228 if (ifa->ifa_rtrequest && ifa->ifa_rtrequest != link_rtrequest)
1229 ifa->ifa_rtrequest(cmd, rt, info);
1230 }
1231 }
1232
1233 /*
1234 * Handle a change in the interface link state.
1235 */
1236 void
1237 if_link_state_change(struct ifnet *ifp, int link_state)
1238 {
1239 if (ifp->if_link_state == link_state)
1240 return;
1241 ifp->if_link_state = link_state;
1242 /* Notify that the link state has changed. */
1243 rt_ifmsg(ifp);
1244 #if NCARP > 0
1245 if (ifp->if_carp)
1246 carp_carpdev_state(ifp);
1247 #endif
1248 }
1249
1250 /*
1251 * Mark an interface down and notify protocols of
1252 * the transition.
1253 * NOTE: must be called at splsoftnet or equivalent.
1254 */
1255 void
1256 if_down(struct ifnet *ifp)
1257 {
1258 struct ifaddr *ifa;
1259
1260 ifp->if_flags &= ~IFF_UP;
1261 microtime(&ifp->if_lastchange);
1262 IFADDR_FOREACH(ifa, ifp)
1263 pfctlinput(PRC_IFDOWN, ifa->ifa_addr);
1264 IFQ_PURGE(&ifp->if_snd);
1265 #if NCARP > 0
1266 if (ifp->if_carp)
1267 carp_carpdev_state(ifp);
1268 #endif
1269 rt_ifmsg(ifp);
1270 }
1271
1272 /*
1273 * Mark an interface up and notify protocols of
1274 * the transition.
1275 * NOTE: must be called at splsoftnet or equivalent.
1276 */
1277 void
1278 if_up(struct ifnet *ifp)
1279 {
1280 #ifdef notyet
1281 struct ifaddr *ifa;
1282 #endif
1283
1284 ifp->if_flags |= IFF_UP;
1285 microtime(&ifp->if_lastchange);
1286 #ifdef notyet
1287 /* this has no effect on IP, and will kill all ISO connections XXX */
1288 IFADDR_FOREACH(ifa, ifp)
1289 pfctlinput(PRC_IFUP, ifa->ifa_addr);
1290 #endif
1291 #if NCARP > 0
1292 if (ifp->if_carp)
1293 carp_carpdev_state(ifp);
1294 #endif
1295 rt_ifmsg(ifp);
1296 #ifdef INET6
1297 in6_if_up(ifp);
1298 #endif
1299 }
1300
1301 /*
1302 * Handle interface watchdog timer routines. Called
1303 * from softclock, we decrement timers (if set) and
1304 * call the appropriate interface routine on expiration.
1305 */
1306 void
1307 if_slowtimo(void *arg)
1308 {
1309 struct ifnet *ifp;
1310 int s = splnet();
1311
1312 IFNET_FOREACH(ifp) {
1313 if (ifp->if_timer == 0 || --ifp->if_timer)
1314 continue;
1315 if (ifp->if_watchdog != NULL)
1316 (*ifp->if_watchdog)(ifp);
1317 }
1318 splx(s);
1319 callout_reset(&if_slowtimo_ch, hz / IFNET_SLOWHZ, if_slowtimo, NULL);
1320 }
1321
1322 /*
1323 * Set/clear promiscuous mode on interface ifp based on the truth value
1324 * of pswitch. The calls are reference counted so that only the first
1325 * "on" request actually has an effect, as does the final "off" request.
1326 * Results are undefined if the "off" and "on" requests are not matched.
1327 */
1328 int
1329 ifpromisc(struct ifnet *ifp, int pswitch)
1330 {
1331 int pcount, ret;
1332 short flags;
1333 struct ifreq ifr;
1334
1335 pcount = ifp->if_pcount;
1336 flags = ifp->if_flags;
1337 if (pswitch) {
1338 /*
1339 * Allow the device to be "placed" into promiscuous
1340 * mode even if it is not configured up. It will
1341 * consult IFF_PROMISC when it is is brought up.
1342 */
1343 if (ifp->if_pcount++ != 0)
1344 return 0;
1345 ifp->if_flags |= IFF_PROMISC;
1346 if ((ifp->if_flags & IFF_UP) == 0)
1347 return 0;
1348 } else {
1349 if (--ifp->if_pcount > 0)
1350 return 0;
1351 ifp->if_flags &= ~IFF_PROMISC;
1352 /*
1353 * If the device is not configured up, we should not need to
1354 * turn off promiscuous mode (device should have turned it
1355 * off when interface went down; and will look at IFF_PROMISC
1356 * again next time interface comes up).
1357 */
1358 if ((ifp->if_flags & IFF_UP) == 0)
1359 return 0;
1360 }
1361 memset(&ifr, 0, sizeof(ifr));
1362 ifr.ifr_flags = ifp->if_flags;
1363 ret = (*ifp->if_ioctl)(ifp, SIOCSIFFLAGS, &ifr);
1364 /* Restore interface state if not successful. */
1365 if (ret != 0) {
1366 ifp->if_pcount = pcount;
1367 ifp->if_flags = flags;
1368 }
1369 return ret;
1370 }
1371
1372 /*
1373 * Map interface name to
1374 * interface structure pointer.
1375 */
1376 struct ifnet *
1377 ifunit(const char *name)
1378 {
1379 struct ifnet *ifp;
1380 const char *cp = name;
1381 u_int unit = 0;
1382 u_int i;
1383
1384 /*
1385 * If the entire name is a number, treat it as an ifindex.
1386 */
1387 for (i = 0; i < IFNAMSIZ && *cp >= '' && *cp <= '9'; i++, cp++) {
1388 unit = unit * 10 + (*cp - '');
1389 }
1390
1391 /*
1392 * If the number took all of the name, then it's a valid ifindex.
1393 */
1394 if (i == IFNAMSIZ || (cp != name && *cp == '\0')) {
1395 if (unit >= if_indexlim)
1396 return NULL;
1397 ifp = ifindex2ifnet[unit];
1398 if (ifp == NULL || ifp->if_output == if_nulloutput)
1399 return NULL;
1400 return ifp;
1401 }
1402
1403 IFNET_FOREACH(ifp) {
1404 if (ifp->if_output == if_nulloutput)
1405 continue;
1406 if (strcmp(ifp->if_xname, name) == 0)
1407 return ifp;
1408 }
1409 return NULL;
1410 }
1411
1412 /* common */
1413 int
1414 ifioctl_common(struct ifnet *ifp, u_long cmd, void *data)
1415 {
1416 int s;
1417 struct ifreq *ifr;
1418 struct ifcapreq *ifcr;
1419 struct ifdatareq *ifdr;
1420
1421 switch (cmd) {
1422 case SIOCSIFCAP:
1423 ifcr = data;
1424 if ((ifcr->ifcr_capenable & ~ifp->if_capabilities) != 0)
1425 return EINVAL;
1426
1427 if (ifcr->ifcr_capenable == ifp->if_capenable)
1428 return 0;
1429
1430 ifp->if_capenable = ifcr->ifcr_capenable;
1431
1432 /* Pre-compute the checksum flags mask. */
1433 ifp->if_csum_flags_tx = 0;
1434 ifp->if_csum_flags_rx = 0;
1435 if (ifp->if_capenable & IFCAP_CSUM_IPv4_Tx) {
1436 ifp->if_csum_flags_tx |= M_CSUM_IPv4;
1437 }
1438 if (ifp->if_capenable & IFCAP_CSUM_IPv4_Rx) {
1439 ifp->if_csum_flags_rx |= M_CSUM_IPv4;
1440 }
1441
1442 if (ifp->if_capenable & IFCAP_CSUM_TCPv4_Tx) {
1443 ifp->if_csum_flags_tx |= M_CSUM_TCPv4;
1444 }
1445 if (ifp->if_capenable & IFCAP_CSUM_TCPv4_Rx) {
1446 ifp->if_csum_flags_rx |= M_CSUM_TCPv4;
1447 }
1448
1449 if (ifp->if_capenable & IFCAP_CSUM_UDPv4_Tx) {
1450 ifp->if_csum_flags_tx |= M_CSUM_UDPv4;
1451 }
1452 if (ifp->if_capenable & IFCAP_CSUM_UDPv4_Rx) {
1453 ifp->if_csum_flags_rx |= M_CSUM_UDPv4;
1454 }
1455
1456 if (ifp->if_capenable & IFCAP_CSUM_TCPv6_Tx) {
1457 ifp->if_csum_flags_tx |= M_CSUM_TCPv6;
1458 }
1459 if (ifp->if_capenable & IFCAP_CSUM_TCPv6_Rx) {
1460 ifp->if_csum_flags_rx |= M_CSUM_TCPv6;
1461 }
1462
1463 if (ifp->if_capenable & IFCAP_CSUM_UDPv6_Tx) {
1464 ifp->if_csum_flags_tx |= M_CSUM_UDPv6;
1465 }
1466 if (ifp->if_capenable & IFCAP_CSUM_UDPv6_Rx) {
1467 ifp->if_csum_flags_rx |= M_CSUM_UDPv6;
1468 }
1469 if (ifp->if_flags & IFF_UP)
1470 return ENETRESET;
1471 return 0;
1472 case SIOCSIFFLAGS:
1473 ifr = data;
1474 if (ifp->if_flags & IFF_UP && (ifr->ifr_flags & IFF_UP) == 0) {
1475 s = splnet();
1476 if_down(ifp);
1477 splx(s);
1478 }
1479 if (ifr->ifr_flags & IFF_UP && (ifp->if_flags & IFF_UP) == 0) {
1480 s = splnet();
1481 if_up(ifp);
1482 splx(s);
1483 }
1484 ifp->if_flags = (ifp->if_flags & IFF_CANTCHANGE) |
1485 (ifr->ifr_flags &~ IFF_CANTCHANGE);
1486 break;
1487 case SIOCGIFFLAGS:
1488 ifr = data;
1489 ifr->ifr_flags = ifp->if_flags;
1490 break;
1491
1492 case SIOCGIFMETRIC:
1493 ifr = data;
1494 ifr->ifr_metric = ifp->if_metric;
1495 break;
1496
1497 case SIOCGIFMTU:
1498 ifr = data;
1499 ifr->ifr_mtu = ifp->if_mtu;
1500 break;
1501
1502 case SIOCGIFDLT:
1503 ifr = data;
1504 ifr->ifr_dlt = ifp->if_dlt;
1505 break;
1506
1507 case SIOCGIFCAP:
1508 ifcr = data;
1509 ifcr->ifcr_capabilities = ifp->if_capabilities;
1510 ifcr->ifcr_capenable = ifp->if_capenable;
1511 break;
1512
1513 case SIOCSIFMETRIC:
1514 ifr = data;
1515 ifp->if_metric = ifr->ifr_metric;
1516 break;
1517
1518 case SIOCGIFDATA:
1519 ifdr = data;
1520 ifdr->ifdr_data = ifp->if_data;
1521 break;
1522
1523 case SIOCZIFDATA:
1524 ifdr = data;
1525 ifdr->ifdr_data = ifp->if_data;
1526 /*
1527 * Assumes that the volatile counters that can be
1528 * zero'ed are at the end of if_data.
1529 */
1530 memset(&ifp->if_data.ifi_ipackets, 0, sizeof(ifp->if_data) -
1531 offsetof(struct if_data, ifi_ipackets));
1532 break;
1533 case SIOCSIFMTU:
1534 ifr = data;
1535 if (ifp->if_mtu == ifr->ifr_mtu)
1536 break;
1537 ifp->if_mtu = ifr->ifr_mtu;
1538 /*
1539 * If the link MTU changed, do network layer specific procedure.
1540 */
1541 #ifdef INET6
1542 nd6_setmtu(ifp);
1543 #endif
1544 return ENETRESET;
1545 default:
1546 return ENOTTY;
1547 }
1548 return 0;
1549 }
1550
1551 /*
1552 * Interface ioctls.
1553 */
1554 int
1555 ifioctl(struct socket *so, u_long cmd, void *data, struct lwp *l)
1556 {
1557 struct ifnet *ifp;
1558 struct ifreq *ifr;
1559 struct ifcapreq *ifcr;
1560 struct ifdatareq *ifdr;
1561 int error = 0;
1562 #if defined(COMPAT_OSOCK) || defined(COMPAT_OIFREQ)
1563 u_long ocmd = cmd;
1564 #endif
1565 short oif_flags;
1566 #ifdef COMPAT_OIFREQ
1567 struct ifreq ifrb;
1568 struct oifreq *oifr = NULL;
1569 #endif
1570
1571 switch (cmd) {
1572 #ifdef COMPAT_OIFREQ
1573 case OSIOCGIFCONF:
1574 case OOSIOCGIFCONF:
1575 return compat_ifconf(cmd, data);
1576 #endif
1577 case SIOCGIFCONF:
1578 return ifconf(cmd, data);
1579 }
1580
1581 #ifdef COMPAT_OIFREQ
1582 cmd = compat_cvtcmd(cmd);
1583 if (cmd != ocmd) {
1584 oifr = data;
1585 data = ifr = &ifrb;
1586 ifreqo2n(oifr, ifr);
1587 } else
1588 #endif
1589 ifr = data;
1590 ifcr = data;
1591 ifdr = data;
1592
1593 ifp = ifunit(ifr->ifr_name);
1594
1595 switch (cmd) {
1596 case SIOCIFCREATE:
1597 case SIOCIFDESTROY:
1598 if (l != NULL) {
1599 error = kauth_authorize_network(l->l_cred,
1600 KAUTH_NETWORK_INTERFACE,
1601 KAUTH_REQ_NETWORK_INTERFACE_SETPRIV, ifp,
1602 (void *)cmd, NULL);
1603 if (error != 0)
1604 return error;
1605 }
1606 return (cmd == SIOCIFCREATE) ?
1607 if_clone_create(ifr->ifr_name) :
1608 if_clone_destroy(ifr->ifr_name);
1609
1610 case SIOCIFGCLONERS:
1611 return if_clone_list((struct if_clonereq *)data);
1612 }
1613
1614 if (ifp == NULL)
1615 return ENXIO;
1616
1617 switch (cmd) {
1618 case SIOCALIFADDR:
1619 case SIOCDLIFADDR:
1620 case SIOCSIFADDRPREF:
1621 case SIOCSIFFLAGS:
1622 case SIOCSIFCAP:
1623 case SIOCSIFMETRIC:
1624 case SIOCZIFDATA:
1625 case SIOCSIFMTU:
1626 case SIOCSIFPHYADDR:
1627 case SIOCDIFPHYADDR:
1628 #ifdef INET6
1629 case SIOCSIFPHYADDR_IN6:
1630 #endif
1631 case SIOCSLIFPHYADDR:
1632 case SIOCADDMULTI:
1633 case SIOCDELMULTI:
1634 case SIOCSIFMEDIA:
1635 case SIOCSDRVSPEC:
1636 case SIOCG80211:
1637 case SIOCS80211:
1638 case SIOCS80211NWID:
1639 case SIOCS80211NWKEY:
1640 case SIOCS80211POWER:
1641 case SIOCS80211BSSID:
1642 case SIOCS80211CHANNEL:
1643 if (l != NULL) {
1644 error = kauth_authorize_network(l->l_cred,
1645 KAUTH_NETWORK_INTERFACE,
1646 KAUTH_REQ_NETWORK_INTERFACE_SETPRIV, ifp,
1647 (void *)cmd, NULL);
1648 if (error != 0)
1649 return error;
1650 }
1651 }
1652
1653 oif_flags = ifp->if_flags;
1654 switch (cmd) {
1655
1656 case SIOCSIFFLAGS:
1657 ifioctl_common(ifp, cmd, data);
1658 if (ifp->if_ioctl)
1659 (void)(*ifp->if_ioctl)(ifp, cmd, data);
1660 break;
1661
1662 case SIOCSIFPHYADDR:
1663 case SIOCDIFPHYADDR:
1664 #ifdef INET6
1665 case SIOCSIFPHYADDR_IN6:
1666 #endif
1667 case SIOCSLIFPHYADDR:
1668 case SIOCADDMULTI:
1669 case SIOCDELMULTI:
1670 case SIOCSIFMEDIA:
1671 case SIOCGIFPSRCADDR:
1672 case SIOCGIFPDSTADDR:
1673 case SIOCGLIFPHYADDR:
1674 case SIOCGIFMEDIA:
1675 case SIOCG80211:
1676 case SIOCS80211:
1677 case SIOCS80211NWID:
1678 case SIOCS80211NWKEY:
1679 case SIOCS80211POWER:
1680 case SIOCS80211BSSID:
1681 case SIOCS80211CHANNEL:
1682 case SIOCSIFCAP:
1683 case SIOCSIFMTU:
1684 if (ifp->if_ioctl == NULL)
1685 return EOPNOTSUPP;
1686 error = (*ifp->if_ioctl)(ifp, cmd, data);
1687 break;
1688
1689 default:
1690 error = ifioctl_common(ifp, cmd, data);
1691 if (error != ENOTTY)
1692 break;
1693 if (so->so_proto == NULL)
1694 return EOPNOTSUPP;
1695 #ifdef COMPAT_OSOCK
1696 error = compat_ifioctl(so, ocmd, cmd, data, l);
1697 #else
1698 error = (*so->so_proto->pr_usrreq)(so, PRU_CONTROL,
1699 (struct mbuf *)cmd, (struct mbuf *)data,
1700 (struct mbuf *)ifp, l);
1701 #endif
1702 break;
1703 }
1704
1705 if (((oif_flags ^ ifp->if_flags) & IFF_UP) != 0) {
1706 #ifdef INET6
1707 if ((ifp->if_flags & IFF_UP) != 0) {
1708 int s = splnet();
1709 in6_if_up(ifp);
1710 splx(s);
1711 }
1712 #endif
1713 }
1714 #ifdef COMPAT_OIFREQ
1715 if (cmd != ocmd)
1716 ifreqn2o(ifr, oifr);
1717 #endif
1718
1719 return error;
1720 }
1721
1722 /*
1723 * Return interface configuration
1724 * of system. List may be used
1725 * in later ioctl's (above) to get
1726 * other information.
1727 *
1728 * Each record is a struct ifreq. Before the addition of
1729 * sockaddr_storage, the API rule was that sockaddr flavors that did
1730 * not fit would extend beyond the struct ifreq, with the next struct
1731 * ifreq starting sa_len beyond the struct sockaddr. Because the
1732 * union in struct ifreq includes struct sockaddr_storage, every kind
1733 * of sockaddr must fit. Thus, there are no longer any overlength
1734 * records.
1735 *
1736 * Records are added to the user buffer if they fit, and ifc_len is
1737 * adjusted to the length that was written. Thus, the user is only
1738 * assured of getting the complete list if ifc_len on return is at
1739 * least sizeof(struct ifreq) less than it was on entry.
1740 *
1741 * If the user buffer pointer is NULL, this routine copies no data and
1742 * returns the amount of space that would be needed.
1743 *
1744 * Invariants:
1745 * ifrp points to the next part of the user's buffer to be used. If
1746 * ifrp != NULL, space holds the number of bytes remaining that we may
1747 * write at ifrp. Otherwise, space holds the number of bytes that
1748 * would have been written had there been adequate space.
1749 */
1750 /*ARGSUSED*/
1751 int
1752 ifconf(u_long cmd, void *data)
1753 {
1754 struct ifconf *ifc = (struct ifconf *)data;
1755 struct ifnet *ifp;
1756 struct ifaddr *ifa;
1757 struct ifreq ifr, *ifrp;
1758 int space, error = 0;
1759 const int sz = (int)sizeof(struct ifreq);
1760
1761 if ((ifrp = ifc->ifc_req) == NULL)
1762 space = 0;
1763 else
1764 space = ifc->ifc_len;
1765 IFNET_FOREACH(ifp) {
1766 (void)strncpy(ifr.ifr_name, ifp->if_xname,
1767 sizeof(ifr.ifr_name));
1768 if (ifr.ifr_name[sizeof(ifr.ifr_name) - 1] != '\0')
1769 return ENAMETOOLONG;
1770 if (IFADDR_EMPTY(ifp)) {
1771 /* Interface with no addresses - send zero sockaddr. */
1772 memset(&ifr.ifr_addr, 0, sizeof(ifr.ifr_addr));
1773 if (ifrp == NULL) {
1774 space += sz;
1775 continue;
1776 }
1777 if (space >= sz) {
1778 error = copyout(&ifr, ifrp, sz);
1779 if (error != 0)
1780 return error;
1781 ifrp++;
1782 space -= sz;
1783 }
1784 }
1785
1786 IFADDR_FOREACH(ifa, ifp) {
1787 struct sockaddr *sa = ifa->ifa_addr;
1788 /* all sockaddrs must fit in sockaddr_storage */
1789 KASSERT(sa->sa_len <= sizeof(ifr.ifr_ifru));
1790
1791 if (ifrp == NULL) {
1792 space += sz;
1793 continue;
1794 }
1795 memcpy(&ifr.ifr_space, sa, sa->sa_len);
1796 if (space >= sz) {
1797 error = copyout(&ifr, ifrp, sz);
1798 if (error != 0)
1799 return (error);
1800 ifrp++; space -= sz;
1801 }
1802 }
1803 }
1804 if (ifrp != NULL) {
1805 KASSERT(0 <= space && space <= ifc->ifc_len);
1806 ifc->ifc_len -= space;
1807 } else {
1808 KASSERT(space >= 0);
1809 ifc->ifc_len = space;
1810 }
1811 return (0);
1812 }
1813
1814 int
1815 ifreq_setaddr(u_long cmd, struct ifreq *ifr, const struct sockaddr *sa)
1816 {
1817 uint8_t len;
1818 #ifdef COMPAT_OIFREQ
1819 struct ifreq ifrb;
1820 struct oifreq *oifr = NULL;
1821 u_long ocmd = cmd;
1822 cmd = compat_cvtcmd(cmd);
1823 if (cmd != ocmd) {
1824 oifr = (struct oifreq *)(void *)ifr;
1825 ifr = &ifrb;
1826 ifreqo2n(oifr, ifr);
1827 len = sizeof(oifr->ifr_addr);
1828 } else
1829 #endif
1830 len = sizeof(ifr->ifr_ifru.ifru_space);
1831
1832 if (len < sa->sa_len)
1833 return EFBIG;
1834
1835 memset(&ifr->ifr_addr, 0, len);
1836 sockaddr_copy(&ifr->ifr_addr, len, sa);
1837
1838 #ifdef COMPAT_OIFREQ
1839 if (cmd != ocmd)
1840 ifreqn2o(oifr, ifr);
1841 #endif
1842 return 0;
1843 }
1844
1845 /*
1846 * Queue message on interface, and start output if interface
1847 * not yet active.
1848 */
1849 int
1850 ifq_enqueue(struct ifnet *ifp, struct mbuf *m
1851 ALTQ_COMMA ALTQ_DECL(struct altq_pktattr *pktattr))
1852 {
1853 int len = m->m_pkthdr.len;
1854 int mflags = m->m_flags;
1855 int s = splnet();
1856 int error;
1857
1858 IFQ_ENQUEUE(&ifp->if_snd, m, pktattr, error);
1859 if (error != 0)
1860 goto out;
1861 ifp->if_obytes += len;
1862 if (mflags & M_MCAST)
1863 ifp->if_omcasts++;
1864 if ((ifp->if_flags & IFF_OACTIVE) == 0)
1865 (*ifp->if_start)(ifp);
1866 out:
1867 splx(s);
1868 return error;
1869 }
1870
1871 /*
1872 * Queue message on interface, possibly using a second fast queue
1873 */
1874 int
1875 ifq_enqueue2(struct ifnet *ifp, struct ifqueue *ifq, struct mbuf *m
1876 ALTQ_COMMA ALTQ_DECL(struct altq_pktattr *pktattr))
1877 {
1878 int error = 0;
1879
1880 if (ifq != NULL
1881 #ifdef ALTQ
1882 && ALTQ_IS_ENABLED(&ifp->if_snd) == 0
1883 #endif
1884 ) {
1885 if (IF_QFULL(ifq)) {
1886 IF_DROP(&ifp->if_snd);
1887 m_freem(m);
1888 if (error == 0)
1889 error = ENOBUFS;
1890 } else
1891 IF_ENQUEUE(ifq, m);
1892 } else
1893 IFQ_ENQUEUE(&ifp->if_snd, m, pktattr, error);
1894 if (error != 0) {
1895 ++ifp->if_oerrors;
1896 return error;
1897 }
1898 return 0;
1899 }
1900
1901
1902 #if defined(INET) || defined(INET6)
1903 static void
1904 sysctl_net_ifq_setup(struct sysctllog **clog,
1905 int pf, const char *pfname,
1906 int ipn, const char *ipname,
1907 int qid, struct ifqueue *ifq)
1908 {
1909
1910 sysctl_createv(clog, 0, NULL, NULL,
1911 CTLFLAG_PERMANENT,
1912 CTLTYPE_NODE, "net", NULL,
1913 NULL, 0, NULL, 0,
1914 CTL_NET, CTL_EOL);
1915 sysctl_createv(clog, 0, NULL, NULL,
1916 CTLFLAG_PERMANENT,
1917 CTLTYPE_NODE, pfname, NULL,
1918 NULL, 0, NULL, 0,
1919 CTL_NET, pf, CTL_EOL);
1920 sysctl_createv(clog, 0, NULL, NULL,
1921 CTLFLAG_PERMANENT,
1922 CTLTYPE_NODE, ipname, NULL,
1923 NULL, 0, NULL, 0,
1924 CTL_NET, pf, ipn, CTL_EOL);
1925 sysctl_createv(clog, 0, NULL, NULL,
1926 CTLFLAG_PERMANENT,
1927 CTLTYPE_NODE, "ifq",
1928 SYSCTL_DESCR("Protocol input queue controls"),
1929 NULL, 0, NULL, 0,
1930 CTL_NET, pf, ipn, qid, CTL_EOL);
1931
1932 sysctl_createv(clog, 0, NULL, NULL,
1933 CTLFLAG_PERMANENT,
1934 CTLTYPE_INT, "len",
1935 SYSCTL_DESCR("Current input queue length"),
1936 NULL, 0, &ifq->ifq_len, 0,
1937 CTL_NET, pf, ipn, qid, IFQCTL_LEN, CTL_EOL);
1938 sysctl_createv(clog, 0, NULL, NULL,
1939 CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1940 CTLTYPE_INT, "maxlen",
1941 SYSCTL_DESCR("Maximum allowed input queue length"),
1942 NULL, 0, &ifq->ifq_maxlen, 0,
1943 CTL_NET, pf, ipn, qid, IFQCTL_MAXLEN, CTL_EOL);
1944 #ifdef notyet
1945 sysctl_createv(clog, 0, NULL, NULL,
1946 CTLFLAG_PERMANENT,
1947 CTLTYPE_INT, "peak",
1948 SYSCTL_DESCR("Highest input queue length"),
1949 NULL, 0, &ifq->ifq_peak, 0,
1950 CTL_NET, pf, ipn, qid, IFQCTL_PEAK, CTL_EOL);
1951 #endif
1952 sysctl_createv(clog, 0, NULL, NULL,
1953 CTLFLAG_PERMANENT,
1954 CTLTYPE_INT, "drops",
1955 SYSCTL_DESCR("Packets dropped due to full input queue"),
1956 NULL, 0, &ifq->ifq_drops, 0,
1957 CTL_NET, pf, ipn, qid, IFQCTL_DROPS, CTL_EOL);
1958 }
1959
1960 #ifdef INET
1961 SYSCTL_SETUP(sysctl_net_inet_ip_ifq_setup,
1962 "sysctl net.inet.ip.ifq subtree setup")
1963 {
1964 extern struct ifqueue ipintrq;
1965
1966 sysctl_net_ifq_setup(clog, PF_INET, "inet", IPPROTO_IP, "ip",
1967 IPCTL_IFQ, &ipintrq);
1968 }
1969 #endif /* INET */
1970
1971 #ifdef INET6
1972 SYSCTL_SETUP(sysctl_net_inet6_ip6_ifq_setup,
1973 "sysctl net.inet6.ip6.ifq subtree setup")
1974 {
1975 extern struct ifqueue ip6intrq;
1976
1977 sysctl_net_ifq_setup(clog, PF_INET6, "inet6", IPPROTO_IPV6, "ip6",
1978 IPV6CTL_IFQ, &ip6intrq);
1979 }
1980 #endif /* INET6 */
1981 #endif /* INET || INET6 */
Cache object: 9055b9f96a382a623d34fd87822abdc2
|