if.c revision 1.271 1 /* $NetBSD: if.c,v 1.271 2014/01/03 12:49:59 pooka 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 1.271 2014/01/03 12:49:59 pooka Exp $");
94
95 #include "opt_inet.h"
96
97 #include "opt_atalk.h"
98 #include "opt_natm.h"
99
100 #include <sys/param.h>
101 #include <sys/mbuf.h>
102 #include <sys/systm.h>
103 #include <sys/callout.h>
104 #include <sys/proc.h>
105 #include <sys/socket.h>
106 #include <sys/socketvar.h>
107 #include <sys/domain.h>
108 #include <sys/protosw.h>
109 #include <sys/kernel.h>
110 #include <sys/ioctl.h>
111 #include <sys/sysctl.h>
112 #include <sys/syslog.h>
113 #include <sys/kauth.h>
114 #include <sys/kmem.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 #include <sys/module.h>
127 #ifdef NETATALK
128 #include <netatalk/at_extern.h>
129 #include <netatalk/at.h>
130 #endif
131 #include <net/pfil.h>
132
133 #ifdef INET6
134 #include <netinet/in.h>
135 #include <netinet6/in6_var.h>
136 #include <netinet6/nd6.h>
137 #endif
138
139 #include "carp.h"
140 #if NCARP > 0
141 #include <netinet/ip_carp.h>
142 #endif
143
144 #include <compat/sys/sockio.h>
145 #include <compat/sys/socket.h>
146
147 MALLOC_DEFINE(M_IFADDR, "ifaddr", "interface address");
148 MALLOC_DEFINE(M_IFMADDR, "ether_multi", "link-level multicast address");
149
150 int ifqmaxlen = IFQ_MAXLEN;
151 callout_t if_slowtimo_ch;
152
153 int netisr; /* scheduling bits for network */
154
155 static int if_rt_walktree(struct rtentry *, void *);
156
157 static struct if_clone *if_clone_lookup(const char *, int *);
158 static int if_clone_list(struct if_clonereq *);
159
160 static LIST_HEAD(, if_clone) if_cloners = LIST_HEAD_INITIALIZER(if_cloners);
161 static int if_cloners_count;
162
163 static uint64_t index_gen;
164 static kmutex_t index_gen_mtx;
165
166 /* Packet filtering hook for interfaces. */
167 pfil_head_t * if_pfil;
168
169 static kauth_listener_t if_listener;
170
171 static int ifioctl_attach(struct ifnet *);
172 static void ifioctl_detach(struct ifnet *);
173 static void ifnet_lock_enter(struct ifnet_lock *);
174 static void ifnet_lock_exit(struct ifnet_lock *);
175 static void if_detach_queues(struct ifnet *, struct ifqueue *);
176 static void sysctl_sndq_setup(struct sysctllog **, const char *,
177 struct ifaltq *);
178
179 #if defined(INET) || defined(INET6)
180 static void sysctl_net_ifq_setup(struct sysctllog **, int, const char *,
181 int, const char *, int, struct ifqueue *);
182 #endif
183
184 static int
185 if_listener_cb(kauth_cred_t cred, kauth_action_t action, void *cookie,
186 void *arg0, void *arg1, void *arg2, void *arg3)
187 {
188 int result;
189 enum kauth_network_req req;
190
191 result = KAUTH_RESULT_DEFER;
192 req = (enum kauth_network_req)arg1;
193
194 if (action != KAUTH_NETWORK_INTERFACE)
195 return result;
196
197 if ((req == KAUTH_REQ_NETWORK_INTERFACE_GET) ||
198 (req == KAUTH_REQ_NETWORK_INTERFACE_SET))
199 result = KAUTH_RESULT_ALLOW;
200
201 return result;
202 }
203
204 /*
205 * Network interface utility routines.
206 *
207 * Routines with ifa_ifwith* names take sockaddr *'s as
208 * parameters.
209 */
210 void
211 ifinit(void)
212 {
213 #ifdef INET
214 {extern struct ifqueue ipintrq;
215 sysctl_net_ifq_setup(NULL, PF_INET, "inet", IPPROTO_IP, "ip",
216 IPCTL_IFQ, &ipintrq);}
217 #endif /* INET */
218 #ifdef INET6
219 {extern struct ifqueue ip6intrq;
220 sysctl_net_ifq_setup(NULL, PF_INET6, "inet6", IPPROTO_IPV6, "ip6",
221 IPV6CTL_IFQ, &ip6intrq);}
222 #endif /* INET6 */
223
224 callout_init(&if_slowtimo_ch, 0);
225 if_slowtimo(NULL);
226
227 if_listener = kauth_listen_scope(KAUTH_SCOPE_NETWORK,
228 if_listener_cb, NULL);
229 }
230
231 /*
232 * XXX Initialization before configure().
233 * XXX hack to get pfil_add_hook working in autoconf.
234 */
235 void
236 ifinit1(void)
237 {
238 mutex_init(&index_gen_mtx, MUTEX_DEFAULT, IPL_NONE);
239 if_pfil = pfil_head_create(PFIL_TYPE_IFNET, NULL);
240 KASSERT(if_pfil != NULL);
241 }
242
243 struct ifnet *
244 if_alloc(u_char type)
245 {
246 return malloc(sizeof(struct ifnet), M_DEVBUF, M_WAITOK|M_ZERO);
247 }
248
249 void
250 if_free(struct ifnet *ifp)
251 {
252 free(ifp, M_DEVBUF);
253 }
254
255 void
256 if_initname(struct ifnet *ifp, const char *name, int unit)
257 {
258 (void)snprintf(ifp->if_xname, sizeof(ifp->if_xname),
259 "%s%d", name, unit);
260 }
261
262 /*
263 * Null routines used while an interface is going away. These routines
264 * just return an error.
265 */
266
267 int
268 if_nulloutput(struct ifnet *ifp, struct mbuf *m,
269 const struct sockaddr *so, struct rtentry *rt)
270 {
271
272 return ENXIO;
273 }
274
275 void
276 if_nullinput(struct ifnet *ifp, struct mbuf *m)
277 {
278
279 /* Nothing. */
280 }
281
282 void
283 if_nullstart(struct ifnet *ifp)
284 {
285
286 /* Nothing. */
287 }
288
289 int
290 if_nullioctl(struct ifnet *ifp, u_long cmd, void *data)
291 {
292
293 /* Wake ifioctl_detach(), who may wait for all threads to
294 * quit the critical section.
295 */
296 cv_signal(&ifp->if_ioctl_lock->il_emptied);
297 return ENXIO;
298 }
299
300 int
301 if_nullinit(struct ifnet *ifp)
302 {
303
304 return ENXIO;
305 }
306
307 void
308 if_nullstop(struct ifnet *ifp, int disable)
309 {
310
311 /* Nothing. */
312 }
313
314 void
315 if_nullwatchdog(struct ifnet *ifp)
316 {
317
318 /* Nothing. */
319 }
320
321 void
322 if_nulldrain(struct ifnet *ifp)
323 {
324
325 /* Nothing. */
326 }
327
328 static u_int if_index = 1;
329 struct ifnet_head ifnet;
330 size_t if_indexlim = 0;
331 struct ifaddr **ifnet_addrs = NULL;
332 struct ifnet **ifindex2ifnet = NULL;
333 struct ifnet *lo0ifp;
334
335 void
336 if_set_sadl(struct ifnet *ifp, const void *lla, u_char addrlen, bool factory)
337 {
338 struct ifaddr *ifa;
339 struct sockaddr_dl *sdl;
340
341 ifp->if_addrlen = addrlen;
342 if_alloc_sadl(ifp);
343 ifa = ifp->if_dl;
344 sdl = satosdl(ifa->ifa_addr);
345
346 (void)sockaddr_dl_setaddr(sdl, sdl->sdl_len, lla, ifp->if_addrlen);
347 if (factory) {
348 ifp->if_hwdl = ifp->if_dl;
349 IFAREF(ifp->if_hwdl);
350 }
351 /* TBD routing socket */
352 }
353
354 struct ifaddr *
355 if_dl_create(const struct ifnet *ifp, const struct sockaddr_dl **sdlp)
356 {
357 unsigned socksize, ifasize;
358 int addrlen, namelen;
359 struct sockaddr_dl *mask, *sdl;
360 struct ifaddr *ifa;
361
362 namelen = strlen(ifp->if_xname);
363 addrlen = ifp->if_addrlen;
364 socksize = roundup(sockaddr_dl_measure(namelen, addrlen), sizeof(long));
365 ifasize = sizeof(*ifa) + 2 * socksize;
366 ifa = (struct ifaddr *)malloc(ifasize, M_IFADDR, M_WAITOK|M_ZERO);
367
368 sdl = (struct sockaddr_dl *)(ifa + 1);
369 mask = (struct sockaddr_dl *)(socksize + (char *)sdl);
370
371 sockaddr_dl_init(sdl, socksize, ifp->if_index, ifp->if_type,
372 ifp->if_xname, namelen, NULL, addrlen);
373 mask->sdl_len = sockaddr_dl_measure(namelen, 0);
374 memset(&mask->sdl_data[0], 0xff, namelen);
375 ifa->ifa_rtrequest = link_rtrequest;
376 ifa->ifa_addr = (struct sockaddr *)sdl;
377 ifa->ifa_netmask = (struct sockaddr *)mask;
378
379 *sdlp = sdl;
380
381 return ifa;
382 }
383
384 static void
385 if_sadl_setrefs(struct ifnet *ifp, struct ifaddr *ifa)
386 {
387 const struct sockaddr_dl *sdl;
388 ifnet_addrs[ifp->if_index] = ifa;
389 IFAREF(ifa);
390 ifp->if_dl = ifa;
391 IFAREF(ifa);
392 sdl = satosdl(ifa->ifa_addr);
393 ifp->if_sadl = sdl;
394 }
395
396 /*
397 * Allocate the link level name for the specified interface. This
398 * is an attachment helper. It must be called after ifp->if_addrlen
399 * is initialized, which may not be the case when if_attach() is
400 * called.
401 */
402 void
403 if_alloc_sadl(struct ifnet *ifp)
404 {
405 struct ifaddr *ifa;
406 const struct sockaddr_dl *sdl;
407
408 /*
409 * If the interface already has a link name, release it
410 * now. This is useful for interfaces that can change
411 * link types, and thus switch link names often.
412 */
413 if (ifp->if_sadl != NULL)
414 if_free_sadl(ifp);
415
416 ifa = if_dl_create(ifp, &sdl);
417
418 ifa_insert(ifp, ifa);
419 if_sadl_setrefs(ifp, ifa);
420 }
421
422 static void
423 if_deactivate_sadl(struct ifnet *ifp)
424 {
425 struct ifaddr *ifa;
426
427 KASSERT(ifp->if_dl != NULL);
428
429 ifa = ifp->if_dl;
430
431 ifp->if_sadl = NULL;
432
433 ifnet_addrs[ifp->if_index] = NULL;
434 IFAFREE(ifa);
435 ifp->if_dl = NULL;
436 IFAFREE(ifa);
437 }
438
439 void
440 if_activate_sadl(struct ifnet *ifp, struct ifaddr *ifa,
441 const struct sockaddr_dl *sdl)
442 {
443 int s;
444
445 s = splnet();
446
447 if_deactivate_sadl(ifp);
448
449 if_sadl_setrefs(ifp, ifa);
450 IFADDR_FOREACH(ifa, ifp)
451 rtinit(ifa, RTM_LLINFO_UPD, 0);
452 splx(s);
453 }
454
455 /*
456 * Free the link level name for the specified interface. This is
457 * a detach helper. This is called from if_detach() or from
458 * link layer type specific detach functions.
459 */
460 void
461 if_free_sadl(struct ifnet *ifp)
462 {
463 struct ifaddr *ifa;
464 int s;
465
466 ifa = ifnet_addrs[ifp->if_index];
467 if (ifa == NULL) {
468 KASSERT(ifp->if_sadl == NULL);
469 KASSERT(ifp->if_dl == NULL);
470 return;
471 }
472
473 KASSERT(ifp->if_sadl != NULL);
474 KASSERT(ifp->if_dl != NULL);
475
476 s = splnet();
477 rtinit(ifa, RTM_DELETE, 0);
478 ifa_remove(ifp, ifa);
479 if_deactivate_sadl(ifp);
480 if (ifp->if_hwdl == ifa) {
481 IFAFREE(ifa);
482 ifp->if_hwdl = NULL;
483 }
484 splx(s);
485 }
486
487 /*
488 * Attach an interface to the
489 * list of "active" interfaces.
490 */
491 void
492 if_attach(struct ifnet *ifp)
493 {
494 int indexlim = 0;
495
496 if (if_indexlim == 0) {
497 TAILQ_INIT(&ifnet);
498 if_indexlim = 8;
499 }
500 TAILQ_INIT(&ifp->if_addrlist);
501 TAILQ_INSERT_TAIL(&ifnet, ifp, if_list);
502
503 if (ifioctl_attach(ifp) != 0)
504 panic("%s: ifioctl_attach() failed", __func__);
505
506 mutex_enter(&index_gen_mtx);
507 ifp->if_index_gen = index_gen++;
508 mutex_exit(&index_gen_mtx);
509
510 ifp->if_index = if_index;
511 if (ifindex2ifnet == NULL)
512 if_index++;
513 else
514 while (ifp->if_index < if_indexlim &&
515 ifindex2ifnet[ifp->if_index] != NULL) {
516 ++if_index;
517 if (if_index == 0)
518 if_index = 1;
519 /*
520 * If we hit USHRT_MAX, we skip back to 0 since
521 * there are a number of places where the value
522 * of if_index or if_index itself is compared
523 * to or stored in an unsigned short. By
524 * jumping back, we won't botch those assignments
525 * or comparisons.
526 */
527 else if (if_index == USHRT_MAX) {
528 /*
529 * However, if we have to jump back to
530 * zero *twice* without finding an empty
531 * slot in ifindex2ifnet[], then there
532 * there are too many (>65535) interfaces.
533 */
534 if (indexlim++)
535 panic("too many interfaces");
536 else
537 if_index = 1;
538 }
539 ifp->if_index = if_index;
540 }
541
542 /*
543 * We have some arrays that should be indexed by if_index.
544 * since if_index will grow dynamically, they should grow too.
545 * struct ifadd **ifnet_addrs
546 * struct ifnet **ifindex2ifnet
547 */
548 if (ifnet_addrs == NULL || ifindex2ifnet == NULL ||
549 ifp->if_index >= if_indexlim) {
550 size_t m, n, oldlim;
551 void *q;
552
553 oldlim = if_indexlim;
554 while (ifp->if_index >= if_indexlim)
555 if_indexlim <<= 1;
556
557 /* grow ifnet_addrs */
558 m = oldlim * sizeof(struct ifaddr *);
559 n = if_indexlim * sizeof(struct ifaddr *);
560 q = malloc(n, M_IFADDR, M_WAITOK|M_ZERO);
561 if (ifnet_addrs != NULL) {
562 memcpy(q, ifnet_addrs, m);
563 free(ifnet_addrs, M_IFADDR);
564 }
565 ifnet_addrs = (struct ifaddr **)q;
566
567 /* grow ifindex2ifnet */
568 m = oldlim * sizeof(struct ifnet *);
569 n = if_indexlim * sizeof(struct ifnet *);
570 q = malloc(n, M_IFADDR, M_WAITOK|M_ZERO);
571 if (ifindex2ifnet != NULL) {
572 memcpy(q, ifindex2ifnet, m);
573 free(ifindex2ifnet, M_IFADDR);
574 }
575 ifindex2ifnet = (struct ifnet **)q;
576 }
577
578 ifindex2ifnet[ifp->if_index] = ifp;
579
580 /*
581 * Link level name is allocated later by a separate call to
582 * if_alloc_sadl().
583 */
584
585 if (ifp->if_snd.ifq_maxlen == 0)
586 ifp->if_snd.ifq_maxlen = ifqmaxlen;
587
588 sysctl_sndq_setup(&ifp->if_sysctl_log, ifp->if_xname, &ifp->if_snd);
589
590 ifp->if_broadcastaddr = 0; /* reliably crash if used uninitialized */
591
592 ifp->if_link_state = LINK_STATE_UNKNOWN;
593
594 ifp->if_capenable = 0;
595 ifp->if_csum_flags_tx = 0;
596 ifp->if_csum_flags_rx = 0;
597
598 #ifdef ALTQ
599 ifp->if_snd.altq_type = 0;
600 ifp->if_snd.altq_disc = NULL;
601 ifp->if_snd.altq_flags &= ALTQF_CANTCHANGE;
602 ifp->if_snd.altq_tbr = NULL;
603 ifp->if_snd.altq_ifp = ifp;
604 #endif
605
606 ifp->if_pfil = pfil_head_create(PFIL_TYPE_IFNET, ifp);
607 (void)pfil_run_hooks(if_pfil,
608 (struct mbuf **)PFIL_IFNET_ATTACH, ifp, PFIL_IFNET);
609
610 if (!STAILQ_EMPTY(&domains))
611 if_attachdomain1(ifp);
612
613 /* Announce the interface. */
614 rt_ifannouncemsg(ifp, IFAN_ARRIVAL);
615 }
616
617 void
618 if_attachdomain(void)
619 {
620 struct ifnet *ifp;
621 int s;
622
623 s = splnet();
624 IFNET_FOREACH(ifp)
625 if_attachdomain1(ifp);
626 splx(s);
627 }
628
629 void
630 if_attachdomain1(struct ifnet *ifp)
631 {
632 struct domain *dp;
633 int s;
634
635 s = splnet();
636
637 /* address family dependent data region */
638 memset(ifp->if_afdata, 0, sizeof(ifp->if_afdata));
639 DOMAIN_FOREACH(dp) {
640 if (dp->dom_ifattach != NULL)
641 ifp->if_afdata[dp->dom_family] =
642 (*dp->dom_ifattach)(ifp);
643 }
644
645 splx(s);
646 }
647
648 /*
649 * Deactivate an interface. This points all of the procedure
650 * handles at error stubs. May be called from interrupt context.
651 */
652 void
653 if_deactivate(struct ifnet *ifp)
654 {
655 int s;
656
657 s = splnet();
658
659 ifp->if_output = if_nulloutput;
660 ifp->if_input = if_nullinput;
661 ifp->if_start = if_nullstart;
662 ifp->if_ioctl = if_nullioctl;
663 ifp->if_init = if_nullinit;
664 ifp->if_stop = if_nullstop;
665 ifp->if_watchdog = if_nullwatchdog;
666 ifp->if_drain = if_nulldrain;
667
668 /* No more packets may be enqueued. */
669 ifp->if_snd.ifq_maxlen = 0;
670
671 splx(s);
672 }
673
674 void
675 if_purgeaddrs(struct ifnet *ifp, int family, void (*purgeaddr)(struct ifaddr *))
676 {
677 struct ifaddr *ifa, *nifa;
678
679 for (ifa = IFADDR_FIRST(ifp); ifa != NULL; ifa = nifa) {
680 nifa = IFADDR_NEXT(ifa);
681 if (ifa->ifa_addr->sa_family != family)
682 continue;
683 (*purgeaddr)(ifa);
684 }
685 }
686
687 /*
688 * Detach an interface from the list of "active" interfaces,
689 * freeing any resources as we go along.
690 *
691 * NOTE: This routine must be called with a valid thread context,
692 * as it may block.
693 */
694 void
695 if_detach(struct ifnet *ifp)
696 {
697 struct socket so;
698 struct ifaddr *ifa;
699 #ifdef IFAREF_DEBUG
700 struct ifaddr *last_ifa = NULL;
701 #endif
702 struct domain *dp;
703 const struct protosw *pr;
704 int s, i, family, purged;
705
706 /*
707 * XXX It's kind of lame that we have to have the
708 * XXX socket structure...
709 */
710 memset(&so, 0, sizeof(so));
711
712 s = splnet();
713
714 /*
715 * Do an if_down() to give protocols a chance to do something.
716 */
717 if_down(ifp);
718
719 #ifdef ALTQ
720 if (ALTQ_IS_ENABLED(&ifp->if_snd))
721 altq_disable(&ifp->if_snd);
722 if (ALTQ_IS_ATTACHED(&ifp->if_snd))
723 altq_detach(&ifp->if_snd);
724 #endif
725
726 sysctl_teardown(&ifp->if_sysctl_log);
727
728 #if NCARP > 0
729 /* Remove the interface from any carp group it is a part of. */
730 if (ifp->if_carp != NULL && ifp->if_type != IFT_CARP)
731 carp_ifdetach(ifp);
732 #endif
733
734 /*
735 * Rip all the addresses off the interface. This should make
736 * all of the routes go away.
737 *
738 * pr_usrreq calls can remove an arbitrary number of ifaddrs
739 * from the list, including our "cursor", ifa. For safety,
740 * and to honor the TAILQ abstraction, I just restart the
741 * loop after each removal. Note that the loop will exit
742 * when all of the remaining ifaddrs belong to the AF_LINK
743 * family. I am counting on the historical fact that at
744 * least one pr_usrreq in each address domain removes at
745 * least one ifaddr.
746 */
747 again:
748 IFADDR_FOREACH(ifa, ifp) {
749 family = ifa->ifa_addr->sa_family;
750 #ifdef IFAREF_DEBUG
751 printf("if_detach: ifaddr %p, family %d, refcnt %d\n",
752 ifa, family, ifa->ifa_refcnt);
753 if (last_ifa != NULL && ifa == last_ifa)
754 panic("if_detach: loop detected");
755 last_ifa = ifa;
756 #endif
757 if (family == AF_LINK)
758 continue;
759 dp = pffinddomain(family);
760 #ifdef DIAGNOSTIC
761 if (dp == NULL)
762 panic("if_detach: no domain for AF %d",
763 family);
764 #endif
765 /*
766 * XXX These PURGEIF calls are redundant with the
767 * purge-all-families calls below, but are left in for
768 * now both to make a smaller change, and to avoid
769 * unplanned interactions with clearing of
770 * ifp->if_addrlist.
771 */
772 purged = 0;
773 for (pr = dp->dom_protosw;
774 pr < dp->dom_protoswNPROTOSW; pr++) {
775 so.so_proto = pr;
776 if (pr->pr_usrreq != NULL) {
777 (void) (*pr->pr_usrreq)(&so,
778 PRU_PURGEIF, NULL, NULL,
779 (struct mbuf *) ifp, curlwp);
780 purged = 1;
781 }
782 }
783 if (purged == 0) {
784 /*
785 * XXX What's really the best thing to do
786 * XXX here? --thorpej (at) NetBSD.org
787 */
788 printf("if_detach: WARNING: AF %d not purged\n",
789 family);
790 ifa_remove(ifp, ifa);
791 }
792 goto again;
793 }
794
795 if_free_sadl(ifp);
796
797 /* Walk the routing table looking for stragglers. */
798 for (i = 0; i <= AF_MAX; i++) {
799 while (rt_walktree(i, if_rt_walktree, ifp) == ERESTART)
800 continue;
801 }
802
803 DOMAIN_FOREACH(dp) {
804 if (dp->dom_ifdetach != NULL && ifp->if_afdata[dp->dom_family])
805 {
806 void *p = ifp->if_afdata[dp->dom_family];
807 if (p) {
808 ifp->if_afdata[dp->dom_family] = NULL;
809 (*dp->dom_ifdetach)(ifp, p);
810 }
811 }
812
813 /*
814 * One would expect multicast memberships (INET and
815 * INET6) on UDP sockets to be purged by the PURGEIF
816 * calls above, but if all addresses were removed from
817 * the interface prior to destruction, the calls will
818 * not be made (e.g. ppp, for which pppd(8) generally
819 * removes addresses before destroying the interface).
820 * Because there is no invariant that multicast
821 * memberships only exist for interfaces with IPv4
822 * addresses, we must call PURGEIF regardless of
823 * addresses. (Protocols which might store ifnet
824 * pointers are marked with PR_PURGEIF.)
825 */
826 for (pr = dp->dom_protosw; pr < dp->dom_protoswNPROTOSW; pr++) {
827 so.so_proto = pr;
828 if (pr->pr_usrreq != NULL && pr->pr_flags & PR_PURGEIF)
829 (void)(*pr->pr_usrreq)(&so, PRU_PURGEIF, NULL,
830 NULL, (struct mbuf *)ifp, curlwp);
831 }
832 }
833
834 (void)pfil_run_hooks(if_pfil,
835 (struct mbuf **)PFIL_IFNET_DETACH, ifp, PFIL_IFNET);
836 (void)pfil_head_destroy(ifp->if_pfil);
837
838 /* Announce that the interface is gone. */
839 rt_ifannouncemsg(ifp, IFAN_DEPARTURE);
840
841 ifindex2ifnet[ifp->if_index] = NULL;
842
843 TAILQ_REMOVE(&ifnet, ifp, if_list);
844
845 ifioctl_detach(ifp);
846
847 /*
848 * remove packets that came from ifp, from software interrupt queues.
849 */
850 DOMAIN_FOREACH(dp) {
851 for (i = 0; i < __arraycount(dp->dom_ifqueues); i++) {
852 struct ifqueue *iq = dp->dom_ifqueues[i];
853 if (iq == NULL)
854 break;
855 dp->dom_ifqueues[i] = NULL;
856 if_detach_queues(ifp, iq);
857 }
858 }
859
860 splx(s);
861 }
862
863 static void
864 if_detach_queues(struct ifnet *ifp, struct ifqueue *q)
865 {
866 struct mbuf *m, *prev, *next;
867
868 prev = NULL;
869 for (m = q->ifq_head; m != NULL; m = next) {
870 next = m->m_nextpkt;
871 #ifdef DIAGNOSTIC
872 if ((m->m_flags & M_PKTHDR) == 0) {
873 prev = m;
874 continue;
875 }
876 #endif
877 if (m->m_pkthdr.rcvif != ifp) {
878 prev = m;
879 continue;
880 }
881
882 if (prev != NULL)
883 prev->m_nextpkt = m->m_nextpkt;
884 else
885 q->ifq_head = m->m_nextpkt;
886 if (q->ifq_tail == m)
887 q->ifq_tail = prev;
888 q->ifq_len--;
889
890 m->m_nextpkt = NULL;
891 m_freem(m);
892 IF_DROP(q);
893 }
894 }
895
896 /*
897 * Callback for a radix tree walk to delete all references to an
898 * ifnet.
899 */
900 static int
901 if_rt_walktree(struct rtentry *rt, void *v)
902 {
903 struct ifnet *ifp = (struct ifnet *)v;
904 int error;
905
906 if (rt->rt_ifp != ifp)
907 return 0;
908
909 /* Delete the entry. */
910 ++rt->rt_refcnt;
911 error = rtrequest(RTM_DELETE, rt_getkey(rt), rt->rt_gateway,
912 rt_mask(rt), rt->rt_flags, NULL);
913 KASSERT((rt->rt_flags & RTF_UP) == 0);
914 rt->rt_ifp = NULL;
915 RTFREE(rt);
916 if (error != 0)
917 printf("%s: warning: unable to delete rtentry @ %p, "
918 "error = %d\n", ifp->if_xname, rt, error);
919 return ERESTART;
920 }
921
922 /*
923 * Create a clone network interface.
924 */
925 int
926 if_clone_create(const char *name)
927 {
928 struct if_clone *ifc;
929 int unit;
930
931 ifc = if_clone_lookup(name, &unit);
932 if (ifc == NULL)
933 return EINVAL;
934
935 if (ifunit(name) != NULL)
936 return EEXIST;
937
938 return (*ifc->ifc_create)(ifc, unit);
939 }
940
941 /*
942 * Destroy a clone network interface.
943 */
944 int
945 if_clone_destroy(const char *name)
946 {
947 struct if_clone *ifc;
948 struct ifnet *ifp;
949
950 ifc = if_clone_lookup(name, NULL);
951 if (ifc == NULL)
952 return EINVAL;
953
954 ifp = ifunit(name);
955 if (ifp == NULL)
956 return ENXIO;
957
958 if (ifc->ifc_destroy == NULL)
959 return EOPNOTSUPP;
960
961 return (*ifc->ifc_destroy)(ifp);
962 }
963
964 /*
965 * Look up a network interface cloner.
966 */
967 static struct if_clone *
968 if_clone_lookup(const char *name, int *unitp)
969 {
970 struct if_clone *ifc;
971 const char *cp;
972 char *dp, ifname[IFNAMSIZ + 3];
973 int unit;
974
975 strcpy(ifname, "if_");
976 /* separate interface name from unit */
977 for (dp = ifname + 3, cp = name; cp - name < IFNAMSIZ &&
978 *cp && (*cp < '0' || *cp > '9');)
979 *dp++ = *cp++;
980
981 if (cp == name || cp - name == IFNAMSIZ || !*cp)
982 return NULL; /* No name or unit number */
983 *dp++ = '\0';
984
985 again:
986 LIST_FOREACH(ifc, &if_cloners, ifc_list) {
987 if (strcmp(ifname + 3, ifc->ifc_name) == 0)
988 break;
989 }
990
991 if (ifc == NULL) {
992 if (*ifname == '\0' ||
993 module_autoload(ifname, MODULE_CLASS_DRIVER))
994 return NULL;
995 *ifname = '\0';
996 goto again;
997 }
998
999 unit = 0;
1000 while (cp - name < IFNAMSIZ && *cp) {
1001 if (*cp < '0' || *cp > '9' || unit >= INT_MAX / 10) {
1002 /* Bogus unit number. */
1003 return NULL;
1004 }
1005 unit = (unit * 10) + (*cp++ - '0');
1006 }
1007
1008 if (unitp != NULL)
1009 *unitp = unit;
1010 return ifc;
1011 }
1012
1013 /*
1014 * Register a network interface cloner.
1015 */
1016 void
1017 if_clone_attach(struct if_clone *ifc)
1018 {
1019
1020 LIST_INSERT_HEAD(&if_cloners, ifc, ifc_list);
1021 if_cloners_count++;
1022 }
1023
1024 /*
1025 * Unregister a network interface cloner.
1026 */
1027 void
1028 if_clone_detach(struct if_clone *ifc)
1029 {
1030
1031 LIST_REMOVE(ifc, ifc_list);
1032 if_cloners_count--;
1033 }
1034
1035 /*
1036 * Provide list of interface cloners to userspace.
1037 */
1038 static int
1039 if_clone_list(struct if_clonereq *ifcr)
1040 {
1041 char outbuf[IFNAMSIZ], *dst;
1042 struct if_clone *ifc;
1043 int count, error = 0;
1044
1045 ifcr->ifcr_total = if_cloners_count;
1046 if ((dst = ifcr->ifcr_buffer) == NULL) {
1047 /* Just asking how many there are. */
1048 return 0;
1049 }
1050
1051 if (ifcr->ifcr_count < 0)
1052 return EINVAL;
1053
1054 count = (if_cloners_count < ifcr->ifcr_count) ?
1055 if_cloners_count : ifcr->ifcr_count;
1056
1057 for (ifc = LIST_FIRST(&if_cloners); ifc != NULL && count != 0;
1058 ifc = LIST_NEXT(ifc, ifc_list), count--, dst += IFNAMSIZ) {
1059 (void)strncpy(outbuf, ifc->ifc_name, sizeof(outbuf));
1060 if (outbuf[sizeof(outbuf) - 1] != '\0')
1061 return ENAMETOOLONG;
1062 error = copyout(outbuf, dst, sizeof(outbuf));
1063 if (error != 0)
1064 break;
1065 }
1066
1067 return error;
1068 }
1069
1070 void
1071 ifa_insert(struct ifnet *ifp, struct ifaddr *ifa)
1072 {
1073 ifa->ifa_ifp = ifp;
1074 TAILQ_INSERT_TAIL(&ifp->if_addrlist, ifa, ifa_list);
1075 IFAREF(ifa);
1076 }
1077
1078 void
1079 ifa_remove(struct ifnet *ifp, struct ifaddr *ifa)
1080 {
1081 KASSERT(ifa->ifa_ifp == ifp);
1082 TAILQ_REMOVE(&ifp->if_addrlist, ifa, ifa_list);
1083 IFAFREE(ifa);
1084 }
1085
1086 static inline int
1087 equal(const struct sockaddr *sa1, const struct sockaddr *sa2)
1088 {
1089 return sockaddr_cmp(sa1, sa2) == 0;
1090 }
1091
1092 /*
1093 * Locate an interface based on a complete address.
1094 */
1095 /*ARGSUSED*/
1096 struct ifaddr *
1097 ifa_ifwithaddr(const struct sockaddr *addr)
1098 {
1099 struct ifnet *ifp;
1100 struct ifaddr *ifa;
1101
1102 IFNET_FOREACH(ifp) {
1103 if (ifp->if_output == if_nulloutput)
1104 continue;
1105 IFADDR_FOREACH(ifa, ifp) {
1106 if (ifa->ifa_addr->sa_family != addr->sa_family)
1107 continue;
1108 if (equal(addr, ifa->ifa_addr))
1109 return ifa;
1110 if ((ifp->if_flags & IFF_BROADCAST) &&
1111 ifa->ifa_broadaddr &&
1112 /* IP6 doesn't have broadcast */
1113 ifa->ifa_broadaddr->sa_len != 0 &&
1114 equal(ifa->ifa_broadaddr, addr))
1115 return ifa;
1116 }
1117 }
1118 return NULL;
1119 }
1120
1121 /*
1122 * Locate the point to point interface with a given destination address.
1123 */
1124 /*ARGSUSED*/
1125 struct ifaddr *
1126 ifa_ifwithdstaddr(const struct sockaddr *addr)
1127 {
1128 struct ifnet *ifp;
1129 struct ifaddr *ifa;
1130
1131 IFNET_FOREACH(ifp) {
1132 if (ifp->if_output == if_nulloutput)
1133 continue;
1134 if ((ifp->if_flags & IFF_POINTOPOINT) == 0)
1135 continue;
1136 IFADDR_FOREACH(ifa, ifp) {
1137 if (ifa->ifa_addr->sa_family != addr->sa_family ||
1138 ifa->ifa_dstaddr == NULL)
1139 continue;
1140 if (equal(addr, ifa->ifa_dstaddr))
1141 return ifa;
1142 }
1143 }
1144 return NULL;
1145 }
1146
1147 /*
1148 * Find an interface on a specific network. If many, choice
1149 * is most specific found.
1150 */
1151 struct ifaddr *
1152 ifa_ifwithnet(const struct sockaddr *addr)
1153 {
1154 struct ifnet *ifp;
1155 struct ifaddr *ifa;
1156 const struct sockaddr_dl *sdl;
1157 struct ifaddr *ifa_maybe = 0;
1158 u_int af = addr->sa_family;
1159 const char *addr_data = addr->sa_data, *cplim;
1160
1161 if (af == AF_LINK) {
1162 sdl = satocsdl(addr);
1163 if (sdl->sdl_index && sdl->sdl_index < if_indexlim &&
1164 ifindex2ifnet[sdl->sdl_index] &&
1165 ifindex2ifnet[sdl->sdl_index]->if_output != if_nulloutput)
1166 return ifnet_addrs[sdl->sdl_index];
1167 }
1168 #ifdef NETATALK
1169 if (af == AF_APPLETALK) {
1170 const struct sockaddr_at *sat, *sat2;
1171 sat = (const struct sockaddr_at *)addr;
1172 IFNET_FOREACH(ifp) {
1173 if (ifp->if_output == if_nulloutput)
1174 continue;
1175 ifa = at_ifawithnet((const struct sockaddr_at *)addr, ifp);
1176 if (ifa == NULL)
1177 continue;
1178 sat2 = (struct sockaddr_at *)ifa->ifa_addr;
1179 if (sat2->sat_addr.s_net == sat->sat_addr.s_net)
1180 return ifa; /* exact match */
1181 if (ifa_maybe == NULL) {
1182 /* else keep the if with the right range */
1183 ifa_maybe = ifa;
1184 }
1185 }
1186 return ifa_maybe;
1187 }
1188 #endif
1189 IFNET_FOREACH(ifp) {
1190 if (ifp->if_output == if_nulloutput)
1191 continue;
1192 IFADDR_FOREACH(ifa, ifp) {
1193 const char *cp, *cp2, *cp3;
1194
1195 if (ifa->ifa_addr->sa_family != af ||
1196 ifa->ifa_netmask == NULL)
1197 next: continue;
1198 cp = addr_data;
1199 cp2 = ifa->ifa_addr->sa_data;
1200 cp3 = ifa->ifa_netmask->sa_data;
1201 cplim = (const char *)ifa->ifa_netmask +
1202 ifa->ifa_netmask->sa_len;
1203 while (cp3 < cplim) {
1204 if ((*cp++ ^ *cp2++) & *cp3++) {
1205 /* want to continue for() loop */
1206 goto next;
1207 }
1208 }
1209 if (ifa_maybe == NULL ||
1210 rn_refines((void *)ifa->ifa_netmask,
1211 (void *)ifa_maybe->ifa_netmask))
1212 ifa_maybe = ifa;
1213 }
1214 }
1215 return ifa_maybe;
1216 }
1217
1218 /*
1219 * Find the interface of the addresss.
1220 */
1221 struct ifaddr *
1222 ifa_ifwithladdr(const struct sockaddr *addr)
1223 {
1224 struct ifaddr *ia;
1225
1226 if ((ia = ifa_ifwithaddr(addr)) || (ia = ifa_ifwithdstaddr(addr)) ||
1227 (ia = ifa_ifwithnet(addr)))
1228 return ia;
1229 return NULL;
1230 }
1231
1232 /*
1233 * Find an interface using a specific address family
1234 */
1235 struct ifaddr *
1236 ifa_ifwithaf(int af)
1237 {
1238 struct ifnet *ifp;
1239 struct ifaddr *ifa;
1240
1241 IFNET_FOREACH(ifp) {
1242 if (ifp->if_output == if_nulloutput)
1243 continue;
1244 IFADDR_FOREACH(ifa, ifp) {
1245 if (ifa->ifa_addr->sa_family == af)
1246 return ifa;
1247 }
1248 }
1249 return NULL;
1250 }
1251
1252 /*
1253 * Find an interface address specific to an interface best matching
1254 * a given address.
1255 */
1256 struct ifaddr *
1257 ifaof_ifpforaddr(const struct sockaddr *addr, struct ifnet *ifp)
1258 {
1259 struct ifaddr *ifa;
1260 const char *cp, *cp2, *cp3;
1261 const char *cplim;
1262 struct ifaddr *ifa_maybe = 0;
1263 u_int af = addr->sa_family;
1264
1265 if (ifp->if_output == if_nulloutput)
1266 return NULL;
1267
1268 if (af >= AF_MAX)
1269 return NULL;
1270
1271 IFADDR_FOREACH(ifa, ifp) {
1272 if (ifa->ifa_addr->sa_family != af)
1273 continue;
1274 ifa_maybe = ifa;
1275 if (ifa->ifa_netmask == NULL) {
1276 if (equal(addr, ifa->ifa_addr) ||
1277 (ifa->ifa_dstaddr &&
1278 equal(addr, ifa->ifa_dstaddr)))
1279 return ifa;
1280 continue;
1281 }
1282 cp = addr->sa_data;
1283 cp2 = ifa->ifa_addr->sa_data;
1284 cp3 = ifa->ifa_netmask->sa_data;
1285 cplim = ifa->ifa_netmask->sa_len + (char *)ifa->ifa_netmask;
1286 for (; cp3 < cplim; cp3++) {
1287 if ((*cp++ ^ *cp2++) & *cp3)
1288 break;
1289 }
1290 if (cp3 == cplim)
1291 return ifa;
1292 }
1293 return ifa_maybe;
1294 }
1295
1296 /*
1297 * Default action when installing a route with a Link Level gateway.
1298 * Lookup an appropriate real ifa to point to.
1299 * This should be moved to /sys/net/link.c eventually.
1300 */
1301 void
1302 link_rtrequest(int cmd, struct rtentry *rt, const struct rt_addrinfo *info)
1303 {
1304 struct ifaddr *ifa;
1305 const struct sockaddr *dst;
1306 struct ifnet *ifp;
1307
1308 if (cmd != RTM_ADD || (ifa = rt->rt_ifa) == NULL ||
1309 (ifp = ifa->ifa_ifp) == NULL || (dst = rt_getkey(rt)) == NULL)
1310 return;
1311 if ((ifa = ifaof_ifpforaddr(dst, ifp)) != NULL) {
1312 rt_replace_ifa(rt, ifa);
1313 if (ifa->ifa_rtrequest && ifa->ifa_rtrequest != link_rtrequest)
1314 ifa->ifa_rtrequest(cmd, rt, info);
1315 }
1316 }
1317
1318 /*
1319 * Handle a change in the interface link state.
1320 * XXX: We should listen to the routing socket in-kernel rather
1321 * than calling in6_if_link_* functions directly from here.
1322 */
1323 void
1324 if_link_state_change(struct ifnet *ifp, int link_state)
1325 {
1326 int s;
1327 #if defined(DEBUG) || defined(INET6)
1328 int old_link_state;
1329 #endif
1330
1331 s = splnet();
1332 if (ifp->if_link_state == link_state) {
1333 splx(s);
1334 return;
1335 }
1336
1337 #if defined(DEBUG) || defined(INET6)
1338 old_link_state = ifp->if_link_state;
1339 #endif
1340 ifp->if_link_state = link_state;
1341 #ifdef DEBUG
1342 log(LOG_DEBUG, "%s: link state %s (was %s)\n", ifp->if_xname,
1343 link_state == LINK_STATE_UP ? "UP" :
1344 link_state == LINK_STATE_DOWN ? "DOWN" :
1345 "UNKNOWN",
1346 old_link_state == LINK_STATE_UP ? "UP" :
1347 old_link_state == LINK_STATE_DOWN ? "DOWN" :
1348 "UNKNOWN");
1349 #endif
1350
1351 #ifdef INET6
1352 /*
1353 * When going from UNKNOWN to UP, we need to mark existing
1354 * IPv6 addresses as tentative and restart DAD as we may have
1355 * erroneously not found a duplicate.
1356 *
1357 * This needs to happen before rt_ifmsg to avoid a race where
1358 * listeners would have an address and expect it to work right
1359 * away.
1360 */
1361 if (in6_present && link_state == LINK_STATE_UP &&
1362 old_link_state == LINK_STATE_UNKNOWN)
1363 in6_if_link_down(ifp);
1364 #endif
1365
1366 /* Notify that the link state has changed. */
1367 rt_ifmsg(ifp);
1368
1369 #if NCARP > 0
1370 if (ifp->if_carp)
1371 carp_carpdev_state(ifp);
1372 #endif
1373
1374 #ifdef INET6
1375 if (in6_present) {
1376 if (link_state == LINK_STATE_DOWN)
1377 in6_if_link_down(ifp);
1378 else if (link_state == LINK_STATE_UP)
1379 in6_if_link_up(ifp);
1380 }
1381 #endif
1382
1383 splx(s);
1384 }
1385
1386 /*
1387 * Mark an interface down and notify protocols of
1388 * the transition.
1389 * NOTE: must be called at splsoftnet or equivalent.
1390 */
1391 void
1392 if_down(struct ifnet *ifp)
1393 {
1394 struct ifaddr *ifa;
1395
1396 ifp->if_flags &= ~IFF_UP;
1397 nanotime(&ifp->if_lastchange);
1398 IFADDR_FOREACH(ifa, ifp)
1399 pfctlinput(PRC_IFDOWN, ifa->ifa_addr);
1400 IFQ_PURGE(&ifp->if_snd);
1401 #if NCARP > 0
1402 if (ifp->if_carp)
1403 carp_carpdev_state(ifp);
1404 #endif
1405 rt_ifmsg(ifp);
1406 #ifdef INET6
1407 if (in6_present)
1408 in6_if_down(ifp);
1409 #endif
1410 }
1411
1412 /*
1413 * Mark an interface up and notify protocols of
1414 * the transition.
1415 * NOTE: must be called at splsoftnet or equivalent.
1416 */
1417 void
1418 if_up(struct ifnet *ifp)
1419 {
1420 #ifdef notyet
1421 struct ifaddr *ifa;
1422 #endif
1423
1424 ifp->if_flags |= IFF_UP;
1425 nanotime(&ifp->if_lastchange);
1426 #ifdef notyet
1427 /* this has no effect on IP, and will kill all ISO connections XXX */
1428 IFADDR_FOREACH(ifa, ifp)
1429 pfctlinput(PRC_IFUP, ifa->ifa_addr);
1430 #endif
1431 #if NCARP > 0
1432 if (ifp->if_carp)
1433 carp_carpdev_state(ifp);
1434 #endif
1435 rt_ifmsg(ifp);
1436 #ifdef INET6
1437 if (in6_present)
1438 in6_if_up(ifp);
1439 #endif
1440 }
1441
1442 /*
1443 * Handle interface watchdog timer routines. Called
1444 * from softclock, we decrement timers (if set) and
1445 * call the appropriate interface routine on expiration.
1446 */
1447 void
1448 if_slowtimo(void *arg)
1449 {
1450 struct ifnet *ifp;
1451 int s = splnet();
1452
1453 IFNET_FOREACH(ifp) {
1454 if (ifp->if_timer == 0 || --ifp->if_timer)
1455 continue;
1456 if (ifp->if_watchdog != NULL)
1457 (*ifp->if_watchdog)(ifp);
1458 }
1459 splx(s);
1460 callout_reset(&if_slowtimo_ch, hz / IFNET_SLOWHZ, if_slowtimo, NULL);
1461 }
1462
1463 /*
1464 * Set/clear promiscuous mode on interface ifp based on the truth value
1465 * of pswitch. The calls are reference counted so that only the first
1466 * "on" request actually has an effect, as does the final "off" request.
1467 * Results are undefined if the "off" and "on" requests are not matched.
1468 */
1469 int
1470 ifpromisc(struct ifnet *ifp, int pswitch)
1471 {
1472 int pcount, ret;
1473 short nflags;
1474
1475 pcount = ifp->if_pcount;
1476 if (pswitch) {
1477 /*
1478 * Allow the device to be "placed" into promiscuous
1479 * mode even if it is not configured up. It will
1480 * consult IFF_PROMISC when it is brought up.
1481 */
1482 if (ifp->if_pcount++ != 0)
1483 return 0;
1484 nflags = ifp->if_flags | IFF_PROMISC;
1485 } else {
1486 if (--ifp->if_pcount > 0)
1487 return 0;
1488 nflags = ifp->if_flags & ~IFF_PROMISC;
1489 }
1490 ret = if_flags_set(ifp, nflags);
1491 /* Restore interface state if not successful. */
1492 if (ret != 0) {
1493 ifp->if_pcount = pcount;
1494 }
1495 return ret;
1496 }
1497
1498 /*
1499 * Map interface name to
1500 * interface structure pointer.
1501 */
1502 struct ifnet *
1503 ifunit(const char *name)
1504 {
1505 struct ifnet *ifp;
1506 const char *cp = name;
1507 u_int unit = 0;
1508 u_int i;
1509
1510 /*
1511 * If the entire name is a number, treat it as an ifindex.
1512 */
1513 for (i = 0; i < IFNAMSIZ && *cp >= '0' && *cp <= '9'; i++, cp++) {
1514 unit = unit * 10 + (*cp - '0');
1515 }
1516
1517 /*
1518 * If the number took all of the name, then it's a valid ifindex.
1519 */
1520 if (i == IFNAMSIZ || (cp != name && *cp == '\0')) {
1521 if (unit >= if_indexlim)
1522 return NULL;
1523 ifp = ifindex2ifnet[unit];
1524 if (ifp == NULL || ifp->if_output == if_nulloutput)
1525 return NULL;
1526 return ifp;
1527 }
1528
1529 IFNET_FOREACH(ifp) {
1530 if (ifp->if_output == if_nulloutput)
1531 continue;
1532 if (strcmp(ifp->if_xname, name) == 0)
1533 return ifp;
1534 }
1535 return NULL;
1536 }
1537
1538 ifnet_t *
1539 if_byindex(u_int idx)
1540 {
1541
1542 return (idx < if_indexlim) ? ifindex2ifnet[idx] : NULL;
1543 }
1544
1545 /* common */
1546 int
1547 ifioctl_common(struct ifnet *ifp, u_long cmd, void *data)
1548 {
1549 int s;
1550 struct ifreq *ifr;
1551 struct ifcapreq *ifcr;
1552 struct ifdatareq *ifdr;
1553
1554 switch (cmd) {
1555 case SIOCSIFCAP:
1556 ifcr = data;
1557 if ((ifcr->ifcr_capenable & ~ifp->if_capabilities) != 0)
1558 return EINVAL;
1559
1560 if (ifcr->ifcr_capenable == ifp->if_capenable)
1561 return 0;
1562
1563 ifp->if_capenable = ifcr->ifcr_capenable;
1564
1565 /* Pre-compute the checksum flags mask. */
1566 ifp->if_csum_flags_tx = 0;
1567 ifp->if_csum_flags_rx = 0;
1568 if (ifp->if_capenable & IFCAP_CSUM_IPv4_Tx) {
1569 ifp->if_csum_flags_tx |= M_CSUM_IPv4;
1570 }
1571 if (ifp->if_capenable & IFCAP_CSUM_IPv4_Rx) {
1572 ifp->if_csum_flags_rx |= M_CSUM_IPv4;
1573 }
1574
1575 if (ifp->if_capenable & IFCAP_CSUM_TCPv4_Tx) {
1576 ifp->if_csum_flags_tx |= M_CSUM_TCPv4;
1577 }
1578 if (ifp->if_capenable & IFCAP_CSUM_TCPv4_Rx) {
1579 ifp->if_csum_flags_rx |= M_CSUM_TCPv4;
1580 }
1581
1582 if (ifp->if_capenable & IFCAP_CSUM_UDPv4_Tx) {
1583 ifp->if_csum_flags_tx |= M_CSUM_UDPv4;
1584 }
1585 if (ifp->if_capenable & IFCAP_CSUM_UDPv4_Rx) {
1586 ifp->if_csum_flags_rx |= M_CSUM_UDPv4;
1587 }
1588
1589 if (ifp->if_capenable & IFCAP_CSUM_TCPv6_Tx) {
1590 ifp->if_csum_flags_tx |= M_CSUM_TCPv6;
1591 }
1592 if (ifp->if_capenable & IFCAP_CSUM_TCPv6_Rx) {
1593 ifp->if_csum_flags_rx |= M_CSUM_TCPv6;
1594 }
1595
1596 if (ifp->if_capenable & IFCAP_CSUM_UDPv6_Tx) {
1597 ifp->if_csum_flags_tx |= M_CSUM_UDPv6;
1598 }
1599 if (ifp->if_capenable & IFCAP_CSUM_UDPv6_Rx) {
1600 ifp->if_csum_flags_rx |= M_CSUM_UDPv6;
1601 }
1602 if (ifp->if_flags & IFF_UP)
1603 return ENETRESET;
1604 return 0;
1605 case SIOCSIFFLAGS:
1606 ifr = data;
1607 if (ifp->if_flags & IFF_UP && (ifr->ifr_flags & IFF_UP) == 0) {
1608 s = splnet();
1609 if_down(ifp);
1610 splx(s);
1611 }
1612 if (ifr->ifr_flags & IFF_UP && (ifp->if_flags & IFF_UP) == 0) {
1613 s = splnet();
1614 if_up(ifp);
1615 splx(s);
1616 }
1617 ifp->if_flags = (ifp->if_flags & IFF_CANTCHANGE) |
1618 (ifr->ifr_flags &~ IFF_CANTCHANGE);
1619 break;
1620 case SIOCGIFFLAGS:
1621 ifr = data;
1622 ifr->ifr_flags = ifp->if_flags;
1623 break;
1624
1625 case SIOCGIFMETRIC:
1626 ifr = data;
1627 ifr->ifr_metric = ifp->if_metric;
1628 break;
1629
1630 case SIOCGIFMTU:
1631 ifr = data;
1632 ifr->ifr_mtu = ifp->if_mtu;
1633 break;
1634
1635 case SIOCGIFDLT:
1636 ifr = data;
1637 ifr->ifr_dlt = ifp->if_dlt;
1638 break;
1639
1640 case SIOCGIFCAP:
1641 ifcr = data;
1642 ifcr->ifcr_capabilities = ifp->if_capabilities;
1643 ifcr->ifcr_capenable = ifp->if_capenable;
1644 break;
1645
1646 case SIOCSIFMETRIC:
1647 ifr = data;
1648 ifp->if_metric = ifr->ifr_metric;
1649 break;
1650
1651 case SIOCGIFDATA:
1652 ifdr = data;
1653 ifdr->ifdr_data = ifp->if_data;
1654 break;
1655
1656 case SIOCGIFINDEX:
1657 ifr = data;
1658 ifr->ifr_index = ifp->if_index;
1659 break;
1660
1661 case SIOCZIFDATA:
1662 ifdr = data;
1663 ifdr->ifdr_data = ifp->if_data;
1664 /*
1665 * Assumes that the volatile counters that can be
1666 * zero'ed are at the end of if_data.
1667 */
1668 memset(&ifp->if_data.ifi_ipackets, 0, sizeof(ifp->if_data) -
1669 offsetof(struct if_data, ifi_ipackets));
1670 /*
1671 * The memset() clears to the bottm of if_data. In the area,
1672 * if_lastchange is included. Please be careful if new entry
1673 * will be added into if_data or rewite this.
1674 *
1675 * And also, update if_lastchnage.
1676 */
1677 getnanotime(&ifp->if_lastchange);
1678 break;
1679 case SIOCSIFMTU:
1680 ifr = data;
1681 if (ifp->if_mtu == ifr->ifr_mtu)
1682 break;
1683 ifp->if_mtu = ifr->ifr_mtu;
1684 /*
1685 * If the link MTU changed, do network layer specific procedure.
1686 */
1687 #ifdef INET6
1688 if (in6_present)
1689 nd6_setmtu(ifp);
1690 #endif
1691 return ENETRESET;
1692 default:
1693 return ENOTTY;
1694 }
1695 return 0;
1696 }
1697
1698 int
1699 ifaddrpref_ioctl(struct socket *so, u_long cmd, void *data, struct ifnet *ifp,
1700 lwp_t *l)
1701 {
1702 struct if_addrprefreq *ifap = (struct if_addrprefreq *)data;
1703 struct ifaddr *ifa;
1704 const struct sockaddr *any, *sa;
1705 union {
1706 struct sockaddr sa;
1707 struct sockaddr_storage ss;
1708 } u, v;
1709
1710 switch (cmd) {
1711 case SIOCSIFADDRPREF:
1712 if (kauth_authorize_network(l->l_cred, KAUTH_NETWORK_INTERFACE,
1713 KAUTH_REQ_NETWORK_INTERFACE_SETPRIV, ifp, (void *)cmd,
1714 NULL) != 0)
1715 return EPERM;
1716 case SIOCGIFADDRPREF:
1717 break;
1718 default:
1719 return EOPNOTSUPP;
1720 }
1721
1722 /* sanity checks */
1723 if (data == NULL || ifp == NULL) {
1724 panic("invalid argument to %s", __func__);
1725 /*NOTREACHED*/
1726 }
1727
1728 /* address must be specified on ADD and DELETE */
1729 sa = sstocsa(&ifap->ifap_addr);
1730 if (sa->sa_family != sofamily(so))
1731 return EINVAL;
1732 if ((any = sockaddr_any(sa)) == NULL || sa->sa_len != any->sa_len)
1733 return EINVAL;
1734
1735 sockaddr_externalize(&v.sa, sizeof(v.ss), sa);
1736
1737 IFADDR_FOREACH(ifa, ifp) {
1738 if (ifa->ifa_addr->sa_family != sa->sa_family)
1739 continue;
1740 sockaddr_externalize(&u.sa, sizeof(u.ss), ifa->ifa_addr);
1741 if (sockaddr_cmp(&u.sa, &v.sa) == 0)
1742 break;
1743 }
1744 if (ifa == NULL)
1745 return EADDRNOTAVAIL;
1746
1747 switch (cmd) {
1748 case SIOCSIFADDRPREF:
1749 ifa->ifa_preference = ifap->ifap_preference;
1750 return 0;
1751 case SIOCGIFADDRPREF:
1752 /* fill in the if_laddrreq structure */
1753 (void)sockaddr_copy(sstosa(&ifap->ifap_addr),
1754 sizeof(ifap->ifap_addr), ifa->ifa_addr);
1755 ifap->ifap_preference = ifa->ifa_preference;
1756 return 0;
1757 default:
1758 return EOPNOTSUPP;
1759 }
1760 }
1761
1762 static void
1763 ifnet_lock_enter(struct ifnet_lock *il)
1764 {
1765 uint64_t *nenter;
1766
1767 /* Before trying to acquire the mutex, increase the count of threads
1768 * who have entered or who wait to enter the critical section.
1769 * Avoid one costly locked memory transaction by keeping a count for
1770 * each CPU.
1771 */
1772 nenter = percpu_getref(il->il_nenter);
1773 (*nenter)++;
1774 percpu_putref(il->il_nenter);
1775 mutex_enter(&il->il_lock);
1776 }
1777
1778 static void
1779 ifnet_lock_exit(struct ifnet_lock *il)
1780 {
1781 /* Increase the count of threads who have exited the critical
1782 * section. Increase while we still hold the lock.
1783 */
1784 il->il_nexit++;
1785 mutex_exit(&il->il_lock);
1786 }
1787
1788 /*
1789 * Interface ioctls.
1790 */
1791 int
1792 ifioctl(struct socket *so, u_long cmd, void *data, struct lwp *l)
1793 {
1794 struct ifnet *ifp;
1795 struct ifreq *ifr;
1796 int error = 0;
1797 #if defined(COMPAT_OSOCK) || defined(COMPAT_OIFREQ)
1798 u_long ocmd = cmd;
1799 #endif
1800 short oif_flags;
1801 #ifdef COMPAT_OIFREQ
1802 struct ifreq ifrb;
1803 struct oifreq *oifr = NULL;
1804 #endif
1805
1806 switch (cmd) {
1807 #ifdef COMPAT_OIFREQ
1808 case OSIOCGIFCONF:
1809 case OOSIOCGIFCONF:
1810 return compat_ifconf(cmd, data);
1811 #endif
1812 #ifdef COMPAT_OIFDATA
1813 case OSIOCGIFDATA:
1814 case OSIOCZIFDATA:
1815 return compat_ifdatareq(l, cmd, data);
1816 #endif
1817 case SIOCGIFCONF:
1818 return ifconf(cmd, data);
1819 case SIOCINITIFADDR:
1820 return EPERM;
1821 }
1822
1823 #ifdef COMPAT_OIFREQ
1824 cmd = compat_cvtcmd(cmd);
1825 if (cmd != ocmd) {
1826 oifr = data;
1827 data = ifr = &ifrb;
1828 ifreqo2n(oifr, ifr);
1829 } else
1830 #endif
1831 ifr = data;
1832
1833 ifp = ifunit(ifr->ifr_name);
1834
1835 switch (cmd) {
1836 case SIOCIFCREATE:
1837 case SIOCIFDESTROY:
1838 if (l != NULL) {
1839 error = kauth_authorize_network(l->l_cred,
1840 KAUTH_NETWORK_INTERFACE,
1841 KAUTH_REQ_NETWORK_INTERFACE_SETPRIV, ifp,
1842 (void *)cmd, NULL);
1843 if (error != 0)
1844 return error;
1845 }
1846 return (cmd == SIOCIFCREATE) ?
1847 if_clone_create(ifr->ifr_name) :
1848 if_clone_destroy(ifr->ifr_name);
1849
1850 case SIOCIFGCLONERS:
1851 return if_clone_list((struct if_clonereq *)data);
1852 }
1853
1854 if (ifp == NULL)
1855 return ENXIO;
1856
1857 switch (cmd) {
1858 case SIOCALIFADDR:
1859 case SIOCDLIFADDR:
1860 case SIOCSIFADDRPREF:
1861 case SIOCSIFFLAGS:
1862 case SIOCSIFCAP:
1863 case SIOCSIFMETRIC:
1864 case SIOCZIFDATA:
1865 case SIOCSIFMTU:
1866 case SIOCSIFPHYADDR:
1867 case SIOCDIFPHYADDR:
1868 #ifdef INET6
1869 case SIOCSIFPHYADDR_IN6:
1870 #endif
1871 case SIOCSLIFPHYADDR:
1872 case SIOCADDMULTI:
1873 case SIOCDELMULTI:
1874 case SIOCSIFMEDIA:
1875 case SIOCSDRVSPEC:
1876 case SIOCG80211:
1877 case SIOCS80211:
1878 case SIOCS80211NWID:
1879 case SIOCS80211NWKEY:
1880 case SIOCS80211POWER:
1881 case SIOCS80211BSSID:
1882 case SIOCS80211CHANNEL:
1883 case SIOCSLINKSTR:
1884 if (l != NULL) {
1885 error = kauth_authorize_network(l->l_cred,
1886 KAUTH_NETWORK_INTERFACE,
1887 KAUTH_REQ_NETWORK_INTERFACE_SETPRIV, ifp,
1888 (void *)cmd, NULL);
1889 if (error != 0)
1890 return error;
1891 }
1892 }
1893
1894 oif_flags = ifp->if_flags;
1895
1896 ifnet_lock_enter(ifp->if_ioctl_lock);
1897 error = (*ifp->if_ioctl)(ifp, cmd, data);
1898 if (error != ENOTTY)
1899 ;
1900 else if (so->so_proto == NULL)
1901 error = EOPNOTSUPP;
1902 else {
1903 #ifdef COMPAT_OSOCK
1904 error = compat_ifioctl(so, ocmd, cmd, data, l);
1905 #else
1906 error = (*so->so_proto->pr_usrreq)(so, PRU_CONTROL,
1907 (struct mbuf *)cmd, (struct mbuf *)data,
1908 (struct mbuf *)ifp, l);
1909 #endif
1910 }
1911
1912 if (((oif_flags ^ ifp->if_flags) & IFF_UP) != 0) {
1913 #ifdef INET6
1914 if (in6_present && (ifp->if_flags & IFF_UP) != 0) {
1915 int s = splnet();
1916 in6_if_up(ifp);
1917 splx(s);
1918 }
1919 #endif
1920 }
1921 #ifdef COMPAT_OIFREQ
1922 if (cmd != ocmd)
1923 ifreqn2o(oifr, ifr);
1924 #endif
1925
1926 ifnet_lock_exit(ifp->if_ioctl_lock);
1927 return error;
1928 }
1929
1930 /* This callback adds to the sum in `arg' the number of
1931 * threads on `ci' who have entered or who wait to enter the
1932 * critical section.
1933 */
1934 static void
1935 ifnet_lock_sum(void *p, void *arg, struct cpu_info *ci)
1936 {
1937 uint64_t *sum = arg, *nenter = p;
1938
1939 *sum += *nenter;
1940 }
1941
1942 /* Return the number of threads who have entered or who wait
1943 * to enter the critical section on all CPUs.
1944 */
1945 static uint64_t
1946 ifnet_lock_entrances(struct ifnet_lock *il)
1947 {
1948 uint64_t sum = 0;
1949
1950 percpu_foreach(il->il_nenter, ifnet_lock_sum, &sum);
1951
1952 return sum;
1953 }
1954
1955 static int
1956 ifioctl_attach(struct ifnet *ifp)
1957 {
1958 struct ifnet_lock *il;
1959
1960 /* If the driver has not supplied its own if_ioctl, then
1961 * supply the default.
1962 */
1963 if (ifp->if_ioctl == NULL)
1964 ifp->if_ioctl = ifioctl_common;
1965
1966 /* Create an ifnet_lock for synchronizing ifioctls. */
1967 if ((il = kmem_zalloc(sizeof(*il), KM_SLEEP)) == NULL)
1968 return ENOMEM;
1969
1970 il->il_nenter = percpu_alloc(sizeof(uint64_t));
1971 if (il->il_nenter == NULL) {
1972 kmem_free(il, sizeof(*il));
1973 return ENOMEM;
1974 }
1975
1976 mutex_init(&il->il_lock, MUTEX_DEFAULT, IPL_NONE);
1977 cv_init(&il->il_emptied, ifp->if_xname);
1978
1979 ifp->if_ioctl_lock = il;
1980
1981 return 0;
1982 }
1983
1984 /*
1985 * This must not be called until after `ifp' has been withdrawn from the
1986 * ifnet tables so that ifioctl() cannot get a handle on it by calling
1987 * ifunit().
1988 */
1989 static void
1990 ifioctl_detach(struct ifnet *ifp)
1991 {
1992 struct ifnet_lock *il;
1993
1994 il = ifp->if_ioctl_lock;
1995 mutex_enter(&il->il_lock);
1996 /* Install if_nullioctl to make sure that any thread that
1997 * subsequently enters the critical section will quit it
1998 * immediately and signal the condition variable that we
1999 * wait on, below.
2000 */
2001 ifp->if_ioctl = if_nullioctl;
2002 /* Sleep while threads are still in the critical section or
2003 * wait to enter it.
2004 */
2005 while (ifnet_lock_entrances(il) != il->il_nexit)
2006 cv_wait(&il->il_emptied, &il->il_lock);
2007 /* At this point, we are the only thread still in the critical
2008 * section, and no new thread can get a handle on the ifioctl
2009 * lock, so it is safe to free its memory.
2010 */
2011 mutex_exit(&il->il_lock);
2012 ifp->if_ioctl_lock = NULL;
2013 percpu_free(il->il_nenter, sizeof(uint64_t));
2014 il->il_nenter = NULL;
2015 cv_destroy(&il->il_emptied);
2016 mutex_destroy(&il->il_lock);
2017 kmem_free(il, sizeof(*il));
2018 }
2019
2020 /*
2021 * Return interface configuration
2022 * of system. List may be used
2023 * in later ioctl's (above) to get
2024 * other information.
2025 *
2026 * Each record is a struct ifreq. Before the addition of
2027 * sockaddr_storage, the API rule was that sockaddr flavors that did
2028 * not fit would extend beyond the struct ifreq, with the next struct
2029 * ifreq starting sa_len beyond the struct sockaddr. Because the
2030 * union in struct ifreq includes struct sockaddr_storage, every kind
2031 * of sockaddr must fit. Thus, there are no longer any overlength
2032 * records.
2033 *
2034 * Records are added to the user buffer if they fit, and ifc_len is
2035 * adjusted to the length that was written. Thus, the user is only
2036 * assured of getting the complete list if ifc_len on return is at
2037 * least sizeof(struct ifreq) less than it was on entry.
2038 *
2039 * If the user buffer pointer is NULL, this routine copies no data and
2040 * returns the amount of space that would be needed.
2041 *
2042 * Invariants:
2043 * ifrp points to the next part of the user's buffer to be used. If
2044 * ifrp != NULL, space holds the number of bytes remaining that we may
2045 * write at ifrp. Otherwise, space holds the number of bytes that
2046 * would have been written had there been adequate space.
2047 */
2048 /*ARGSUSED*/
2049 int
2050 ifconf(u_long cmd, void *data)
2051 {
2052 struct ifconf *ifc = (struct ifconf *)data;
2053 struct ifnet *ifp;
2054 struct ifaddr *ifa;
2055 struct ifreq ifr, *ifrp;
2056 int space, error = 0;
2057 const int sz = (int)sizeof(struct ifreq);
2058
2059 if ((ifrp = ifc->ifc_req) == NULL)
2060 space = 0;
2061 else
2062 space = ifc->ifc_len;
2063 IFNET_FOREACH(ifp) {
2064 (void)strncpy(ifr.ifr_name, ifp->if_xname,
2065 sizeof(ifr.ifr_name));
2066 if (ifr.ifr_name[sizeof(ifr.ifr_name) - 1] != '\0')
2067 return ENAMETOOLONG;
2068 if (IFADDR_EMPTY(ifp)) {
2069 /* Interface with no addresses - send zero sockaddr. */
2070 memset(&ifr.ifr_addr, 0, sizeof(ifr.ifr_addr));
2071 if (ifrp == NULL) {
2072 space += sz;
2073 continue;
2074 }
2075 if (space >= sz) {
2076 error = copyout(&ifr, ifrp, sz);
2077 if (error != 0)
2078 return error;
2079 ifrp++;
2080 space -= sz;
2081 }
2082 }
2083
2084 IFADDR_FOREACH(ifa, ifp) {
2085 struct sockaddr *sa = ifa->ifa_addr;
2086 /* all sockaddrs must fit in sockaddr_storage */
2087 KASSERT(sa->sa_len <= sizeof(ifr.ifr_ifru));
2088
2089 if (ifrp == NULL) {
2090 space += sz;
2091 continue;
2092 }
2093 memcpy(&ifr.ifr_space, sa, sa->sa_len);
2094 if (space >= sz) {
2095 error = copyout(&ifr, ifrp, sz);
2096 if (error != 0)
2097 return (error);
2098 ifrp++; space -= sz;
2099 }
2100 }
2101 }
2102 if (ifrp != NULL) {
2103 KASSERT(0 <= space && space <= ifc->ifc_len);
2104 ifc->ifc_len -= space;
2105 } else {
2106 KASSERT(space >= 0);
2107 ifc->ifc_len = space;
2108 }
2109 return (0);
2110 }
2111
2112 int
2113 ifreq_setaddr(u_long cmd, struct ifreq *ifr, const struct sockaddr *sa)
2114 {
2115 uint8_t len;
2116 #ifdef COMPAT_OIFREQ
2117 struct ifreq ifrb;
2118 struct oifreq *oifr = NULL;
2119 u_long ocmd = cmd;
2120 cmd = compat_cvtcmd(cmd);
2121 if (cmd != ocmd) {
2122 oifr = (struct oifreq *)(void *)ifr;
2123 ifr = &ifrb;
2124 ifreqo2n(oifr, ifr);
2125 len = sizeof(oifr->ifr_addr);
2126 } else
2127 #endif
2128 len = sizeof(ifr->ifr_ifru.ifru_space);
2129
2130 if (len < sa->sa_len)
2131 return EFBIG;
2132
2133 memset(&ifr->ifr_addr, 0, len);
2134 sockaddr_copy(&ifr->ifr_addr, len, sa);
2135
2136 #ifdef COMPAT_OIFREQ
2137 if (cmd != ocmd)
2138 ifreqn2o(oifr, ifr);
2139 #endif
2140 return 0;
2141 }
2142
2143 /*
2144 * Queue message on interface, and start output if interface
2145 * not yet active.
2146 */
2147 int
2148 ifq_enqueue(struct ifnet *ifp, struct mbuf *m
2149 ALTQ_COMMA ALTQ_DECL(struct altq_pktattr *pktattr))
2150 {
2151 int len = m->m_pkthdr.len;
2152 int mflags = m->m_flags;
2153 int s = splnet();
2154 int error;
2155
2156 IFQ_ENQUEUE(&ifp->if_snd, m, pktattr, error);
2157 if (error != 0)
2158 goto out;
2159 ifp->if_obytes += len;
2160 if (mflags & M_MCAST)
2161 ifp->if_omcasts++;
2162 if ((ifp->if_flags & IFF_OACTIVE) == 0)
2163 (*ifp->if_start)(ifp);
2164 out:
2165 splx(s);
2166 return error;
2167 }
2168
2169 /*
2170 * Queue message on interface, possibly using a second fast queue
2171 */
2172 int
2173 ifq_enqueue2(struct ifnet *ifp, struct ifqueue *ifq, struct mbuf *m
2174 ALTQ_COMMA ALTQ_DECL(struct altq_pktattr *pktattr))
2175 {
2176 int error = 0;
2177
2178 if (ifq != NULL
2179 #ifdef ALTQ
2180 && ALTQ_IS_ENABLED(&ifp->if_snd) == 0
2181 #endif
2182 ) {
2183 if (IF_QFULL(ifq)) {
2184 IF_DROP(&ifp->if_snd);
2185 m_freem(m);
2186 if (error == 0)
2187 error = ENOBUFS;
2188 } else
2189 IF_ENQUEUE(ifq, m);
2190 } else
2191 IFQ_ENQUEUE(&ifp->if_snd, m, pktattr, error);
2192 if (error != 0) {
2193 ++ifp->if_oerrors;
2194 return error;
2195 }
2196 return 0;
2197 }
2198
2199 int
2200 if_addr_init(ifnet_t *ifp, struct ifaddr *ifa, const bool src)
2201 {
2202 int rc;
2203
2204 if (ifp->if_initaddr != NULL)
2205 rc = (*ifp->if_initaddr)(ifp, ifa, src);
2206 else if (src ||
2207 (rc = (*ifp->if_ioctl)(ifp, SIOCSIFDSTADDR, ifa)) == ENOTTY)
2208 rc = (*ifp->if_ioctl)(ifp, SIOCINITIFADDR, ifa);
2209
2210 return rc;
2211 }
2212
2213 int
2214 if_flags_set(ifnet_t *ifp, const short flags)
2215 {
2216 int rc;
2217
2218 if (ifp->if_setflags != NULL)
2219 rc = (*ifp->if_setflags)(ifp, flags);
2220 else {
2221 short cantflags, chgdflags;
2222 struct ifreq ifr;
2223
2224 chgdflags = ifp->if_flags ^ flags;
2225 cantflags = chgdflags & IFF_CANTCHANGE;
2226
2227 if (cantflags != 0)
2228 ifp->if_flags ^= cantflags;
2229
2230 /* Traditionally, we do not call if_ioctl after
2231 * setting/clearing only IFF_PROMISC if the interface
2232 * isn't IFF_UP. Uphold that tradition.
2233 */
2234 if (chgdflags == IFF_PROMISC && (ifp->if_flags & IFF_UP) == 0)
2235 return 0;
2236
2237 memset(&ifr, 0, sizeof(ifr));
2238
2239 ifr.ifr_flags = flags & ~IFF_CANTCHANGE;
2240 rc = (*ifp->if_ioctl)(ifp, SIOCSIFFLAGS, &ifr);
2241
2242 if (rc != 0 && cantflags != 0)
2243 ifp->if_flags ^= cantflags;
2244 }
2245
2246 return rc;
2247 }
2248
2249 int
2250 if_mcast_op(ifnet_t *ifp, const unsigned long cmd, const struct sockaddr *sa)
2251 {
2252 int rc;
2253 struct ifreq ifr;
2254
2255 if (ifp->if_mcastop != NULL)
2256 rc = (*ifp->if_mcastop)(ifp, cmd, sa);
2257 else {
2258 ifreq_setaddr(cmd, &ifr, sa);
2259 rc = (*ifp->if_ioctl)(ifp, cmd, &ifr);
2260 }
2261
2262 return rc;
2263 }
2264
2265 static void
2266 sysctl_sndq_setup(struct sysctllog **clog, const char *ifname,
2267 struct ifaltq *ifq)
2268 {
2269 const struct sysctlnode *cnode, *rnode;
2270
2271 if (sysctl_createv(clog, 0, NULL, &rnode,
2272 CTLFLAG_PERMANENT,
2273 CTLTYPE_NODE, "net", NULL,
2274 NULL, 0, NULL, 0,
2275 CTL_NET, CTL_EOL) != 0)
2276 goto bad;
2277
2278 if (sysctl_createv(clog, 0, &rnode, &rnode,
2279 CTLFLAG_PERMANENT,
2280 CTLTYPE_NODE, "interfaces",
2281 SYSCTL_DESCR("Per-interface controls"),
2282 NULL, 0, NULL, 0,
2283 CTL_CREATE, CTL_EOL) != 0)
2284 goto bad;
2285
2286 if (sysctl_createv(clog, 0, &rnode, &rnode,
2287 CTLFLAG_PERMANENT,
2288 CTLTYPE_NODE, ifname,
2289 SYSCTL_DESCR("Interface controls"),
2290 NULL, 0, NULL, 0,
2291 CTL_CREATE, CTL_EOL) != 0)
2292 goto bad;
2293
2294 if (sysctl_createv(clog, 0, &rnode, &rnode,
2295 CTLFLAG_PERMANENT,
2296 CTLTYPE_NODE, "sndq",
2297 SYSCTL_DESCR("Interface output queue controls"),
2298 NULL, 0, NULL, 0,
2299 CTL_CREATE, CTL_EOL) != 0)
2300 goto bad;
2301
2302 if (sysctl_createv(clog, 0, &rnode, &cnode,
2303 CTLFLAG_PERMANENT,
2304 CTLTYPE_INT, "len",
2305 SYSCTL_DESCR("Current output queue length"),
2306 NULL, 0, &ifq->ifq_len, 0,
2307 CTL_CREATE, CTL_EOL) != 0)
2308 goto bad;
2309
2310 if (sysctl_createv(clog, 0, &rnode, &cnode,
2311 CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2312 CTLTYPE_INT, "maxlen",
2313 SYSCTL_DESCR("Maximum allowed output queue length"),
2314 NULL, 0, &ifq->ifq_maxlen, 0,
2315 CTL_CREATE, CTL_EOL) != 0)
2316 goto bad;
2317
2318 if (sysctl_createv(clog, 0, &rnode, &cnode,
2319 CTLFLAG_PERMANENT,
2320 CTLTYPE_INT, "drops",
2321 SYSCTL_DESCR("Packets dropped due to full output queue"),
2322 NULL, 0, &ifq->ifq_drops, 0,
2323 CTL_CREATE, CTL_EOL) != 0)
2324 goto bad;
2325
2326 return;
2327 bad:
2328 printf("%s: could not attach sysctl nodes\n", ifname);
2329 return;
2330 }
2331
2332 #if defined(INET) || defined(INET6)
2333 static void
2334 sysctl_net_ifq_setup(struct sysctllog **clog,
2335 int pf, const char *pfname,
2336 int ipn, const char *ipname,
2337 int qid, struct ifqueue *ifq)
2338 {
2339
2340 sysctl_createv(clog, 0, NULL, NULL,
2341 CTLFLAG_PERMANENT,
2342 CTLTYPE_NODE, "net", NULL,
2343 NULL, 0, NULL, 0,
2344 CTL_NET, CTL_EOL);
2345 sysctl_createv(clog, 0, NULL, NULL,
2346 CTLFLAG_PERMANENT,
2347 CTLTYPE_NODE, pfname, NULL,
2348 NULL, 0, NULL, 0,
2349 CTL_NET, pf, CTL_EOL);
2350 sysctl_createv(clog, 0, NULL, NULL,
2351 CTLFLAG_PERMANENT,
2352 CTLTYPE_NODE, ipname, NULL,
2353 NULL, 0, NULL, 0,
2354 CTL_NET, pf, ipn, CTL_EOL);
2355 sysctl_createv(clog, 0, NULL, NULL,
2356 CTLFLAG_PERMANENT,
2357 CTLTYPE_NODE, "ifq",
2358 SYSCTL_DESCR("Protocol input queue controls"),
2359 NULL, 0, NULL, 0,
2360 CTL_NET, pf, ipn, qid, CTL_EOL);
2361
2362 sysctl_createv(clog, 0, NULL, NULL,
2363 CTLFLAG_PERMANENT,
2364 CTLTYPE_INT, "len",
2365 SYSCTL_DESCR("Current input queue length"),
2366 NULL, 0, &ifq->ifq_len, 0,
2367 CTL_NET, pf, ipn, qid, IFQCTL_LEN, CTL_EOL);
2368 sysctl_createv(clog, 0, NULL, NULL,
2369 CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
2370 CTLTYPE_INT, "maxlen",
2371 SYSCTL_DESCR("Maximum allowed input queue length"),
2372 NULL, 0, &ifq->ifq_maxlen, 0,
2373 CTL_NET, pf, ipn, qid, IFQCTL_MAXLEN, CTL_EOL);
2374 #ifdef notyet
2375 sysctl_createv(clog, 0, NULL, NULL,
2376 CTLFLAG_PERMANENT,
2377 CTLTYPE_INT, "peak",
2378 SYSCTL_DESCR("Highest input queue length"),
2379 NULL, 0, &ifq->ifq_peak, 0,
2380 CTL_NET, pf, ipn, qid, IFQCTL_PEAK, CTL_EOL);
2381 #endif
2382 sysctl_createv(clog, 0, NULL, NULL,
2383 CTLFLAG_PERMANENT,
2384 CTLTYPE_INT, "drops",
2385 SYSCTL_DESCR("Packets dropped due to full input queue"),
2386 NULL, 0, &ifq->ifq_drops, 0,
2387 CTL_NET, pf, ipn, qid, IFQCTL_DROPS, CTL_EOL);
2388 }
2389 #endif /* INET || INET6 */
2390