if.c revision 1.161 1 /* $NetBSD: if.c,v 1.161 2005/09/24 15:52:03 christos Exp $ */
2
3 /*-
4 * Copyright (c) 1999, 2000, 2001 The NetBSD Foundation, Inc.
5 * All rights reserved.
6 *
7 * This code is derived from software contributed to The NetBSD Foundation
8 * by William Studenmund and Jason R. Thorpe.
9 *
10 * Redistribution and use in source and binary forms, with or without
11 * modification, are permitted provided that the following conditions
12 * are met:
13 * 1. Redistributions of source code must retain the above copyright
14 * notice, this list of conditions and the following disclaimer.
15 * 2. Redistributions in binary form must reproduce the above copyright
16 * notice, this list of conditions and the following disclaimer in the
17 * documentation and/or other materials provided with the distribution.
18 * 3. All advertising materials mentioning features or use of this software
19 * must display the following acknowledgement:
20 * This product includes software developed by the NetBSD
21 * Foundation, Inc. and its contributors.
22 * 4. Neither the name of The NetBSD Foundation nor the names of its
23 * contributors may be used to endorse or promote products derived
24 * from this software without specific prior written permission.
25 *
26 * THIS SOFTWARE IS PROVIDED BY THE NETBSD FOUNDATION, INC. AND CONTRIBUTORS
27 * ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED
28 * TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
29 * PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE FOUNDATION OR CONTRIBUTORS
30 * BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR
31 * CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF
32 * SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS
33 * INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN
34 * CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE)
35 * ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
36 * POSSIBILITY OF SUCH DAMAGE.
37 */
38
39 /*
40 * Copyright (C) 1995, 1996, 1997, and 1998 WIDE Project.
41 * All rights reserved.
42 *
43 * Redistribution and use in source and binary forms, with or without
44 * modification, are permitted provided that the following conditions
45 * are met:
46 * 1. Redistributions of source code must retain the above copyright
47 * notice, this list of conditions and the following disclaimer.
48 * 2. Redistributions in binary form must reproduce the above copyright
49 * notice, this list of conditions and the following disclaimer in the
50 * documentation and/or other materials provided with the distribution.
51 * 3. Neither the name of the project nor the names of its contributors
52 * may be used to endorse or promote products derived from this software
53 * without specific prior written permission.
54 *
55 * THIS SOFTWARE IS PROVIDED BY THE PROJECT AND CONTRIBUTORS ``AS IS'' AND
56 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
57 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
58 * ARE DISCLAIMED. IN NO EVENT SHALL THE PROJECT OR CONTRIBUTORS BE LIABLE
59 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
60 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
61 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
62 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
63 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
64 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
65 * SUCH DAMAGE.
66 */
67
68 /*
69 * Copyright (c) 1980, 1986, 1993
70 * The Regents of the University of California. All rights reserved.
71 *
72 * Redistribution and use in source and binary forms, with or without
73 * modification, are permitted provided that the following conditions
74 * are met:
75 * 1. Redistributions of source code must retain the above copyright
76 * notice, this list of conditions and the following disclaimer.
77 * 2. Redistributions in binary form must reproduce the above copyright
78 * notice, this list of conditions and the following disclaimer in the
79 * documentation and/or other materials provided with the distribution.
80 * 3. Neither the name of the University nor the names of its contributors
81 * may be used to endorse or promote products derived from this software
82 * without specific prior written permission.
83 *
84 * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
85 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
86 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
87 * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
88 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
89 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
90 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
91 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
92 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
93 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
94 * SUCH DAMAGE.
95 *
96 * @(#)if.c 8.5 (Berkeley) 1/9/95
97 */
98
99 #include <sys/cdefs.h>
100 __KERNEL_RCSID(0, "$NetBSD: if.c,v 1.161 2005/09/24 15:52:03 christos Exp $");
101
102 #include "opt_inet.h"
103
104 #include "opt_compat_linux.h"
105 #include "opt_compat_svr4.h"
106 #include "opt_compat_ultrix.h"
107 #include "opt_compat_43.h"
108 #include "opt_atalk.h"
109 #include "opt_ccitt.h"
110 #include "opt_natm.h"
111 #include "opt_pfil_hooks.h"
112
113 #include <sys/param.h>
114 #include <sys/mbuf.h>
115 #include <sys/systm.h>
116 #include <sys/callout.h>
117 #include <sys/proc.h>
118 #include <sys/socket.h>
119 #include <sys/socketvar.h>
120 #include <sys/domain.h>
121 #include <sys/protosw.h>
122 #include <sys/kernel.h>
123 #include <sys/ioctl.h>
124 #include <sys/sysctl.h>
125 #include <sys/syslog.h>
126
127 #include <net/if.h>
128 #include <net/if_dl.h>
129 #include <net/if_ether.h>
130 #include <net/if_media.h>
131 #include <net80211/ieee80211.h>
132 #include <net80211/ieee80211_ioctl.h>
133 #include <net/if_types.h>
134 #include <net/radix.h>
135 #include <net/route.h>
136 #include <net/netisr.h>
137 #ifdef NETATALK
138 #include <netatalk/at_extern.h>
139 #include <netatalk/at.h>
140 #endif
141 #include <net/pfil.h>
142
143 #ifdef INET6
144 #include <netinet/in.h>
145 #include <netinet6/in6_var.h>
146 #include <netinet6/nd6.h>
147 #endif
148
149 #if defined(COMPAT_43) || defined(COMPAT_LINUX) || defined(COMPAT_SVR4) || defined(COMPAT_ULTRIX) || defined(LKM)
150 #define COMPAT_OSOCK
151 #include <compat/sys/socket.h>
152 #endif
153
154 MALLOC_DEFINE(M_IFADDR, "ifaddr", "interface address");
155 MALLOC_DEFINE(M_IFMADDR, "ether_multi", "link-level multicast address");
156
157 int ifqmaxlen = IFQ_MAXLEN;
158 struct callout if_slowtimo_ch;
159
160 int netisr; /* scheduling bits for network */
161
162 int if_rt_walktree __P((struct radix_node *, void *));
163
164 struct if_clone *if_clone_lookup __P((const char *, int *));
165 int if_clone_list __P((struct if_clonereq *));
166
167 LIST_HEAD(, if_clone) if_cloners = LIST_HEAD_INITIALIZER(if_cloners);
168 int if_cloners_count;
169
170 #ifdef PFIL_HOOKS
171 struct pfil_head if_pfil; /* packet filtering hook for interfaces */
172 #endif
173
174 #if defined(INET) || defined(INET6) || defined(NETATALK) || defined(NS) || \
175 defined(ISO) || defined(CCITT) || defined(NATM)
176 static void if_detach_queues __P((struct ifnet *, struct ifqueue *));
177 #endif
178
179 /*
180 * Network interface utility routines.
181 *
182 * Routines with ifa_ifwith* names take sockaddr *'s as
183 * parameters.
184 */
185 void
186 ifinit()
187 {
188
189 callout_init(&if_slowtimo_ch);
190 if_slowtimo(NULL);
191 #ifdef PFIL_HOOKS
192 if_pfil.ph_type = PFIL_TYPE_IFNET;
193 if_pfil.ph_ifnet = NULL;
194 if (pfil_head_register(&if_pfil) != 0)
195 printf("WARNING: unable to register pfil hook\n");
196 #endif
197 }
198
199 /*
200 * Null routines used while an interface is going away. These routines
201 * just return an error.
202 */
203
204 int
205 if_nulloutput(ifp, m, so, rt)
206 struct ifnet *ifp;
207 struct mbuf *m;
208 struct sockaddr *so;
209 struct rtentry *rt;
210 {
211
212 return (ENXIO);
213 }
214
215 void
216 if_nullinput(ifp, m)
217 struct ifnet *ifp;
218 struct mbuf *m;
219 {
220
221 /* Nothing. */
222 }
223
224 void
225 if_nullstart(ifp)
226 struct ifnet *ifp;
227 {
228
229 /* Nothing. */
230 }
231
232 int
233 if_nullioctl(ifp, cmd, data)
234 struct ifnet *ifp;
235 u_long cmd;
236 caddr_t data;
237 {
238
239 return (ENXIO);
240 }
241
242 int
243 if_nullinit(ifp)
244 struct ifnet *ifp;
245 {
246
247 return (ENXIO);
248 }
249
250 void
251 if_nullstop(ifp, disable)
252 struct ifnet *ifp;
253 int disable;
254 {
255
256 /* Nothing. */
257 }
258
259 void
260 if_nullwatchdog(ifp)
261 struct ifnet *ifp;
262 {
263
264 /* Nothing. */
265 }
266
267 void
268 if_nulldrain(ifp)
269 struct ifnet *ifp;
270 {
271
272 /* Nothing. */
273 }
274
275 static u_int if_index = 1;
276 struct ifnet_head ifnet;
277 size_t if_indexlim = 0;
278 struct ifaddr **ifnet_addrs = NULL;
279 struct ifnet **ifindex2ifnet = NULL;
280 struct ifnet *lo0ifp;
281
282 /*
283 * Allocate the link level name for the specified interface. This
284 * is an attachment helper. It must be called after ifp->if_addrlen
285 * is initialized, which may not be the case when if_attach() is
286 * called.
287 */
288 void
289 if_alloc_sadl(struct ifnet *ifp)
290 {
291 unsigned socksize, ifasize;
292 int namelen, masklen;
293 struct sockaddr_dl *sdl;
294 struct ifaddr *ifa;
295
296 /*
297 * If the interface already has a link name, release it
298 * now. This is useful for interfaces that can change
299 * link types, and thus switch link names often.
300 */
301 if (ifp->if_sadl != NULL)
302 if_free_sadl(ifp);
303
304 namelen = strlen(ifp->if_xname);
305 masklen = offsetof(struct sockaddr_dl, sdl_data[0]) + namelen;
306 socksize = masklen + ifp->if_addrlen;
307 #define ROUNDUP(a) (1 + (((a) - 1) | (sizeof(long) - 1)))
308 if (socksize < sizeof(*sdl))
309 socksize = sizeof(*sdl);
310 socksize = ROUNDUP(socksize);
311 ifasize = sizeof(*ifa) + 2 * socksize;
312 ifa = (struct ifaddr *)malloc(ifasize, M_IFADDR, M_WAITOK);
313 memset((caddr_t)ifa, 0, ifasize);
314 sdl = (struct sockaddr_dl *)(ifa + 1);
315 sdl->sdl_len = socksize;
316 sdl->sdl_family = AF_LINK;
317 bcopy(ifp->if_xname, sdl->sdl_data, namelen);
318 sdl->sdl_nlen = namelen;
319 sdl->sdl_alen = ifp->if_addrlen;
320 sdl->sdl_index = ifp->if_index;
321 sdl->sdl_type = ifp->if_type;
322 ifnet_addrs[ifp->if_index] = ifa;
323 IFAREF(ifa);
324 ifa->ifa_ifp = ifp;
325 ifa->ifa_rtrequest = link_rtrequest;
326 TAILQ_INSERT_HEAD(&ifp->if_addrlist, ifa, ifa_list);
327 IFAREF(ifa);
328 ifa->ifa_addr = (struct sockaddr *)sdl;
329 ifp->if_sadl = sdl;
330 sdl = (struct sockaddr_dl *)(socksize + (caddr_t)sdl);
331 ifa->ifa_netmask = (struct sockaddr *)sdl;
332 sdl->sdl_len = masklen;
333 while (namelen != 0)
334 sdl->sdl_data[--namelen] = 0xff;
335 }
336
337 /*
338 * Free the link level name for the specified interface. This is
339 * a detach helper. This is called from if_detach() or from
340 * link layer type specific detach functions.
341 */
342 void
343 if_free_sadl(struct ifnet *ifp)
344 {
345 struct ifaddr *ifa;
346 int s;
347
348 ifa = ifnet_addrs[ifp->if_index];
349 if (ifa == NULL) {
350 KASSERT(ifp->if_sadl == NULL);
351 return;
352 }
353
354 KASSERT(ifp->if_sadl != NULL);
355
356 s = splnet();
357 rtinit(ifa, RTM_DELETE, 0);
358 TAILQ_REMOVE(&ifp->if_addrlist, ifa, ifa_list);
359 IFAFREE(ifa);
360
361 ifp->if_sadl = NULL;
362
363 ifnet_addrs[ifp->if_index] = NULL;
364 IFAFREE(ifa);
365 splx(s);
366 }
367
368 /*
369 * Attach an interface to the
370 * list of "active" interfaces.
371 */
372 void
373 if_attach(ifp)
374 struct ifnet *ifp;
375 {
376 int indexlim = 0;
377
378 if (if_indexlim == 0) {
379 TAILQ_INIT(&ifnet);
380 if_indexlim = 8;
381 }
382 TAILQ_INIT(&ifp->if_addrlist);
383 TAILQ_INSERT_TAIL(&ifnet, ifp, if_list);
384 ifp->if_index = if_index;
385 if (ifindex2ifnet == 0)
386 if_index++;
387 else
388 while (ifp->if_index < if_indexlim &&
389 ifindex2ifnet[ifp->if_index] != NULL) {
390 ++if_index;
391 if (if_index == 0)
392 if_index = 1;
393 /*
394 * If we hit USHRT_MAX, we skip back to 0 since
395 * there are a number of places where the value
396 * of if_index or if_index itself is compared
397 * to or stored in an unsigned short. By
398 * jumping back, we won't botch those assignments
399 * or comparisons.
400 */
401 else if (if_index == USHRT_MAX) {
402 /*
403 * However, if we have to jump back to
404 * zero *twice* without finding an empty
405 * slot in ifindex2ifnet[], then there
406 * there are too many (>65535) interfaces.
407 */
408 if (indexlim++)
409 panic("too many interfaces");
410 else
411 if_index = 1;
412 }
413 ifp->if_index = if_index;
414 }
415
416 /*
417 * We have some arrays that should be indexed by if_index.
418 * since if_index will grow dynamically, they should grow too.
419 * struct ifadd **ifnet_addrs
420 * struct ifnet **ifindex2ifnet
421 */
422 if (ifnet_addrs == 0 || ifindex2ifnet == 0 ||
423 ifp->if_index >= if_indexlim) {
424 size_t m, n, oldlim;
425 caddr_t q;
426
427 oldlim = if_indexlim;
428 while (ifp->if_index >= if_indexlim)
429 if_indexlim <<= 1;
430
431 /* grow ifnet_addrs */
432 m = oldlim * sizeof(struct ifaddr *);
433 n = if_indexlim * sizeof(struct ifaddr *);
434 q = (caddr_t)malloc(n, M_IFADDR, M_WAITOK);
435 memset(q, 0, n);
436 if (ifnet_addrs) {
437 bcopy((caddr_t)ifnet_addrs, q, m);
438 free((caddr_t)ifnet_addrs, M_IFADDR);
439 }
440 ifnet_addrs = (struct ifaddr **)q;
441
442 /* grow ifindex2ifnet */
443 m = oldlim * sizeof(struct ifnet *);
444 n = if_indexlim * sizeof(struct ifnet *);
445 q = (caddr_t)malloc(n, M_IFADDR, M_WAITOK);
446 memset(q, 0, n);
447 if (ifindex2ifnet) {
448 bcopy((caddr_t)ifindex2ifnet, q, m);
449 free((caddr_t)ifindex2ifnet, M_IFADDR);
450 }
451 ifindex2ifnet = (struct ifnet **)q;
452 }
453
454 ifindex2ifnet[ifp->if_index] = ifp;
455
456 /*
457 * Link level name is allocated later by a separate call to
458 * if_alloc_sadl().
459 */
460
461 if (ifp->if_snd.ifq_maxlen == 0)
462 ifp->if_snd.ifq_maxlen = ifqmaxlen;
463 ifp->if_broadcastaddr = 0; /* reliably crash if used uninitialized */
464
465 ifp->if_link_state = LINK_STATE_UNKNOWN;
466
467 ifp->if_capenable = 0;
468 ifp->if_csum_flags_tx = 0;
469 ifp->if_csum_flags_rx = 0;
470
471 #ifdef ALTQ
472 ifp->if_snd.altq_type = 0;
473 ifp->if_snd.altq_disc = NULL;
474 ifp->if_snd.altq_flags &= ALTQF_CANTCHANGE;
475 ifp->if_snd.altq_tbr = NULL;
476 ifp->if_snd.altq_ifp = ifp;
477 #endif
478
479 #ifdef PFIL_HOOKS
480 ifp->if_pfil.ph_type = PFIL_TYPE_IFNET;
481 ifp->if_pfil.ph_ifnet = ifp;
482 if (pfil_head_register(&ifp->if_pfil) != 0)
483 printf("%s: WARNING: unable to register pfil hook\n",
484 ifp->if_xname);
485 (void)pfil_run_hooks(&if_pfil,
486 (struct mbuf **)PFIL_IFNET_ATTACH, ifp, PFIL_IFNET);
487 #endif
488
489 if (!STAILQ_EMPTY(&domains))
490 if_attachdomain1(ifp);
491
492 /* Announce the interface. */
493 rt_ifannouncemsg(ifp, IFAN_ARRIVAL);
494 }
495
496 void
497 if_attachdomain()
498 {
499 struct ifnet *ifp;
500 int s;
501
502 s = splnet();
503 for (ifp = TAILQ_FIRST(&ifnet); ifp; ifp = TAILQ_NEXT(ifp, if_list))
504 if_attachdomain1(ifp);
505 splx(s);
506 }
507
508 void
509 if_attachdomain1(ifp)
510 struct ifnet *ifp;
511 {
512 struct domain *dp;
513 int s;
514
515 s = splnet();
516
517 /* address family dependent data region */
518 memset(ifp->if_afdata, 0, sizeof(ifp->if_afdata));
519 DOMAIN_FOREACH(dp) {
520 if (dp->dom_ifattach)
521 ifp->if_afdata[dp->dom_family] =
522 (*dp->dom_ifattach)(ifp);
523 }
524
525 splx(s);
526 }
527
528 /*
529 * Deactivate an interface. This points all of the procedure
530 * handles at error stubs. May be called from interrupt context.
531 */
532 void
533 if_deactivate(ifp)
534 struct ifnet *ifp;
535 {
536 int s;
537
538 s = splnet();
539
540 ifp->if_output = if_nulloutput;
541 ifp->if_input = if_nullinput;
542 ifp->if_start = if_nullstart;
543 ifp->if_ioctl = if_nullioctl;
544 ifp->if_init = if_nullinit;
545 ifp->if_stop = if_nullstop;
546 ifp->if_watchdog = if_nullwatchdog;
547 ifp->if_drain = if_nulldrain;
548
549 /* No more packets may be enqueued. */
550 ifp->if_snd.ifq_maxlen = 0;
551
552 splx(s);
553 }
554
555 /*
556 * Detach an interface from the list of "active" interfaces,
557 * freeing any resources as we go along.
558 *
559 * NOTE: This routine must be called with a valid thread context,
560 * as it may block.
561 */
562 void
563 if_detach(ifp)
564 struct ifnet *ifp;
565 {
566 struct socket so;
567 struct ifaddr *ifa, **ifap;
568 #ifdef IFAREF_DEBUG
569 struct ifaddr *last_ifa = NULL;
570 #endif
571 struct domain *dp;
572 const struct protosw *pr;
573 struct radix_node_head *rnh;
574 int s, i, family, purged;
575
576 /*
577 * XXX It's kind of lame that we have to have the
578 * XXX socket structure...
579 */
580 memset(&so, 0, sizeof(so));
581
582 s = splnet();
583
584 /*
585 * Do an if_down() to give protocols a chance to do something.
586 */
587 if_down(ifp);
588
589 #ifdef ALTQ
590 if (ALTQ_IS_ENABLED(&ifp->if_snd))
591 altq_disable(&ifp->if_snd);
592 if (ALTQ_IS_ATTACHED(&ifp->if_snd))
593 altq_detach(&ifp->if_snd);
594 #endif
595
596 #ifdef PFIL_HOOKS
597 (void)pfil_run_hooks(&if_pfil,
598 (struct mbuf **)PFIL_IFNET_DETACH, ifp, PFIL_IFNET);
599 (void)pfil_head_unregister(&ifp->if_pfil);
600 #endif
601
602 /*
603 * Rip all the addresses off the interface. This should make
604 * all of the routes go away.
605 */
606 ifap = &TAILQ_FIRST(&ifp->if_addrlist); /* XXX abstraction violation */
607 while ((ifa = *ifap)) {
608 family = ifa->ifa_addr->sa_family;
609 #ifdef IFAREF_DEBUG
610 printf("if_detach: ifaddr %p, family %d, refcnt %d\n",
611 ifa, family, ifa->ifa_refcnt);
612 if (last_ifa != NULL && ifa == last_ifa)
613 panic("if_detach: loop detected");
614 last_ifa = ifa;
615 #endif
616 if (family == AF_LINK) {
617 ifap = &TAILQ_NEXT(ifa, ifa_list);
618 continue;
619 }
620 dp = pffinddomain(family);
621 #ifdef DIAGNOSTIC
622 if (dp == NULL)
623 panic("if_detach: no domain for AF %d",
624 family);
625 #endif
626 /*
627 * XXX These PURGEIF calls are redundant with the
628 * purge-all-families calls below, but are left in for
629 * now both to make a smaller change, and to avoid
630 * unplanned interactions with clearing of
631 * ifp->if_addrlist.
632 */
633 purged = 0;
634 for (pr = dp->dom_protosw;
635 pr < dp->dom_protoswNPROTOSW; pr++) {
636 so.so_proto = pr;
637 if (pr->pr_usrreq != NULL) {
638 (void) (*pr->pr_usrreq)(&so,
639 PRU_PURGEIF, NULL, NULL,
640 (struct mbuf *) ifp, curproc);
641 purged = 1;
642 }
643 }
644 if (purged == 0) {
645 /*
646 * XXX What's really the best thing to do
647 * XXX here? --thorpej (at) NetBSD.org
648 */
649 printf("if_detach: WARNING: AF %d not purged\n",
650 family);
651 TAILQ_REMOVE(&ifp->if_addrlist, ifa, ifa_list);
652 }
653 }
654
655 if_free_sadl(ifp);
656
657 /* Walk the routing table looking for straglers. */
658 for (i = 0; i <= AF_MAX; i++) {
659 if ((rnh = rt_tables[i]) != NULL)
660 (void) (*rnh->rnh_walktree)(rnh, if_rt_walktree, ifp);
661 }
662
663 DOMAIN_FOREACH(dp) {
664 if (dp->dom_ifdetach && ifp->if_afdata[dp->dom_family])
665 (*dp->dom_ifdetach)(ifp,
666 ifp->if_afdata[dp->dom_family]);
667
668 /*
669 * One would expect multicast memberships (INET and
670 * INET6) on UDP sockets to be purged by the PURGEIF
671 * calls above, but if all addresses were removed from
672 * the interface prior to destruction, the calls will
673 * not be made (e.g. ppp, for which pppd(8) generally
674 * removes addresses before destroying the interface).
675 * Because there is no invariant that multicast
676 * memberships only exist for interfaces with IPv4
677 * addresses, we must call PURGEIF regardless of
678 * addresses. (Protocols which might store ifnet
679 * pointers are marked with PR_PURGEIF.)
680 */
681 for (pr = dp->dom_protosw;
682 pr < dp->dom_protoswNPROTOSW; pr++) {
683 so.so_proto = pr;
684 if (pr->pr_usrreq != NULL &&
685 pr->pr_flags & PR_PURGEIF)
686 (void) (*pr->pr_usrreq)(&so,
687 PRU_PURGEIF, NULL, NULL,
688 (struct mbuf *) ifp, curproc);
689 }
690 }
691
692 /* Announce that the interface is gone. */
693 rt_ifannouncemsg(ifp, IFAN_DEPARTURE);
694
695 ifindex2ifnet[ifp->if_index] = NULL;
696
697 TAILQ_REMOVE(&ifnet, ifp, if_list);
698
699 /*
700 * remove packets came from ifp, from software interrupt queues.
701 * net/netisr_dispatch.h is not usable, as some of them use
702 * strange queue names.
703 */
704 #define IF_DETACH_QUEUES(x) \
705 do { \
706 extern struct ifqueue x; \
707 if_detach_queues(ifp, & x); \
708 } while (/*CONSTCOND*/ 0)
709 #ifdef INET
710 #if NARP > 0
711 IF_DETACH_QUEUES(arpintrq);
712 #endif
713 IF_DETACH_QUEUES(ipintrq);
714 #endif
715 #ifdef INET6
716 IF_DETACH_QUEUES(ip6intrq);
717 #endif
718 #ifdef NETATALK
719 IF_DETACH_QUEUES(atintrq1);
720 IF_DETACH_QUEUES(atintrq2);
721 #endif
722 #ifdef NS
723 IF_DETACH_QUEUES(nsintrq);
724 #endif
725 #ifdef ISO
726 IF_DETACH_QUEUES(clnlintrq);
727 #endif
728 #ifdef CCITT
729 IF_DETACH_QUEUES(llcintrq);
730 IF_DETACH_QUEUES(hdintrq);
731 #endif
732 #ifdef NATM
733 IF_DETACH_QUEUES(natmintrq);
734 #endif
735 #ifdef DECNET
736 IF_DETACH_QUEUES(decnetintrq);
737 #endif
738 #undef IF_DETACH_QUEUES
739
740 splx(s);
741 }
742
743 #if defined(INET) || defined(INET6) || defined(NETATALK) || defined(NS) || \
744 defined(ISO) || defined(CCITT) || defined(NATM) || defined(DECNET)
745 static void
746 if_detach_queues(ifp, q)
747 struct ifnet *ifp;
748 struct ifqueue *q;
749 {
750 struct mbuf *m, *prev, *next;
751
752 prev = NULL;
753 for (m = q->ifq_head; m; m = next) {
754 next = m->m_nextpkt;
755 #ifdef DIAGNOSTIC
756 if ((m->m_flags & M_PKTHDR) == 0) {
757 prev = m;
758 continue;
759 }
760 #endif
761 if (m->m_pkthdr.rcvif != ifp) {
762 prev = m;
763 continue;
764 }
765
766 if (prev)
767 prev->m_nextpkt = m->m_nextpkt;
768 else
769 q->ifq_head = m->m_nextpkt;
770 if (q->ifq_tail == m)
771 q->ifq_tail = prev;
772 q->ifq_len--;
773
774 m->m_nextpkt = NULL;
775 m_freem(m);
776 IF_DROP(q);
777 }
778 }
779 #endif /* defined(INET) || ... */
780
781 /*
782 * Callback for a radix tree walk to delete all references to an
783 * ifnet.
784 */
785 int
786 if_rt_walktree(rn, v)
787 struct radix_node *rn;
788 void *v;
789 {
790 struct ifnet *ifp = (struct ifnet *)v;
791 struct rtentry *rt = (struct rtentry *)rn;
792 int error;
793
794 if (rt->rt_ifp == ifp) {
795 /* Delete the entry. */
796 error = rtrequest(RTM_DELETE, rt_key(rt), rt->rt_gateway,
797 rt_mask(rt), rt->rt_flags, NULL);
798 if (error)
799 printf("%s: warning: unable to delete rtentry @ %p, "
800 "error = %d\n", ifp->if_xname, rt, error);
801 }
802 return (0);
803 }
804
805 /*
806 * Create a clone network interface.
807 */
808 int
809 if_clone_create(name)
810 const char *name;
811 {
812 struct if_clone *ifc;
813 int unit;
814
815 ifc = if_clone_lookup(name, &unit);
816 if (ifc == NULL)
817 return (EINVAL);
818
819 if (ifunit(name) != NULL)
820 return (EEXIST);
821
822 return ((*ifc->ifc_create)(ifc, unit));
823 }
824
825 /*
826 * Destroy a clone network interface.
827 */
828 int
829 if_clone_destroy(name)
830 const char *name;
831 {
832 struct if_clone *ifc;
833 struct ifnet *ifp;
834
835 ifc = if_clone_lookup(name, NULL);
836 if (ifc == NULL)
837 return (EINVAL);
838
839 ifp = ifunit(name);
840 if (ifp == NULL)
841 return (ENXIO);
842
843 if (ifc->ifc_destroy == NULL)
844 return (EOPNOTSUPP);
845
846 return ((*ifc->ifc_destroy)(ifp));
847 }
848
849 /*
850 * Look up a network interface cloner.
851 */
852 struct if_clone *
853 if_clone_lookup(name, unitp)
854 const char *name;
855 int *unitp;
856 {
857 struct if_clone *ifc;
858 const char *cp;
859 int unit;
860
861 /* separate interface name from unit */
862 for (cp = name;
863 cp - name < IFNAMSIZ && *cp && (*cp < '0' || *cp > '9');
864 cp++)
865 continue;
866
867 if (cp == name || cp - name == IFNAMSIZ || !*cp)
868 return (NULL); /* No name or unit number */
869
870 LIST_FOREACH(ifc, &if_cloners, ifc_list) {
871 if (strlen(ifc->ifc_name) == cp - name &&
872 !strncmp(name, ifc->ifc_name, cp - name))
873 break;
874 }
875
876 if (ifc == NULL)
877 return (NULL);
878
879 unit = 0;
880 while (cp - name < IFNAMSIZ && *cp) {
881 if (*cp < '0' || *cp > '9' || unit > INT_MAX / 10) {
882 /* Bogus unit number. */
883 return (NULL);
884 }
885 unit = (unit * 10) + (*cp++ - '0');
886 }
887
888 if (unitp != NULL)
889 *unitp = unit;
890 return (ifc);
891 }
892
893 /*
894 * Register a network interface cloner.
895 */
896 void
897 if_clone_attach(ifc)
898 struct if_clone *ifc;
899 {
900
901 LIST_INSERT_HEAD(&if_cloners, ifc, ifc_list);
902 if_cloners_count++;
903 }
904
905 /*
906 * Unregister a network interface cloner.
907 */
908 void
909 if_clone_detach(ifc)
910 struct if_clone *ifc;
911 {
912
913 LIST_REMOVE(ifc, ifc_list);
914 if_cloners_count--;
915 }
916
917 /*
918 * Provide list of interface cloners to userspace.
919 */
920 int
921 if_clone_list(ifcr)
922 struct if_clonereq *ifcr;
923 {
924 char outbuf[IFNAMSIZ], *dst;
925 struct if_clone *ifc;
926 int count, error = 0;
927
928 ifcr->ifcr_total = if_cloners_count;
929 if ((dst = ifcr->ifcr_buffer) == NULL) {
930 /* Just asking how many there are. */
931 return (0);
932 }
933
934 if (ifcr->ifcr_count < 0)
935 return (EINVAL);
936
937 count = (if_cloners_count < ifcr->ifcr_count) ?
938 if_cloners_count : ifcr->ifcr_count;
939
940 for (ifc = LIST_FIRST(&if_cloners); ifc != NULL && count != 0;
941 ifc = LIST_NEXT(ifc, ifc_list), count--, dst += IFNAMSIZ) {
942 strncpy(outbuf, ifc->ifc_name, IFNAMSIZ);
943 outbuf[IFNAMSIZ - 1] = '\0'; /* sanity */
944 error = copyout(outbuf, dst, IFNAMSIZ);
945 if (error)
946 break;
947 }
948
949 return (error);
950 }
951
952 /*
953 * Locate an interface based on a complete address.
954 */
955 /*ARGSUSED*/
956 struct ifaddr *
957 ifa_ifwithaddr(addr)
958 const struct sockaddr *addr;
959 {
960 struct ifnet *ifp;
961 struct ifaddr *ifa;
962
963 #define equal(a1, a2) \
964 (bcmp((a1), (a2), ((const struct sockaddr *)(a1))->sa_len) == 0)
965
966 for (ifp = TAILQ_FIRST(&ifnet); ifp != NULL;
967 ifp = TAILQ_NEXT(ifp, if_list)) {
968 if (ifp->if_output == if_nulloutput)
969 continue;
970 for (ifa = TAILQ_FIRST(&ifp->if_addrlist); ifa != NULL;
971 ifa = TAILQ_NEXT(ifa, ifa_list)) {
972 if (ifa->ifa_addr->sa_family != addr->sa_family)
973 continue;
974 if (equal(addr, ifa->ifa_addr))
975 return (ifa);
976 if ((ifp->if_flags & IFF_BROADCAST) &&
977 ifa->ifa_broadaddr &&
978 /* IP6 doesn't have broadcast */
979 ifa->ifa_broadaddr->sa_len != 0 &&
980 equal(ifa->ifa_broadaddr, addr))
981 return (ifa);
982 }
983 }
984 return (NULL);
985 }
986
987 /*
988 * Locate the point to point interface with a given destination address.
989 */
990 /*ARGSUSED*/
991 struct ifaddr *
992 ifa_ifwithdstaddr(addr)
993 const struct sockaddr *addr;
994 {
995 struct ifnet *ifp;
996 struct ifaddr *ifa;
997
998 for (ifp = TAILQ_FIRST(&ifnet); ifp != NULL;
999 ifp = TAILQ_NEXT(ifp, if_list)) {
1000 if (ifp->if_output == if_nulloutput)
1001 continue;
1002 if (ifp->if_flags & IFF_POINTOPOINT) {
1003 for (ifa = TAILQ_FIRST(&ifp->if_addrlist); ifa != NULL;
1004 ifa = TAILQ_NEXT(ifa, ifa_list)) {
1005 if (ifa->ifa_addr->sa_family !=
1006 addr->sa_family ||
1007 ifa->ifa_dstaddr == NULL)
1008 continue;
1009 if (equal(addr, ifa->ifa_dstaddr))
1010 return (ifa);
1011 }
1012 }
1013 }
1014 return (NULL);
1015 }
1016
1017 /*
1018 * Find an interface on a specific network. If many, choice
1019 * is most specific found.
1020 */
1021 struct ifaddr *
1022 ifa_ifwithnet(addr)
1023 const struct sockaddr *addr;
1024 {
1025 struct ifnet *ifp;
1026 struct ifaddr *ifa;
1027 const struct sockaddr_dl *sdl;
1028 struct ifaddr *ifa_maybe = 0;
1029 u_int af = addr->sa_family;
1030 char *addr_data = addr->sa_data, *cplim;
1031
1032 if (af == AF_LINK) {
1033 sdl = (const struct sockaddr_dl *)addr;
1034 if (sdl->sdl_index && sdl->sdl_index < if_indexlim &&
1035 ifindex2ifnet[sdl->sdl_index] &&
1036 ifindex2ifnet[sdl->sdl_index]->if_output != if_nulloutput)
1037 return (ifnet_addrs[sdl->sdl_index]);
1038 }
1039 #ifdef NETATALK
1040 if (af == AF_APPLETALK) {
1041 const struct sockaddr_at *sat, *sat2;
1042 sat = (const struct sockaddr_at *)addr;
1043 for (ifp = TAILQ_FIRST(&ifnet); ifp != NULL;
1044 ifp = TAILQ_NEXT(ifp, if_list)) {
1045 if (ifp->if_output == if_nulloutput)
1046 continue;
1047 ifa = at_ifawithnet((const struct sockaddr_at *)addr, ifp);
1048 if (ifa == NULL)
1049 continue;
1050 sat2 = (struct sockaddr_at *)ifa->ifa_addr;
1051 if (sat2->sat_addr.s_net == sat->sat_addr.s_net)
1052 return (ifa); /* exact match */
1053 if (ifa_maybe == NULL) {
1054 /* else keep the if with the right range */
1055 ifa_maybe = ifa;
1056 }
1057 }
1058 return (ifa_maybe);
1059 }
1060 #endif
1061 for (ifp = TAILQ_FIRST(&ifnet); ifp != NULL;
1062 ifp = TAILQ_NEXT(ifp, if_list)) {
1063 if (ifp->if_output == if_nulloutput)
1064 continue;
1065 for (ifa = TAILQ_FIRST(&ifp->if_addrlist); ifa != NULL;
1066 ifa = TAILQ_NEXT(ifa, ifa_list)) {
1067 char *cp, *cp2, *cp3;
1068
1069 if (ifa->ifa_addr->sa_family != af ||
1070 ifa->ifa_netmask == 0)
1071 next: continue;
1072 cp = addr_data;
1073 cp2 = ifa->ifa_addr->sa_data;
1074 cp3 = ifa->ifa_netmask->sa_data;
1075 cplim = (char *)ifa->ifa_netmask +
1076 ifa->ifa_netmask->sa_len;
1077 while (cp3 < cplim) {
1078 if ((*cp++ ^ *cp2++) & *cp3++) {
1079 /* want to continue for() loop */
1080 goto next;
1081 }
1082 }
1083 if (ifa_maybe == 0 ||
1084 rn_refines((caddr_t)ifa->ifa_netmask,
1085 (caddr_t)ifa_maybe->ifa_netmask))
1086 ifa_maybe = ifa;
1087 }
1088 }
1089 return (ifa_maybe);
1090 }
1091
1092 /*
1093 * Find the interface of the addresss.
1094 */
1095 struct ifaddr *
1096 ifa_ifwithladdr(addr)
1097 const struct sockaddr *addr;
1098 {
1099 struct ifaddr *ia;
1100
1101 if ((ia = ifa_ifwithaddr(addr)) || (ia = ifa_ifwithdstaddr(addr)) ||
1102 (ia = ifa_ifwithnet(addr)))
1103 return (ia);
1104 return (NULL);
1105 }
1106
1107 /*
1108 * Find an interface using a specific address family
1109 */
1110 struct ifaddr *
1111 ifa_ifwithaf(af)
1112 int af;
1113 {
1114 struct ifnet *ifp;
1115 struct ifaddr *ifa;
1116
1117 for (ifp = TAILQ_FIRST(&ifnet); ifp != NULL;
1118 ifp = TAILQ_NEXT(ifp, if_list)) {
1119 if (ifp->if_output == if_nulloutput)
1120 continue;
1121 for (ifa = TAILQ_FIRST(&ifp->if_addrlist); ifa != NULL;
1122 ifa = TAILQ_NEXT(ifa, ifa_list)) {
1123 if (ifa->ifa_addr->sa_family == af)
1124 return (ifa);
1125 }
1126 }
1127 return (NULL);
1128 }
1129
1130 /*
1131 * Find an interface address specific to an interface best matching
1132 * a given address.
1133 */
1134 struct ifaddr *
1135 ifaof_ifpforaddr(addr, ifp)
1136 const struct sockaddr *addr;
1137 struct ifnet *ifp;
1138 {
1139 struct ifaddr *ifa;
1140 const char *cp, *cp2, *cp3;
1141 const char *cplim;
1142 struct ifaddr *ifa_maybe = 0;
1143 u_int af = addr->sa_family;
1144
1145 if (ifp->if_output == if_nulloutput)
1146 return (NULL);
1147
1148 if (af >= AF_MAX)
1149 return (NULL);
1150
1151 for (ifa = TAILQ_FIRST(&ifp->if_addrlist); ifa != NULL;
1152 ifa = TAILQ_NEXT(ifa, ifa_list)) {
1153 if (ifa->ifa_addr->sa_family != af)
1154 continue;
1155 ifa_maybe = ifa;
1156 if (ifa->ifa_netmask == 0) {
1157 if (equal(addr, ifa->ifa_addr) ||
1158 (ifa->ifa_dstaddr &&
1159 equal(addr, ifa->ifa_dstaddr)))
1160 return (ifa);
1161 continue;
1162 }
1163 cp = addr->sa_data;
1164 cp2 = ifa->ifa_addr->sa_data;
1165 cp3 = ifa->ifa_netmask->sa_data;
1166 cplim = ifa->ifa_netmask->sa_len + (char *)ifa->ifa_netmask;
1167 for (; cp3 < cplim; cp3++) {
1168 if ((*cp++ ^ *cp2++) & *cp3)
1169 break;
1170 }
1171 if (cp3 == cplim)
1172 return (ifa);
1173 }
1174 return (ifa_maybe);
1175 }
1176
1177 /*
1178 * Default action when installing a route with a Link Level gateway.
1179 * Lookup an appropriate real ifa to point to.
1180 * This should be moved to /sys/net/link.c eventually.
1181 */
1182 void
1183 link_rtrequest(cmd, rt, info)
1184 int cmd;
1185 struct rtentry *rt;
1186 struct rt_addrinfo *info;
1187 {
1188 struct ifaddr *ifa;
1189 struct sockaddr *dst;
1190 struct ifnet *ifp;
1191
1192 if (cmd != RTM_ADD || ((ifa = rt->rt_ifa) == 0) ||
1193 ((ifp = ifa->ifa_ifp) == 0) || ((dst = rt_key(rt)) == 0))
1194 return;
1195 if ((ifa = ifaof_ifpforaddr(dst, ifp)) != NULL) {
1196 IFAFREE(rt->rt_ifa);
1197 rt->rt_ifa = ifa;
1198 IFAREF(ifa);
1199 if (ifa->ifa_rtrequest && ifa->ifa_rtrequest != link_rtrequest)
1200 ifa->ifa_rtrequest(cmd, rt, info);
1201 }
1202 }
1203
1204 /*
1205 * Handle a change in the interface link state.
1206 */
1207 void
1208 if_link_state_change(struct ifnet *ifp, int link_state)
1209 {
1210 /* Notify that the link state has changed. */
1211 if (ifp->if_link_state != link_state) {
1212 ifp->if_link_state = link_state;
1213 rt_ifmsg(ifp);
1214 log(LOG_NOTICE, "%s: link state changed to %s\n", ifp->if_xname,
1215 (link_state == LINK_STATE_UP) ? "UP" : "DOWN" );
1216 }
1217 }
1218
1219 /*
1220 * Mark an interface down and notify protocols of
1221 * the transition.
1222 * NOTE: must be called at splsoftnet or equivalent.
1223 */
1224 void
1225 if_down(ifp)
1226 struct ifnet *ifp;
1227 {
1228 struct ifaddr *ifa;
1229
1230 ifp->if_flags &= ~IFF_UP;
1231 microtime(&ifp->if_lastchange);
1232 for (ifa = TAILQ_FIRST(&ifp->if_addrlist); ifa != NULL;
1233 ifa = TAILQ_NEXT(ifa, ifa_list))
1234 pfctlinput(PRC_IFDOWN, ifa->ifa_addr);
1235 IFQ_PURGE(&ifp->if_snd);
1236 rt_ifmsg(ifp);
1237 }
1238
1239 /*
1240 * Mark an interface up and notify protocols of
1241 * the transition.
1242 * NOTE: must be called at splsoftnet or equivalent.
1243 */
1244 void
1245 if_up(ifp)
1246 struct ifnet *ifp;
1247 {
1248 #ifdef notyet
1249 struct ifaddr *ifa;
1250 #endif
1251
1252 ifp->if_flags |= IFF_UP;
1253 microtime(&ifp->if_lastchange);
1254 #ifdef notyet
1255 /* this has no effect on IP, and will kill all ISO connections XXX */
1256 for (ifa = TAILQ_FIRST(&ifp->if_addrlist); ifa != NULL;
1257 ifa = TAILQ_NEXT(ifa, ifa_list))
1258 pfctlinput(PRC_IFUP, ifa->ifa_addr);
1259 #endif
1260 rt_ifmsg(ifp);
1261 #ifdef INET6
1262 in6_if_up(ifp);
1263 #endif
1264 }
1265
1266 /*
1267 * Handle interface watchdog timer routines. Called
1268 * from softclock, we decrement timers (if set) and
1269 * call the appropriate interface routine on expiration.
1270 */
1271 void
1272 if_slowtimo(arg)
1273 void *arg;
1274 {
1275 struct ifnet *ifp;
1276 int s = splnet();
1277
1278 for (ifp = TAILQ_FIRST(&ifnet); ifp != NULL;
1279 ifp = TAILQ_NEXT(ifp, if_list)) {
1280 if (ifp->if_timer == 0 || --ifp->if_timer)
1281 continue;
1282 if (ifp->if_watchdog)
1283 (*ifp->if_watchdog)(ifp);
1284 }
1285 splx(s);
1286 callout_reset(&if_slowtimo_ch, hz / IFNET_SLOWHZ,
1287 if_slowtimo, NULL);
1288 }
1289
1290 /*
1291 * Set/clear promiscuous mode on interface ifp based on the truth value
1292 * of pswitch. The calls are reference counted so that only the first
1293 * "on" request actually has an effect, as does the final "off" request.
1294 * Results are undefined if the "off" and "on" requests are not matched.
1295 */
1296 int
1297 ifpromisc(ifp, pswitch)
1298 struct ifnet *ifp;
1299 int pswitch;
1300 {
1301 int pcount, ret;
1302 short flags;
1303 struct ifreq ifr;
1304
1305 pcount = ifp->if_pcount;
1306 flags = ifp->if_flags;
1307 if (pswitch) {
1308 /*
1309 * Allow the device to be "placed" into promiscuous
1310 * mode even if it is not configured up. It will
1311 * consult IFF_PROMISC when it is is brought up.
1312 */
1313 if (ifp->if_pcount++ != 0)
1314 return (0);
1315 ifp->if_flags |= IFF_PROMISC;
1316 if ((ifp->if_flags & IFF_UP) == 0)
1317 return (0);
1318 } else {
1319 if (--ifp->if_pcount > 0)
1320 return (0);
1321 ifp->if_flags &= ~IFF_PROMISC;
1322 /*
1323 * If the device is not configured up, we should not need to
1324 * turn off promiscuous mode (device should have turned it
1325 * off when interface went down; and will look at IFF_PROMISC
1326 * again next time interface comes up).
1327 */
1328 if ((ifp->if_flags & IFF_UP) == 0)
1329 return (0);
1330 }
1331 memset(&ifr, 0, sizeof(ifr));
1332 ifr.ifr_flags = ifp->if_flags;
1333 ret = (*ifp->if_ioctl)(ifp, SIOCSIFFLAGS, (caddr_t) &ifr);
1334 /* Restore interface state if not successful. */
1335 if (ret != 0) {
1336 ifp->if_pcount = pcount;
1337 ifp->if_flags = flags;
1338 }
1339 return (ret);
1340 }
1341
1342 /*
1343 * Map interface name to
1344 * interface structure pointer.
1345 */
1346 struct ifnet *
1347 ifunit(name)
1348 const char *name;
1349 {
1350 struct ifnet *ifp;
1351 const char *cp = name;
1352 u_int unit = 0;
1353 u_int i;
1354
1355 /*
1356 * If the entire name is a number, treat it as an ifindex.
1357 */
1358 for (i = 0; i < IFNAMSIZ && *cp >= '0' && *cp <= '9'; i++, cp++) {
1359 unit = unit * 10 + (*cp - '0');
1360 }
1361
1362 /*
1363 * If the number took all of the name, then it's a valid ifindex.
1364 */
1365 if (i == IFNAMSIZ || (cp != name && *cp == '\0')) {
1366 if (unit >= if_indexlim)
1367 return (NULL);
1368 ifp = ifindex2ifnet[unit];
1369 if (ifp == NULL || ifp->if_output == if_nulloutput)
1370 return (NULL);
1371 return (ifp);
1372 }
1373
1374 for (ifp = TAILQ_FIRST(&ifnet); ifp != NULL;
1375 ifp = TAILQ_NEXT(ifp, if_list)) {
1376 if (ifp->if_output == if_nulloutput)
1377 continue;
1378 if (strcmp(ifp->if_xname, name) == 0)
1379 return (ifp);
1380 }
1381 return (NULL);
1382 }
1383
1384 /*
1385 * Interface ioctls.
1386 */
1387 int
1388 ifioctl(so, cmd, data, p)
1389 struct socket *so;
1390 u_long cmd;
1391 caddr_t data;
1392 struct proc *p;
1393 {
1394 struct ifnet *ifp;
1395 struct ifreq *ifr;
1396 struct ifcapreq *ifcr;
1397 struct ifdatareq *ifdr;
1398 int s, error = 0;
1399 short oif_flags;
1400
1401 switch (cmd) {
1402
1403 case SIOCGIFCONF:
1404 case OSIOCGIFCONF:
1405 return (ifconf(cmd, data));
1406 }
1407 ifr = (struct ifreq *)data;
1408 ifcr = (struct ifcapreq *)data;
1409 ifdr = (struct ifdatareq *)data;
1410
1411 switch (cmd) {
1412 case SIOCIFCREATE:
1413 case SIOCIFDESTROY:
1414 if (p) {
1415 error = suser(p->p_ucred, &p->p_acflag);
1416 if (error)
1417 return error;
1418 }
1419 return ((cmd == SIOCIFCREATE) ?
1420 if_clone_create(ifr->ifr_name) :
1421 if_clone_destroy(ifr->ifr_name));
1422
1423 case SIOCIFGCLONERS:
1424 return (if_clone_list((struct if_clonereq *)data));
1425 }
1426
1427 ifp = ifunit(ifr->ifr_name);
1428 if (ifp == 0)
1429 return (ENXIO);
1430
1431 switch (cmd) {
1432 case SIOCSIFFLAGS:
1433 case SIOCSIFCAP:
1434 case SIOCSIFMETRIC:
1435 case SIOCZIFDATA:
1436 case SIOCSIFMTU:
1437 case SIOCSIFPHYADDR:
1438 case SIOCDIFPHYADDR:
1439 #ifdef INET6
1440 case SIOCSIFPHYADDR_IN6:
1441 #endif
1442 case SIOCSLIFPHYADDR:
1443 case SIOCADDMULTI:
1444 case SIOCDELMULTI:
1445 case SIOCSIFMEDIA:
1446 case SIOCSDRVSPEC:
1447 case SIOCS80211NWID:
1448 case SIOCS80211NWKEY:
1449 case SIOCS80211POWER:
1450 case SIOCS80211BSSID:
1451 case SIOCS80211CHANNEL:
1452 if (p) {
1453 error = suser(p->p_ucred, &p->p_acflag);
1454 if (error)
1455 return error;
1456 }
1457 }
1458
1459 oif_flags = ifp->if_flags;
1460 switch (cmd) {
1461
1462 case SIOCGIFFLAGS:
1463 ifr->ifr_flags = ifp->if_flags;
1464 break;
1465
1466 case SIOCGIFMETRIC:
1467 ifr->ifr_metric = ifp->if_metric;
1468 break;
1469
1470 case SIOCGIFMTU:
1471 ifr->ifr_mtu = ifp->if_mtu;
1472 break;
1473
1474 case SIOCGIFDLT:
1475 ifr->ifr_dlt = ifp->if_dlt;
1476 break;
1477
1478 case SIOCSIFFLAGS:
1479 if (ifp->if_flags & IFF_UP && (ifr->ifr_flags & IFF_UP) == 0) {
1480 s = splnet();
1481 if_down(ifp);
1482 splx(s);
1483 }
1484 if (ifr->ifr_flags & IFF_UP && (ifp->if_flags & IFF_UP) == 0) {
1485 s = splnet();
1486 if_up(ifp);
1487 splx(s);
1488 }
1489 ifp->if_flags = (ifp->if_flags & IFF_CANTCHANGE) |
1490 (ifr->ifr_flags &~ IFF_CANTCHANGE);
1491 if (ifp->if_ioctl)
1492 (void) (*ifp->if_ioctl)(ifp, cmd, data);
1493 break;
1494
1495 case SIOCGIFCAP:
1496 ifcr->ifcr_capabilities = ifp->if_capabilities;
1497 ifcr->ifcr_capenable = ifp->if_capenable;
1498 break;
1499
1500 case SIOCSIFCAP:
1501 if ((ifcr->ifcr_capenable & ~ifp->if_capabilities) != 0)
1502 return (EINVAL);
1503 if (ifp->if_ioctl == NULL)
1504 return (EOPNOTSUPP);
1505
1506 /* Must prevent race with packet reception here. */
1507 s = splnet();
1508 if (ifcr->ifcr_capenable != ifp->if_capenable) {
1509 struct ifreq ifrq;
1510
1511 ifrq.ifr_flags = ifp->if_flags;
1512 ifp->if_capenable = ifcr->ifcr_capenable;
1513
1514 /* Pre-compute the checksum flags mask. */
1515 ifp->if_csum_flags_tx = 0;
1516 ifp->if_csum_flags_rx = 0;
1517 if (ifp->if_capenable & IFCAP_CSUM_IPv4_Tx) {
1518 ifp->if_csum_flags_tx |= M_CSUM_IPv4;
1519 }
1520 if (ifp->if_capenable & IFCAP_CSUM_IPv4_Rx) {
1521 ifp->if_csum_flags_rx |= M_CSUM_IPv4;
1522 }
1523
1524 if (ifp->if_capenable & IFCAP_CSUM_TCPv4_Tx) {
1525 ifp->if_csum_flags_tx |= M_CSUM_TCPv4;
1526 }
1527 if (ifp->if_capenable & IFCAP_CSUM_TCPv4_Rx) {
1528 ifp->if_csum_flags_rx |= M_CSUM_TCPv4;
1529 }
1530
1531 if (ifp->if_capenable & IFCAP_CSUM_UDPv4_Tx) {
1532 ifp->if_csum_flags_tx |= M_CSUM_UDPv4;
1533 }
1534 if (ifp->if_capenable & IFCAP_CSUM_UDPv4_Rx) {
1535 ifp->if_csum_flags_rx |= M_CSUM_UDPv4;
1536 }
1537
1538 if (ifp->if_capenable & IFCAP_CSUM_TCPv6_Tx) {
1539 ifp->if_csum_flags_tx |= M_CSUM_TCPv6;
1540 }
1541 if (ifp->if_capenable & IFCAP_CSUM_TCPv6_Rx) {
1542 ifp->if_csum_flags_rx |= M_CSUM_TCPv6;
1543 }
1544
1545 if (ifp->if_capenable & IFCAP_CSUM_UDPv6_Tx) {
1546 ifp->if_csum_flags_tx |= M_CSUM_UDPv6;
1547 }
1548 if (ifp->if_capenable & IFCAP_CSUM_UDPv6_Rx) {
1549 ifp->if_csum_flags_rx |= M_CSUM_UDPv6;
1550 }
1551
1552 /*
1553 * Only kick the interface if it's up. If it's
1554 * not up now, it will notice the cap enables
1555 * when it is brought up later.
1556 */
1557 if (ifp->if_flags & IFF_UP)
1558 (void) (*ifp->if_ioctl)(ifp, SIOCSIFFLAGS,
1559 (caddr_t) &ifrq);
1560 }
1561 splx(s);
1562 break;
1563
1564 case SIOCSIFMETRIC:
1565 ifp->if_metric = ifr->ifr_metric;
1566 break;
1567
1568 case SIOCGIFDATA:
1569 ifdr->ifdr_data = ifp->if_data;
1570 break;
1571
1572 case SIOCZIFDATA:
1573 ifdr->ifdr_data = ifp->if_data;
1574 /*
1575 * Assumes that the volatile counters that can be
1576 * zero'ed are at the end of if_data.
1577 */
1578 memset(&ifp->if_data.ifi_ipackets, 0, sizeof(ifp->if_data) -
1579 offsetof(struct if_data, ifi_ipackets));
1580 break;
1581
1582 case SIOCSIFMTU:
1583 {
1584 u_long oldmtu = ifp->if_mtu;
1585
1586 if (ifp->if_ioctl == NULL)
1587 return (EOPNOTSUPP);
1588 error = (*ifp->if_ioctl)(ifp, cmd, data);
1589
1590 /*
1591 * If the link MTU changed, do network layer specific procedure.
1592 */
1593 if (ifp->if_mtu != oldmtu) {
1594 #ifdef INET6
1595 nd6_setmtu(ifp);
1596 #endif
1597 }
1598 break;
1599 }
1600 case SIOCSIFPHYADDR:
1601 case SIOCDIFPHYADDR:
1602 #ifdef INET6
1603 case SIOCSIFPHYADDR_IN6:
1604 #endif
1605 case SIOCSLIFPHYADDR:
1606 case SIOCADDMULTI:
1607 case SIOCDELMULTI:
1608 case SIOCSIFMEDIA:
1609 case SIOCGIFPSRCADDR:
1610 case SIOCGIFPDSTADDR:
1611 case SIOCGLIFPHYADDR:
1612 case SIOCGIFMEDIA:
1613 if (ifp->if_ioctl == 0)
1614 return (EOPNOTSUPP);
1615 error = (*ifp->if_ioctl)(ifp, cmd, data);
1616 break;
1617
1618 case SIOCSDRVSPEC:
1619 case SIOCS80211NWID:
1620 case SIOCS80211NWKEY:
1621 case SIOCS80211POWER:
1622 case SIOCS80211BSSID:
1623 case SIOCS80211CHANNEL:
1624 default:
1625 if (so->so_proto == 0)
1626 return (EOPNOTSUPP);
1627 #ifdef COMPAT_OSOCK
1628 error = compat_ifioctl(so, cmd, data, p);
1629 #else
1630 error = ((*so->so_proto->pr_usrreq)(so, PRU_CONTROL,
1631 (struct mbuf *)cmd, (struct mbuf *)data,
1632 (struct mbuf *)ifp, p));
1633 #endif
1634 break;
1635 }
1636
1637 if (((oif_flags ^ ifp->if_flags) & IFF_UP) != 0) {
1638 #ifdef INET6
1639 if ((ifp->if_flags & IFF_UP) != 0) {
1640 s = splnet();
1641 in6_if_up(ifp);
1642 splx(s);
1643 }
1644 #endif
1645 }
1646
1647 return (error);
1648 }
1649
1650 /*
1651 * Return interface configuration
1652 * of system. List may be used
1653 * in later ioctl's (above) to get
1654 * other information.
1655 */
1656 /*ARGSUSED*/
1657 int
1658 ifconf(cmd, data)
1659 u_long cmd;
1660 caddr_t data;
1661 {
1662 struct ifconf *ifc = (struct ifconf *)data;
1663 struct ifnet *ifp;
1664 struct ifaddr *ifa;
1665 struct ifreq ifr, *ifrp;
1666 int space = ifc->ifc_len, error = 0;
1667 const int sz = (int)sizeof(ifr);
1668 int sign;
1669
1670 if ((ifrp = ifc->ifc_req) == NULL) {
1671 space = 0;
1672 sign = -1;
1673 } else {
1674 sign = 1;
1675 }
1676 IFNET_FOREACH(ifp) {
1677 bcopy(ifp->if_xname, ifr.ifr_name, IFNAMSIZ);
1678 if ((ifa = TAILQ_FIRST(&ifp->if_addrlist)) == 0) {
1679 memset(&ifr.ifr_addr, 0, sizeof(ifr.ifr_addr));
1680 if (ifrp != NULL && space >= sz) {
1681 error = copyout(&ifr, ifrp, sz);
1682 if (error)
1683 break;
1684 ifrp++;
1685 }
1686 space -= sizeof(ifr) * sign;
1687 continue;
1688 }
1689
1690 for (; ifa != 0; ifa = TAILQ_NEXT(ifa, ifa_list)) {
1691 struct sockaddr *sa = ifa->ifa_addr;
1692 #ifdef COMPAT_OSOCK
1693 if (cmd == OSIOCGIFCONF) {
1694 struct osockaddr *osa =
1695 (struct osockaddr *)&ifr.ifr_addr;
1696 /*
1697 * If it does not fit, we don't bother with it
1698 */
1699 if (sa->sa_len > sizeof(*osa))
1700 continue;
1701 ifr.ifr_addr = *sa;
1702 osa->sa_family = sa->sa_family;
1703 if (ifrp != NULL && space >= sz) {
1704 error = copyout(&ifr, ifrp, sz);
1705 ifrp++;
1706 }
1707 } else
1708 #endif
1709 if (sa->sa_len <= sizeof(*sa)) {
1710 ifr.ifr_addr = *sa;
1711 if (ifrp != NULL && space >= sz) {
1712 error = copyout(&ifr, ifrp, sz);
1713 ifrp++;
1714 }
1715 } else {
1716 space -= (sa->sa_len - sizeof(*sa)) * sign;
1717 if (ifrp != NULL && space >= sz) {
1718 error = copyout(&ifr, ifrp,
1719 sizeof(ifr.ifr_name));
1720 if (error == 0) {
1721 error = copyout(sa,
1722 &ifrp->ifr_addr,
1723 sa->sa_len);
1724 }
1725 ifrp = (struct ifreq *)
1726 (sa->sa_len +
1727 (caddr_t)&ifrp->ifr_addr);
1728 }
1729 }
1730 if (error)
1731 break;
1732 space -= sz * sign;
1733 }
1734 }
1735 if (ifrp != NULL)
1736 ifc->ifc_len -= space;
1737 else
1738 ifc->ifc_len = space;
1739 return (error);
1740 }
1741
1742 /*
1743 * Queue message on interface, and start output if interface
1744 * not yet active.
1745 */
1746 int
1747 ifq_enqueue(struct ifnet *ifp, struct mbuf *m
1748 ALTQ_COMMA ALTQ_DECL(struct altq_pktattr *pktattr))
1749 {
1750 int len = m->m_pkthdr.len;
1751 int mflags = m->m_flags;
1752 int s = splnet();
1753 int error;
1754
1755 IFQ_ENQUEUE(&ifp->if_snd, m, pktattr, error);
1756 if (error) {
1757 splx(s);
1758 return error;
1759 }
1760 ifp->if_obytes += len;
1761 if (mflags & M_MCAST)
1762 ifp->if_omcasts++;
1763 if ((ifp->if_flags & IFF_OACTIVE) == 0)
1764 (*ifp->if_start)(ifp);
1765 splx(s);
1766 return error;
1767 }
1768
1769 /*
1770 * Queue message on interface, possibly using a second fast queue
1771 */
1772 int
1773 ifq_enqueue2(struct ifnet *ifp, struct ifqueue *ifq, struct mbuf *m
1774 ALTQ_COMMA ALTQ_DECL(struct altq_pktattr *pktattr))
1775 {
1776 int error = 0;
1777
1778 if (ifq != NULL
1779 #ifdef ALTQ
1780 && ALTQ_IS_ENABLED(&ifp->if_snd) == 0
1781 #endif
1782 ) {
1783 if (IF_QFULL(ifq)) {
1784 IF_DROP(&ifp->if_snd);
1785 m_freem(m);
1786 if (error == 0)
1787 error = ENOBUFS;
1788 }
1789 else
1790 IF_ENQUEUE(ifq, m);
1791 } else
1792 IFQ_ENQUEUE(&ifp->if_snd, m, pktattr, error);
1793 if (error != 0) {
1794 ++ifp->if_oerrors;
1795 return error;
1796 }
1797
1798 return 0;
1799 }
1800
1801
1802 #if defined(INET) || defined(INET6)
1803 static void
1804 sysctl_net_ifq_setup(struct sysctllog **clog,
1805 int pf, const char *pfname,
1806 int ipn, const char *ipname,
1807 int qid, struct ifqueue *ifq)
1808 {
1809
1810 sysctl_createv(clog, 0, NULL, NULL,
1811 CTLFLAG_PERMANENT,
1812 CTLTYPE_NODE, "net", NULL,
1813 NULL, 0, NULL, 0,
1814 CTL_NET, CTL_EOL);
1815 sysctl_createv(clog, 0, NULL, NULL,
1816 CTLFLAG_PERMANENT,
1817 CTLTYPE_NODE, pfname, NULL,
1818 NULL, 0, NULL, 0,
1819 CTL_NET, pf, CTL_EOL);
1820 sysctl_createv(clog, 0, NULL, NULL,
1821 CTLFLAG_PERMANENT,
1822 CTLTYPE_NODE, ipname, NULL,
1823 NULL, 0, NULL, 0,
1824 CTL_NET, pf, ipn, CTL_EOL);
1825 sysctl_createv(clog, 0, NULL, NULL,
1826 CTLFLAG_PERMANENT,
1827 CTLTYPE_NODE, "ifq",
1828 SYSCTL_DESCR("Protocol input queue controls"),
1829 NULL, 0, NULL, 0,
1830 CTL_NET, pf, ipn, qid, CTL_EOL);
1831
1832 sysctl_createv(clog, 0, NULL, NULL,
1833 CTLFLAG_PERMANENT,
1834 CTLTYPE_INT, "len",
1835 SYSCTL_DESCR("Current input queue length"),
1836 NULL, 0, &ifq->ifq_len, 0,
1837 CTL_NET, pf, ipn, qid, IFQCTL_LEN, CTL_EOL);
1838 sysctl_createv(clog, 0, NULL, NULL,
1839 CTLFLAG_PERMANENT|CTLFLAG_READWRITE,
1840 CTLTYPE_INT, "maxlen",
1841 SYSCTL_DESCR("Maximum allowed input queue length"),
1842 NULL, 0, &ifq->ifq_maxlen, 0,
1843 CTL_NET, pf, ipn, qid, IFQCTL_MAXLEN, CTL_EOL);
1844 #ifdef notyet
1845 sysctl_createv(clog, 0, NULL, NULL,
1846 CTLFLAG_PERMANENT,
1847 CTLTYPE_INT, "peak",
1848 SYSCTL_DESCR("Highest input queue length"),
1849 NULL, 0, &ifq->ifq_peak, 0,
1850 CTL_NET, pf, ipn, qid, IFQCTL_PEAK, CTL_EOL);
1851 #endif
1852 sysctl_createv(clog, 0, NULL, NULL,
1853 CTLFLAG_PERMANENT,
1854 CTLTYPE_INT, "drops",
1855 SYSCTL_DESCR("Packets dropped due to full input queue"),
1856 NULL, 0, &ifq->ifq_drops, 0,
1857 CTL_NET, pf, ipn, qid, IFQCTL_DROPS, CTL_EOL);
1858 }
1859
1860 #ifdef INET
1861 SYSCTL_SETUP(sysctl_net_inet_ip_ifq_setup,
1862 "sysctl net.inet.ip.ifq subtree setup")
1863 {
1864 extern struct ifqueue ipintrq;
1865
1866 sysctl_net_ifq_setup(clog, PF_INET, "inet", IPPROTO_IP, "ip",
1867 IPCTL_IFQ, &ipintrq);
1868 }
1869 #endif /* INET */
1870
1871 #ifdef INET6
1872 SYSCTL_SETUP(sysctl_net_inet6_ip6_ifq_setup,
1873 "sysctl net.inet6.ip6.ifq subtree setup")
1874 {
1875 extern struct ifqueue ip6intrq;
1876
1877 sysctl_net_ifq_setup(clog, PF_INET6, "inet6", IPPROTO_IPV6, "ip6",
1878 IPV6CTL_IFQ, &ip6intrq);
1879 }
1880 #endif /* INET6 */
1881 #endif /* INET || INET6 */
1882